技术领域Technical field
本公开涉及数据处理领域中的路网技术领域，尤其涉及一种基于遥感卫星图像的地图更新方法、训练方法和设备。The present disclosure relates to the field of road network technology in the field of data processing, and in particular to a map update method, training method and equipment based on remote sensing satellite images.
背景技术Background technique
随着移动互联网和智能设备的发展，地图已经成为人们出行的重要依据。路网中的道路会发生变化，进而需要对地图进行更新。With the development of mobile Internet and smart devices, maps have become an important basis for people to travel. Roads in the road network change, requiring map updates.
现有技术中，可以人工的根据采集设备采集道路的数据，然后基于道路的数据对地图进行更新。In the existing technology, road data can be collected manually based on collection equipment, and then the map can be updated based on the road data.
然而现有技术中，人工采集道路的数据需要耗费大量的人力和物力，进而导致更新地图的成本较高；并且，上述方式的作业效率较低且容易出现错误，导致地图更新不及时、地图更新错误。However, in the existing technology, manual collection of road data requires a large amount of manpower and material resources, which in turn leads to high cost of updating the map. Moreover, the operation efficiency of the above method is low and error-prone, resulting in untimely map updates and poor map updating. mistake.
发明内容Contents of the invention
本公开提供了一种基于遥感卫星图像的地图更新方法、训练方法和设备。The present disclosure provides a map update method, training method and equipment based on remote sensing satellite images.
根据本公开的第一方面，提供了一种基于遥感卫星图像的地图更新方法，包括：According to the first aspect of the present disclosure, a map updating method based on remote sensing satellite images is provided, including:
获取遥感卫星图像；Obtain remote sensing satellite images;
对所述遥感卫星图像进行语义分割处理，得到第一拓扑图，所述第一拓扑图包括所述遥感卫星图像所对应的位置上的道路；并根据所述遥感卫星图像，确定三维张量图，所述三维张量图表征所述遥感卫星图像所对应的位置上的道路的编码信息；Perform semantic segmentation processing on the remote sensing satellite image to obtain a first topological map, which includes the road at the position corresponding to the remote sensing satellite image; and determine a three-dimensional tensor map based on the remote sensing satellite image , the three-dimensional tensor diagram represents the coded information of the road at the location corresponding to the remote sensing satellite image;
对所述三维张量图进行解码处理，得到第二拓扑图，其中，所述第二拓扑图包括所述遥感卫星图像所对应的位置上的道路；Decoding the three-dimensional tensor map to obtain a second topological map, wherein the second topological map includes the road at the location corresponding to the remote sensing satellite image;
根据所述第一拓扑图和所述第二拓扑图，确定道路拓扑图，并根据所述道路拓扑图更新地图。A road topology map is determined based on the first topology map and the second topology map, and a map is updated based on the road topology map.
根据本公开的第二方面，提供了一种应用于地图更新的图编码模型的训练方法，包括：According to a second aspect of the present disclosure, a training method for a graph coding model applied to map updating is provided, including:
获取多个待训练的遥感卫星图像，所述待训练的遥感卫星图像具有标准的三维张量图；Acquire multiple remote sensing satellite images to be trained, where the remote sensing satellite images to be trained have standard three-dimensional tensor diagrams;
重复以下各步骤，直至达到预设条件：将所述待训练的遥感卫星图像输入至图编码模型中，得到预测的三维张量图，所述预测的三维张量图表征所述待训练的遥感卫星图像所对应的位置上的道路的编码信息；根据所述预测的三维张量图和所述标准的三维张量图对所述图编码模型进行参数调整；Repeat the following steps until the preset conditions are reached: input the remote sensing satellite image to be trained into the image coding model to obtain a predicted three-dimensional tensor map, and the predicted three-dimensional tensor map represents the remote sensing to be trained Encoding information of the road at the position corresponding to the satellite image; adjusting parameters of the image encoding model according to the predicted three-dimensional tensor image and the standard three-dimensional tensor image;
其中，达到预设条件所得到的图编码模型，用于确定本公开的第一方面所述方法中的遥感卫星图像的三维张量图。Among them, the graph coding model obtained by meeting the preset conditions is used to determine the three-dimensional tensor graph of the remote sensing satellite image in the method described in the first aspect of the present disclosure.
根据本公开的第三方面，提供了一种基于遥感卫星图像的地图更新装置，包括：According to a third aspect of the present disclosure, a map updating device based on remote sensing satellite images is provided, including:
获取单元，用于获取遥感卫星图像；An acquisition unit is used to acquire remote sensing satellite images;
第一确定单元，用于对所述遥感卫星图像进行语义分割处理，得到第一拓扑图，所述第一拓扑图包括所述遥感卫星图像所对应的位置上的道路；并根据所述遥感卫星图像，确定三维张量图，所述三维张量图表征所述遥感卫星图像所对应的位置上的道路的编码信息；A first determination unit configured to perform semantic segmentation processing on the remote sensing satellite image to obtain a first topological map, where the first topological map includes the road at the location corresponding to the remote sensing satellite image; and according to the remote sensing satellite image Image, determine a three-dimensional tensor map, the three-dimensional tensor map represents the coded information of the road at the position corresponding to the remote sensing satellite image;
第二确定单元，用于对所述三维张量图进行解码处理，得到第二拓扑图，其中，所述第二拓扑图包括所述遥感卫星图像所对应的位置上的道路；A second determination unit configured to decode the three-dimensional tensor map to obtain a second topological map, wherein the second topological map includes the road at the location corresponding to the remote sensing satellite image;
第三确定单元，用于根据所述第一拓扑图和所述第二拓扑图，确定道路拓扑图，并根据所述道路拓扑图更新地图。A third determination unit is configured to determine a road topology map based on the first topology map and the second topology map, and update a map based on the road topology map.
根据本公开的第四方面，提供了一种应用于地图更新的图编码模型的训练装置，包括：According to a fourth aspect of the present disclosure, a training device for a graph coding model applied to map updating is provided, including:
第一获取单元，用于获取多个待训练的遥感卫星图像，所述待训练的遥感卫星图像具有标准的三维张量图；The first acquisition unit is used to acquire multiple remote sensing satellite images to be trained, where the remote sensing satellite images to be trained have a standard three-dimensional tensor map;
第一确定单元，用于重复以下各步骤，直至达到预设条件：将所述待训练的遥感卫星图像输入至图编码模型中，得到预测的三维张量图，所述预测的三维张量图表征所述待训练的遥感卫星图像所对应的位置上的道路的编码信息；根据所述预测的三维张量图和所述标准的三维张量图对所述图编码模型进行参数调整；The first determination unit is used to repeat the following steps until the preset conditions are reached: input the remote sensing satellite image to be trained into the graph coding model to obtain a predicted three-dimensional tensor graph, and the predicted three-dimensional tensor graph Characterizing the coding information of the road at the location corresponding to the remote sensing satellite image to be trained; adjusting the parameters of the graph coding model according to the predicted three-dimensional tensor graph and the standard three-dimensional tensor graph;
其中，达到预设条件所得到的图编码模型，用于确定本公开的第三方面所述装置中的遥感卫星图像的三维张量图。Among them, the graph coding model obtained by meeting the preset conditions is used to determine the three-dimensional tensor graph of the remote sensing satellite image in the device described in the third aspect of the present disclosure.
根据本公开的第五方面，提供了一种计算机设备，包括：至少一个处理器；以及According to a fifth aspect of the present disclosure, a computer device is provided, including: at least one processor; and
与所述至少一个处理器通信连接的存储器；其中，a memory communicatively connected to the at least one processor; wherein,
所述存储器存储有可被所述至少一个处理器执行的指令，所述指令被所述至少一个处理器执行，以使所述至少一个处理器能够执行第一方面或者第二方面所述的方法。The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method described in the first aspect or the second aspect. .
根据本公开的第六方面，提供了一种存储有计算机指令的非瞬时计算机可读存储介质，其中，所述计算机指令用于使所述计算机执行第一方面或者第二方面所述的方法。According to a sixth aspect of the present disclosure, there is provided a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to execute the method described in the first or second aspect.
根据本公开的第七方面，提供了一种计算机程序产品，所述计算机程序产品包括：计算机程序，所述计算机程序存储在可读存储介质中，计算机设备的至少一个处理器可以从所述可读存储介质读取所述计算机程序，所述至少一个处理器执行所述计算机程序使得计算机设备执行第一方面或者第二方面所述的方法。According to a seventh aspect of the present disclosure, a computer program product is provided, the computer program product comprising: a computer program stored in a readable storage medium, from which at least one processor of a computer device can obtain Reading the storage medium reads the computer program, and the at least one processor executes the computer program so that the computer device performs the method described in the first aspect or the second aspect.
根据本公开的技术解决了由于人工采集道路数据导致的更新地图成本较高以及地图更新错误的问题。The technology according to the present disclosure solves the problems of high map update costs and map update errors caused by manual collection of road data.
应当理解，本部分所描述的内容并非旨在标识本公开的实施例的关键或重要特征，也不用于限制本公开的范围。本公开的其它特征将通过以下的说明书而变得容易理解。It should be understood that what is described in this section is not intended to identify key or important features of the embodiments of the disclosure, nor is it intended to limit the scope of the disclosure. Other features of the present disclosure will become readily understood from the following description.
附图说明Description of the drawings
附图用于更好地理解本方案，不构成对本公开的限定。其中：The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure. in:
图1是根据本公开第一实施例的示意图；Figure 1 is a schematic diagram according to a first embodiment of the present disclosure;
图2是根据本公开第一实施例的一种遥感卫星图像的示意图；Figure 2 is a schematic diagram of a remote sensing satellite image according to the first embodiment of the present disclosure;
图3是根据本公开第二实施例的示意图；Figure 3 is a schematic diagram according to a second embodiment of the present disclosure;
图4是根据本公开第二实施例的一种局部特征图生成过程的示意图；Figure 4 is a schematic diagram of a local feature map generation process according to the second embodiment of the present disclosure;
图5是根据本公开第三实施例的示意图；Figure 5 is a schematic diagram according to a third embodiment of the present disclosure;
图6是根据本公开第四实施例的示意图；Figure 6 is a schematic diagram according to a fourth embodiment of the present disclosure;
图7是根据本公开第四实施例的一种用户对待训练的遥感卫星图像中的道路关键点进行标注后的示意图；Figure 7 is a schematic diagram of a user marking road key points in remote sensing satellite images to be trained according to the fourth embodiment of the present disclosure;
图8是根据本公开第四实施例的一种道路关键点的像素点的相邻的其他道路关键点的个数统计示意图；Figure 8 is a schematic diagram showing the number of adjacent road key points of a pixel of a road key point according to the fourth embodiment of the present disclosure;
图9是根据本公开第四实施例的一种待训练的遥感卫星图像的每一像素点的编码数据以及三维张量图；Figure 9 is the encoded data and three-dimensional tensor diagram of each pixel of a remote sensing satellite image to be trained according to the fourth embodiment of the present disclosure;
图10是根据本公开第五实施例的示意图；Figure 10 is a schematic diagram according to a fifth embodiment of the present disclosure;
图11是根据本公开第六实施例的示意图；Figure 11 is a schematic diagram according to a sixth embodiment of the present disclosure;
图12是根据本公开第七实施例的示意图；Figure 12 is a schematic diagram according to a seventh embodiment of the present disclosure;
图13是根据本公开第八实施例的示意图；Figure 13 is a schematic diagram according to an eighth embodiment of the present disclosure;
图14是根据本公开第九实施例的示意图；Figure 14 is a schematic diagram according to a ninth embodiment of the present disclosure;
图15是根据本公开第十实施例的示意图；Figure 15 is a schematic diagram according to a tenth embodiment of the present disclosure;
图16是用来实现本公开实施例的基于遥感卫星图像的地图更新方法的计算机设备的框图；Figure 16 is a block diagram of a computer device used to implement a map update method based on remote sensing satellite images according to an embodiment of the present disclosure;
图17是用来实现本公开实施例的应用于地图更新的图编码模型的训练方法的计算机设备的框图。17 is a block diagram of a computer device used to implement a training method of a graph coding model applied to map updating according to an embodiment of the present disclosure.
具体实施方式Detailed ways
以下结合附图对本公开的示范性实施例做出说明，其中包括本公开实施例的各种细节以助于理解，应当将它们认为仅仅是示范性的。因此，本领域普通技术人员应当认识到，可以对这里描述的实施例做出各种改变和修改，而不会背离本公开的范围和精神。同样，为了清楚和简明，以下的描述中省略了对公知功能和结构的描述。Exemplary embodiments of the present disclosure are described below with reference to the accompanying drawings, in which various details of the embodiments of the present disclosure are included to facilitate understanding and should be considered to be exemplary only. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the disclosure. Also, descriptions of well-known functions and constructions are omitted from the following description for clarity and conciseness.
本公开适用于采集地图信息的场景下，当前的地图信息中可能会缺失一些道路信息，尤其是内部道路，即小区内的一些道路。若单独采用语义分割处理遥感卫星图像，或者单独采用图编码处理遥感卫星图像，均会出现召回的地图信息不全面的问题。This disclosure is applicable to scenarios where map information is collected. Some road information may be missing from the current map information, especially internal roads, that is, some roads within the community. If semantic segmentation is used alone to process remote sensing satellite images, or image coding is used alone to process remote sensing satellite images, there will be a problem of incomplete map information being recalled.
本公开提供一种基于遥感卫星图像的地图更新方法、训练方法和设备，应用于数据处理领域中的路网技术领域，以解决由于人工采集道路数据导致的更新地图成本较高以及地图更新错误的问题。The present disclosure provides a map update method, training method and equipment based on remote sensing satellite images, which are applied to the road network technology field in the field of data processing to solve the problem of high update cost and map update errors caused by manual collection of road data. question.
图1是根据本公开第一实施例的示意图，如图1所示，图1示出的是一种基于遥感卫星图像的地图更新方法，该方法包括：Figure 1 is a schematic diagram according to the first embodiment of the present disclosure. As shown in Figure 1, Figure 1 shows a map update method based on remote sensing satellite images. The method includes:
S101、获取遥感卫星图像。S101. Obtain remote sensing satellite images.
示例性地，遥感卫星图像是指记录各种地物电磁波大小的胶片或照片，具有很高的分辨率，其中，分辨率包括空间分辨率、光谱分辨率、辐射分辨率和时间分辨率。其中，空间分辨率是遥感卫星图像上能够详细区分的最小单元的尺寸或大小，或指遥感器区分两个目标的最小角度或线性距离的度量。光谱分辨率是指遥感器接收目标辐射时能分辨的最小波间隔。辐射分辨率是遥感器感测元件在接收光谱信号时能分辨的最小辐射度差。时间分辨率是关于遥感卫星图像间隔时间的一项性能指标。For example, remote sensing satellite images refer to films or photos that record the size of electromagnetic waves of various ground objects, and have a very high resolution, where the resolution includes spatial resolution, spectral resolution, radiation resolution and time resolution. Among them, spatial resolution is the size or size of the smallest unit that can be distinguished in detail on a remote sensing satellite image, or refers to the measurement of the minimum angle or linear distance at which a remote sensor distinguishes two targets. Spectral resolution refers to the minimum wave interval that a remote sensor can resolve when receiving target radiation. The radiometric resolution is the minimum radiometric difference that the remote sensor sensing element can resolve when receiving spectral signals. Temporal resolution is a performance indicator regarding the time interval between remote sensing satellite images.
S102、对遥感卫星图像进行语义分割处理，得到第一拓扑图，第一拓扑图包括遥感卫星图像所对应的位置上的道路；并根据遥感卫星图像，确定三维张量图，三维张量图表征遥感卫星图像所对应的位置上的道路的编码信息。S102. Perform semantic segmentation processing on the remote sensing satellite image to obtain the first topological map. The first topological map includes the road at the position corresponding to the remote sensing satellite image; and determine the three-dimensional tensor map and the three-dimensional tensor map representation based on the remote sensing satellite image. The encoded information of the road at the location corresponding to the remote sensing satellite image.
示例性地，语义分割处理是指对遥感卫星图像按照区域进行分割，每一个区域由具备同一个属性的像素点组成。其中，语义分割处理可以分为编码器网络和解码器网络，编码器网络是一个预先训练的分类网络，解码器网络是将编码器学习到的识别特征语义投影到像素空间上，得到密集的分类。进一步地，语音分割处理不仅需要在像素级别上进行区分，还需要一种机制将编码器不同阶段学习到的区分特征投影到像素空间上。例如，图2示出了一种遥感卫星图像的示意图。分析图2的遥感卫星图像可以看到地面的道路结构。For example, semantic segmentation processing refers to segmenting remote sensing satellite images according to regions, and each region is composed of pixels with the same attribute. Among them, semantic segmentation processing can be divided into encoder network and decoder network. The encoder network is a pre-trained classification network, and the decoder network projects the recognition feature semantics learned by the encoder onto the pixel space to obtain dense classification. . Furthermore, speech segmentation processing not only requires distinction at the pixel level, but also requires a mechanism to project the distinguishing features learned at different stages of the encoder onto the pixel space. For example, Figure 2 shows a schematic diagram of a remote sensing satellite image. By analyzing the remote sensing satellite image in Figure 2, we can see the road structure on the ground.
图2中所示出的遥感卫星图像经过语义分割处理后，可以得到第一拓扑图。其中，第一拓扑图能够表征遥感卫星图像中道路和背景的界限。After the remote sensing satellite image shown in Figure 2 is processed through semantic segmentation, the first topological map can be obtained. Among them, the first topological map can represent the boundaries of roads and backgrounds in remote sensing satellite images.
本实施例中，三维张量图表征遥感卫星图像所对应的位置上的道路的编码信息。三维张量图是指由三个部分组成的图像，可以表征遥感卫星图像的参数。第一个部分可以表征当前的像素点是否为道路关键点，第二个部分可以表征当前的像素点是否存在相邻的道路关键点，第三个部分可以表征当前的像素点与其他像素点之间的位置关系，其中，位置关系可以是当前的像素点与其他像素点的位置偏移量。In this embodiment, the three-dimensional tensor diagram represents the encoded information of the road at the location corresponding to the remote sensing satellite image. A three-dimensional tensor map refers to an image composed of three parts, which can characterize the parameters of remote sensing satellite images. The first part can characterize whether the current pixel is a road key point, the second part can characterize whether the current pixel has adjacent road key points, and the third part can characterize the relationship between the current pixel and other pixels. The positional relationship between the pixels can be the positional offset between the current pixel and other pixels.
S103、对三维张量图进行解码处理，得到第二拓扑图，其中，第二拓扑图包括遥感卫星图像所对应的位置上的道路。S103. Decode the three-dimensional tensor map to obtain a second topological map, where the second topological map includes roads at positions corresponding to remote sensing satellite images.
示例性地，对三维张量图进行解码处理的过程为将三维张量图的三个部分的编码数据位按照每一个部分的编码数据位的含义进行反向解释，得到第二拓扑图。其中，第二拓扑图也能够表征遥感卫星图像中道路和背景的界限。For example, the process of decoding the three-dimensional tensor graph is to reversely interpret the encoded data bits of the three parts of the three-dimensional tensor graph according to the meaning of the encoded data bits of each part to obtain the second topological graph. Among them, the second topological map can also represent the boundaries between roads and backgrounds in remote sensing satellite images.
S104、根据第一拓扑图和第二拓扑图，确定道路拓扑图，并根据道路拓扑图更新地图。S104. Determine the road topology map based on the first topology map and the second topology map, and update the map based on the road topology map.
示例性地，将第一拓扑图和第二拓扑图进行比较，确定两者之间的差异，根据两者的差异，得到道路拓扑图。在得到道路拓扑图后，将最新的道路拓扑图更新地图。For example, the first topology map and the second topology map are compared to determine the difference between the two, and a road topology map is obtained based on the difference between the two. After obtaining the road topology map, update the map with the latest road topology map.
本公开提供一种基于遥感卫星图像的地图更新方法，包括：获取车辆在历史轨迹上的历史定位信息，根据帧的全局定位信息所指示的与帧对应的定位信号的强度，确定帧的权重信息，根据帧的全局定位信息、帧间定位信息以及权重信息，确定帧的优化定位信息。通过该技术方案，能够解决由于人工采集道路数据导致的更新地图成本较高以及地图更新错误的问题。The present disclosure provides a map update method based on remote sensing satellite images, which includes: obtaining historical positioning information of vehicles on historical trajectories, and determining the weight information of the frame according to the strength of the positioning signal corresponding to the frame indicated by the global positioning information of the frame. , determine the optimized positioning information of the frame based on the global positioning information, inter-frame positioning information and weight information of the frame. Through this technical solution, the problems of high map update costs and map update errors caused by manual collection of road data can be solved.
图3是根据本公开第二实施例的示意图，如图3所示，图3示出的是一种基于遥感卫星图像的地图更新方法，该方法包括：Figure 3 is a schematic diagram according to the second embodiment of the present disclosure. As shown in Figure 3, Figure 3 shows a map update method based on remote sensing satellite images. The method includes:
S301、获取遥感卫星图像。S301. Obtain remote sensing satellite images.
示例性地，本步骤可以参见步骤S101，在此不再赘述。For example, this step can refer to step S101, which will not be described again here.
S302、对遥感卫星图像进行语义分割处理，得到第一拓扑图，第一拓扑图包括遥感卫星图像所对应的位置上的道路。S302. Perform semantic segmentation processing on the remote sensing satellite image to obtain a first topological map. The first topological map includes the road at the location corresponding to the remote sensing satellite image.
示例性地，本步骤可以参见步骤S102，在此不再赘述。For example, this step can refer to step S102, which will not be described again here.
S303、将遥感卫星图像输入至图编码模型中，得到三维张量图；其中，图编码模型为基于具有标准的三维张量图的遥感卫星图像进行训练所得到的。S303. Input the remote sensing satellite image into the graph coding model to obtain a three-dimensional tensor graph; wherein the graph coding model is obtained by training based on the remote sensing satellite image with a standard three-dimensional tensor graph.
示例性地，图编码模型是用于输出三维张量图的模型，将多张待识别的遥感卫星图像输入至图编码模型中，可以分别得到与每一张待识别的遥感卫星图像对应的三维张量图。其中，图编码模型是预先进行训练得到的。这样设置的好处是能够使用图编码模型实现端对端的输入和输出，能够快速地输出与遥感卫星图像对应的三维张量图。For example, the graph encoding model is a model used to output a three-dimensional tensor graph. By inputting multiple remote sensing satellite images to be identified into the graph encoding model, the three-dimensional image corresponding to each remote sensing satellite image to be identified can be obtained. Tensor graph. Among them, the graph coding model is trained in advance. The advantage of this setting is that it can use the graph coding model to realize end-to-end input and output, and can quickly output the three-dimensional tensor map corresponding to the remote sensing satellite image.
示例性地，将遥感卫星图像输入至图编码模型中，得到三维张量图，包括：For example, remote sensing satellite images are input into the graph coding model to obtain a three-dimensional tensor graph, including:
基于图编码模型对遥感卫星图像进行特征提取，得到全局特征图和局部特征图；其中，全局特征图表征遥感卫星图像的全局特征，局部特征图表征遥感卫星图像的道路特征；Based on the graph coding model, feature extraction is performed on remote sensing satellite images to obtain global feature maps and local feature maps; among them, the global feature map represents the global features of remote sensing satellite images, and the local feature map represents the road features of remote sensing satellite images;
基于图编码模型对全局特征图和局部特征图进行特征融合，得到融合后的特征图；Based on the graph coding model, the global feature map and the local feature map are feature fused to obtain the fused feature map;
根据融合后的特征图，生成三维张量图。Based on the fused feature map, a three-dimensional tensor map is generated.
示例性地，全局特征图是用于描述遥感卫星图像的颜色特征、纹理特征和/或形状特征的整体特征。局部特征图是用于描述遥感卫星图像局部的特征，具体的，可以是从遥感卫星图像中提取的特征，包括边缘、角点、线、曲线和特殊属性的区域等。局部特征图具备特征间相关度小，遮挡情况下不会因为部分局部特征的消失而影响其他局部特征的检测和匹配。将全局特征图和局部特征图通过图编码模型进行融合，融合后的特征图是由全局特征图和局部特征图组成的，融合后的特征图并未对全局特征图和局部特征图进行修改，只是将两者作为一个整体。根据融合后的特征图，将该张遥感卫星图像生成对应的三维张量图。这样设置的好处是能够充分结合遥感卫星图像的全局特征和局部特征，使得生成的三维张量图的信息更加准确和丰富。Illustratively, the global feature map is an overall feature used to describe the color features, texture features and/or shape features of remote sensing satellite images. Local feature maps are used to describe local features of remote sensing satellite images. Specifically, they can be features extracted from remote sensing satellite images, including edges, corners, lines, curves, and areas with special attributes. The local feature map has a small correlation between features. In the case of occlusion, the disappearance of some local features will not affect the detection and matching of other local features. The global feature map and the local feature map are fused through the graph coding model. The fused feature map is composed of the global feature map and the local feature map. The fused feature map does not modify the global feature map and the local feature map. Just take the two as a whole. According to the fused feature map, the corresponding three-dimensional tensor map is generated from the remote sensing satellite image. The advantage of this setting is that it can fully combine the global features and local features of remote sensing satellite images, making the information of the generated three-dimensional tensor map more accurate and rich.
示例性地，基于图编码模型对遥感卫星图像进行特征提取，得到全局特征图和局部特征图，包括：For example, feature extraction is performed on remote sensing satellite images based on the graph coding model to obtain global feature maps and local feature maps, including:
基于图编码模型对遥感卫星图像进行特征提取，得到全局特征图；Based on the graph coding model, feature extraction is performed on remote sensing satellite images to obtain global feature maps;
对全局特征图进行二值化处理，得到二值化的特征图，二值化的特征图中包括道路的特征；Binarize the global feature map to obtain a binarized feature map, which includes the characteristics of the road;
基于二值化的特征图中的道路点，在全局特征图中确定与道路点对应的道路位置区域；并根据与道路点对应的道路位置区域，生成局部特征图。Based on the road points in the binarized feature map, the road location area corresponding to the road point is determined in the global feature map; and a local feature map is generated based on the road location area corresponding to the road point.
示例性地，在获取到全局特征图后，对全局特征图进行二值化处理，将全局特征图中道路点取为1，将全局特征图中非道路点取为0，得到二值化的特征图，并在二值化的特征图中任意选取一个道路点，在全局特征图找到对应的道路点，以该道路点确定道路位置区域，其中，道路位置区域的数量和形状不进行限制。例如，道路位置区域可以是以该道路点为圆心，以预设的距离为半径确定的圆，将这个圆作为该道路点的道路位置区域。还可以是以该道路点为质心，以预设的范围作4个矩形，将上述4个矩形作为该道路点的道路位置区域。值得注意的是，道路位置区域是包含该道路点的一个范围，该范围的划分标准并不进行限制。具体的，可以参见图4中示出的一种局部特征图生成过程的示意图。可以从图中看出，将遥感卫星图像A通过图编码模型的第一层神经网络模型输出全局特征图A后，对该全局特征图A进行二值化处理，得到二值化的特征图B，在二值化的特征图B中选取一个道路点C，在该全局特征图A确定道路位置区域，图示中的道路位置区域为4个矩形。For example, after obtaining the global feature map, perform binarization processing on the global feature map, set the road points in the global feature map to 1, set the non-road points in the global feature map to 0, and obtain the binarized Feature map, and randomly select a road point in the binary feature map, find the corresponding road point in the global feature map, and use the road point to determine the road location area, where the number and shape of the road location area are not limited. For example, the road location area may be a circle with the road point as the center and a preset distance as the radius, and this circle is regarded as the road location area of the road point. It is also possible to use the road point as the centroid, create four rectangles in a preset range, and use the above four rectangles as the road location area of the road point. It is worth noting that the road location area is a range that includes the road point, and the division standard of this range is not limited. Specifically, please refer to the schematic diagram of a local feature map generation process shown in Figure 4. It can be seen from the figure that after the remote sensing satellite image A is outputted through the first layer neural network model of the graph coding model to output the global feature map A, the global feature map A is binarized to obtain the binarized feature map B. , select a road point C in the binary feature map B, and determine the road location area in the global feature map A. The road location area in the illustration is four rectangles.
这样设置的好处是通过图编码模型中的两层神经网络模型结构，分别获取全局特征图和局部特征图，这样比单一的一层神经网络模型获取到的信息更全面。The advantage of this setting is that through the two-layer neural network model structure in the graph coding model, global feature maps and local feature maps are obtained respectively, which obtains more comprehensive information than a single layer of neural network model.
示例性地，基于图编码模型对全局特征图和局部特征图进行特征融合，得到融合后的特征图，包括：For example, the global feature map and the local feature map are feature fused based on the graph coding model to obtain the fused feature map, including:
对局部特征图进行上采样处理，得到上采样后的局部特征图；上采样后的局部特征图的尺寸与全局特征图的尺寸相同；Upsample the local feature map to obtain the upsampled local feature map; the size of the upsampled local feature map is the same as the size of the global feature map;
基于图编码模型对全局特征图和上采样后的局部特征图进行特征融合，得到融合后的特征图。Based on the graph coding model, the global feature map and the upsampled local feature map are feature fused to obtain the fused feature map.
示例性地，上采样处理是以数据量较多的样本数量为标准，将数据量少的样本数量生成与数据量较多的样本数量相同的样本数量。例如，在本实施例中，局部特征图可以为4*4*n，全局特征图可以为8*8*N，则局部特征图以全局特征图为标准进行上采样后得到的上采样后的局部特征图为8*8*n。将上采样后的局部特征图8*8*n与全局特征图8*8*N，基于图编码模型得到融合后的特征图8*8*(N+n)。这样设置的好处是能够使得全局特征图与局部特征图是在同一个维度下进行比较，是更加合理的一种方式。For example, the upsampling process uses the number of samples with a large amount of data as a standard, and generates the same number of samples from the number of samples with a small amount of data as the number of samples with a large amount of data. For example, in this embodiment, the local feature map can be 4*4*n and the global feature map can be 8*8*N. Then the upsampled result obtained by upsampling the local feature map is based on the global feature map. The local feature map is 8*8*n. The upsampled local feature map 8*8*n and the global feature map 8*8*N are combined to obtain the fused feature map 8*8*(N+n) based on the graph coding model. The advantage of this setting is that it allows the global feature map and the local feature map to be compared in the same dimension, which is a more reasonable way.
S304、对三维张量图进行解码处理，得到第二拓扑图，其中，第二拓扑图包括遥感卫星图像所对应的位置上的道路。S304. Decode the three-dimensional tensor map to obtain a second topological map, where the second topological map includes roads at positions corresponding to remote sensing satellite images.
示例性地，本步骤可以参见步骤S103，在此不再赘述。For example, this step can refer to step S103, which will not be described again here.
S305、若确定第一拓扑图中的道路像素点，不存在于第二拓扑图中，则将第一拓扑图中的道路像素点，加入到第二拓扑图中，以生成道路拓扑图。S305. If it is determined that the road pixels in the first topology map do not exist in the second topology map, add the road pixels in the first topology map to the second topology map to generate a road topology map.
示例性地，按照第一拓扑图中的坐标信息查找第一拓扑图中的道路像素点，若该坐标信息在第二拓扑图为非道路像素点，则将第一拓扑图中的该道路像素点加入到第二拓扑图中，并将修改后的第二拓扑图作为道路拓扑图。For example, the road pixels in the first topology map are searched according to the coordinate information in the first topology map. If the coordinate information is a non-road pixel in the second topology map, the road pixels in the first topology map are Points are added to the second topology map, and the modified second topology map is used as the road topology map.
例如，第一拓扑图中的坐标信息A(a，b)为一个道路像素点，则在第二拓扑图中查找坐标信息A(a，b)的像素点是否为道路像素点，若是，则不对第二拓扑图做任何处理，若不是，则将第二拓扑图中坐标信息A(a，b)的像素点添加为道路像素点。值得注意的是，第二拓扑图和第一拓扑图中坐标信息的划分标准是相同的，例如，第一拓扑图和第二拓扑图左下角均为原点，以第一拓扑图和第二拓扑图最下面的水平边为x轴，以第一拓扑图和第二拓扑图最左面的垂直边为y轴。这样设置的好处是能够结合两张拓扑图综合确定道路拓扑图，能够弥补一张拓扑图对道路召回不足的问题。For example, if the coordinate information A(a, b) in the first topological map is a road pixel, then search whether the pixel of the coordinate information A(a, b) in the second topological map is a road pixel. If so, then No processing is performed on the second topology map. If not, the pixels of coordinate information A(a, b) in the second topology map are added as road pixels. It is worth noting that the criteria for dividing coordinate information in the second topology map and the first topology map are the same. For example, the lower left corner of the first topology map and the second topology map is the origin. The bottom horizontal side of the graph is the x-axis, and the leftmost vertical side of the first topology graph and the second topology graph is the y-axis. The advantage of this setting is that it can combine two topology maps to comprehensively determine the road topology map, which can make up for the problem of insufficient road recall in one topology map.
进一步地，其中，第一拓扑图为二值图，第二拓扑图为二值图。这样设置的好处是在能够提高第一拓扑图和第二拓扑图的比对效率。Further, the first topological graph is a binary graph, and the second topological graph is a binary graph. The advantage of this setting is that it can improve the comparison efficiency between the first topology map and the second topology map.
S306、对道路拓扑图进行图像增强处理，得到增强处理后的道路拓扑图。S306. Perform image enhancement processing on the road topology map to obtain an enhanced road topology map.
示例性地，图像增强处理可根据过程所在的空间不同，可分为基于空域和频域的方法。基于空域的方法直接对道路拓扑图进行处理；基于频域的方法是在道路拓扑图的某种变换域内对道路拓扑图的变换系数进行修正，然后再反变换到原来的空域，得到增强处理后的道路拓扑图。这样设置的好处是为了改善道路拓扑图的视觉效果，提高道路拓扑图的清晰度；或者是针对道路拓扑图的应用场合，突出某些感兴趣的特征，抑制不感兴趣的特征，以扩大道路拓扑图中不同物体特征之间的差别。For example, image enhancement processing can be divided into methods based on spatial domain and frequency domain according to the space where the process occurs. The method based on the spatial domain directly processes the road topology map; the method based on the frequency domain corrects the transformation coefficient of the road topology map in a certain transformation domain of the road topology map, and then inversely transforms it to the original spatial domain to obtain the enhanced processing road topology map. The advantage of this setting is to improve the visual effect of the road topology map and improve the clarity of the road topology map; or to highlight certain interesting features and suppress uninteresting features for the application of the road topology map to expand the road topology. The difference between the features of different objects in the picture.
图5是根据本公开第三实施例的示意图，如图5所示，图5示出的是一种应用于地图更新的图编码模型的训练方法，该方法包括：Figure 5 is a schematic diagram according to a third embodiment of the present disclosure. As shown in Figure 5, Figure 5 shows a training method for a graph coding model applied to map update. The method includes:
S501、获取多个待训练的遥感卫星图像，待训练的遥感卫星图像具有标准的三维张量图。S501. Acquire multiple remote sensing satellite images to be trained. The remote sensing satellite images to be trained have standard three-dimensional tensor diagrams.
示例性地，每一个待训练的遥感卫星图像均具有唯一的标准的三维张量图，多个待训练的遥感卫星图像则对应多个标准的三维张量图。For example, each remote sensing satellite image to be trained has a unique standard three-dimensional tensor map, and multiple remote sensing satellite images to be trained correspond to multiple standard three-dimensional tensor maps.
S502、重复以下各步骤，直至达到预设条件：将待训练的遥感卫星图像输入至图编码模型中，得到预测的三维张量图，预测的三维张量图表征待训练的遥感卫星图像所对应的位置上的道路的编码信息；根据预测的三维张量图和标准的三维张量图对图编码模型进行参数调整；其中，达到预设条件所得到的图编码模型，用于确定上述实施例方法中的遥感卫星图像的三维张量图。S502. Repeat the following steps until the preset conditions are reached: input the remote sensing satellite image to be trained into the image coding model to obtain a predicted three-dimensional tensor map. The predicted three-dimensional tensor map represents the remote sensing satellite image to be trained. The coding information of the road at the position; parameter adjustment of the graph coding model is performed based on the predicted three-dimensional tensor graph and the standard three-dimensional tensor graph; wherein, the graph coding model obtained by meeting the preset conditions is used to determine the above embodiment Three-dimensional tensor map of remote sensing satellite images in the method.
示例性地，图编码模型是由两层神经网络构成的模型，将待训练的遥感卫星图像输入至图编码模型中，由图编码模型输出预测的三维张量图，然后将预测的三维张量图与标准的三维张量图通过损失函数进行比较，得到图编码模型中每一层神经网络的参数，直到将待训练的遥感卫星图像输入至图编码模型，能够输出标准的三维张量图，则将此时达到预设条件的图编码模型确定待识别的遥感卫星图像的三维张量图。For example, the graph coding model is a model composed of two layers of neural networks. Remote sensing satellite images to be trained are input into the graph coding model, and the graph coding model outputs a predicted three-dimensional tensor map, and then the predicted three-dimensional tensor is The graph is compared with the standard three-dimensional tensor graph through the loss function to obtain the parameters of each layer of the neural network in the graph encoding model. Until the remote sensing satellite image to be trained is input to the graph encoding model, a standard three-dimensional tensor graph can be output. Then the image coding model that meets the preset conditions at this time is used to determine the three-dimensional tensor image of the remote sensing satellite image to be identified.
本公开提供一种应用于地图更新的图编码模型的训练方法，通过多个待训练的遥感卫星图像与多个待训练的遥感卫星图像对应的多个标准的三维张量图训练图编码模型，将待识别的遥感卫星图像输入到得到的图编码模型中确定待识别的遥感卫星图像的三维张量图。采用上述技术手段，能够得到比较准确的图编码模型，进而可以将待识别的遥感卫星图像输入至准确的图编码模型，得到比较准确的待识别的遥感卫星图像的三维张量图。The present disclosure provides a training method for a graph coding model applied to map updating. The graph coding model is trained through multiple remote sensing satellite images to be trained and multiple standard three-dimensional tensor graphs corresponding to the multiple remote sensing satellite images to be trained. The remote sensing satellite image to be identified is input into the obtained graph coding model to determine the three-dimensional tensor map of the remote sensing satellite image to be identified. Using the above technical means, a relatively accurate graph coding model can be obtained, and then the remote sensing satellite image to be identified can be input into the accurate graph encoding model to obtain a relatively accurate three-dimensional tensor map of the remote sensing satellite image to be identified.
图6是根据本公开第四实施例的示意图，如图6所示，图6示出的是一种应用于地图更新的图编码模型的训练方法，该方法包括：Figure 6 is a schematic diagram according to the fourth embodiment of the present disclosure. As shown in Figure 6, Figure 6 shows a training method of a graph coding model applied to map update. The method includes:
S601、获取多个待训练的遥感卫星图像，待训练的遥感卫星图像具有标准的三维张量图。S601. Acquire multiple remote sensing satellite images to be trained. The remote sensing satellite images to be trained have standard three-dimensional tensor diagrams.
示例性地，本步骤可以参见步骤S501，在此不再赘述。For example, this step can refer to step S501, which will not be described again here.
S602、基于图编码模型对待训练的遥感卫星图像进行特征提取，得到全局特征图和局部特征图；其中，全局特征图表征待训练的遥感卫星图像的全局特征，局部特征图表征待训练的遥感卫星图像的道路特征。S602. Extract features from the remote sensing satellite images to be trained based on the graph coding model to obtain global feature maps and local feature maps; among which, the global feature map represents the global features of the remote sensing satellite images to be trained, and the local feature map represents the remote sensing satellite images to be trained. Road features of the image.
示例性地，基于图编码模型对待训练的遥感卫星图像进行特征提取，得到全局特征图和局部特征图，包括：For example, feature extraction is performed on the remote sensing satellite images to be trained based on the graph coding model to obtain global feature maps and local feature maps, including:
基于图编码模型对待训练的遥感卫星图像进行特征提取，得到全局特征图；Based on the graph coding model, feature extraction is performed on the remote sensing satellite images to be trained to obtain the global feature map;
对全局特征图进行二值化处理，得到二值化的特征图，二值化的特征图中包括道路的特征；Binarize the global feature map to obtain a binarized feature map, which includes the characteristics of the road;
基于二值化的特征图中的道路点，在全局特征图中确定与道路点对应的道路位置区域；并根据与道路点对应的道路位置区域，生成局部特征图。Based on the road points in the binarized feature map, the road location area corresponding to the road point is determined in the global feature map; and a local feature map is generated based on the road location area corresponding to the road point.
示例性地，本步骤可以参见步骤S303，在此不再赘述。For example, this step can refer to step S303, which will not be described again here.
S603、基于图编码模型对全局特征图和局部特征图进行特征融合，得到融合后的特征图。S603. Perform feature fusion on the global feature map and the local feature map based on the graph coding model to obtain the fused feature map.
示例性地，本步骤可以参见步骤S303，在此不再赘述。For example, this step can refer to step S303, which will not be described again here.
本实施例中，基于图编码模型对全局特征图和局部特征图进行特征融合，得到融合后的特征图，包括：In this embodiment, the global feature map and the local feature map are feature fused based on the graph coding model to obtain the fused feature map, which includes:
对局部特征图进行上采样处理，得到上采样后的局部特征图；上采样后的局部特征图的尺寸与全局特征图的尺寸相同；Upsample the local feature map to obtain the upsampled local feature map; the size of the upsampled local feature map is the same as the size of the global feature map;
基于图编码模型对全局特征图和上采样后的局部特征图进行特征融合，得到融合后的特征图。Based on the graph coding model, the global feature map and the upsampled local feature map are feature fused to obtain the fused feature map.
示例性地，本步骤可以参见步骤S303，在此不再赘述。For example, this step can refer to step S303, which will not be described again here.
S604、根据融合后的特征图，生成三维张量图。S604. Generate a three-dimensional tensor map based on the fused feature map.
示例性地，本步骤可以参见步骤S303，在此不再赘述。For example, this step can refer to step S303, which will not be described again here.
S605、根据预测的三维张量图和标准的三维张量图对图编码模型进行参数调整；其中，达到预设条件所得到的图编码模型，用于确定本实施例方法中的遥感卫星图像的三维张量图。S605. Adjust the parameters of the graph coding model according to the predicted three-dimensional tensor graph and the standard three-dimensional tensor graph; wherein, the graph coding model obtained by meeting the preset conditions is used to determine the remote sensing satellite image in the method of this embodiment. 3D tensor plot.
示例性的，响应于用户的标注操作，获取待训练的遥感卫星图像中的道路关键点；根据待训练的遥感卫星图像中的道路关键点，以及与道路关键点相邻的其他道路关键点，生成待训练的遥感卫星图像的道路编码信息；其中，道路编码信息包括待训练的遥感卫星图像的每一像素点的编码数据；根据待训练的遥感卫星图像的道路编码信息，生成待训练的遥感卫星图像的标准的三维张量图。For example, in response to the user's annotation operation, the road key points in the remote sensing satellite image to be trained are obtained; according to the road key points in the remote sensing satellite image to be trained and other road key points adjacent to the road key point, Generate road coding information of remote sensing satellite images to be trained; wherein the road coding information includes coding data of each pixel of the remote sensing satellite images to be trained; generate remote sensing remote sensing images to be trained based on the road coding information of the remote sensing satellite images to be trained A standard three-dimensional tensor plot of a satellite image.
本实施例中，用户对待训练的遥感卫星图像中的道路关键点进行标注，其中，待训练的遥感卫星图像中的道路关键点为不同道路转接处以及不同道路的端点。例如，图7为用户对待训练的遥感卫星图像中的道路关键点进行标注后的示意图，可以从图7中看出，道路关键点为A、B、C、D、E和F，从道路关键点A连接道路关键点B，由道路关键点B连接道路关键点C，道路关键点C同时连接道路关键点D和道路关键点E。道路关键点E连接道路关键点F。In this embodiment, the user annotates road key points in the remote sensing satellite images to be trained, where the road key points in the remote sensing satellite images to be trained are different road junctions and end points of different roads. For example, Figure 7 is a schematic diagram after the user annotates the key points of the road in the remote sensing satellite images to be trained. It can be seen from Figure 7 that the key points of the road are A, B, C, D, E and F. From the key points of the road Point A is connected to road key point B, road key point B is connected to road key point C, and road key point C is connected to road key point D and road key point E at the same time. Road key point E connects to road key point F.
进一步地，本实施例中，在确定了待训练的遥感卫星图像中的道路关键点后，能够确定与该道路关键点相邻的其他道路关键点。例如，初步确定的待训练的遥感卫星图像中的道路关键点为道路关键点C，则与该道路关键点相邻的其他道路关键点为道路关键点B、道路关键点D和道路关键点E。根据道路关键点C以及与该道路关键点相邻的其他道路关键点为道路关键点B、道路关键点D和道路关键点E生成道路关键点C的编码数据。这样设置的好处是用户标注操作后的遥感卫星图像不能够直接用于训练图编码模型，因此，需要将用户标注操作后的遥感卫星图像转化为能够训练图编码模型的数据形式。Further, in this embodiment, after the road key points in the remote sensing satellite images to be trained are determined, other road key points adjacent to the road key points can be determined. For example, the road key point in the remote sensing satellite image to be trained is initially determined to be road key point C, and the other road key points adjacent to the road key point are road key point B, road key point D, and road key point E. . The encoded data of the road key point C is generated for the road key point B, the road key point D and the road key point E based on the road key point C and other road key points adjacent to the road key point. The advantage of this setting is that the remote sensing satellite images after user annotation operations cannot be directly used to train the image coding model. Therefore, the remote sensing satellite images after user annotation operations need to be converted into a data form that can train the image coding model.
其中，待训练的遥感卫星图像的每一像素点的编码数据包括以下信息中的一种或多种：Among them, the coded data of each pixel of the remote sensing satellite image to be trained includes one or more of the following information:
待训练的遥感卫星图像中的每一像素点是否为道路关键点；Whether each pixel in the remote sensing satellite image to be trained is a road key point;
与作为道路关键点的像素点的相邻的其他道路关键点的个数；The number of other road key points adjacent to the pixel that is the road key point;
作为道路关键点的像素点、与作为道路关键点的像素点的相邻的其他道路关键点两者之间的距离信息。The distance information between a pixel that is a road key point and other adjacent road key points adjacent to the pixel that is a road key point.
本实施例中，编码数据包括待训练的遥感卫星图像中的每一像素点是否为道路关键点，或者与作为道路关键点的像素点的相邻的其他道路关键点的个数，或者作为道路关键点的像素点、与作为道路关键点的像素点的相邻的其他道路关键点两者之间的距离信息。In this embodiment, the encoding data includes whether each pixel in the remote sensing satellite image to be trained is a road key point, or the number of other road key points adjacent to the pixel that is a road key point, or whether it is a road key point. The distance information between the pixel of the key point and other adjacent road key points adjacent to the pixel that is the road key point.
编码数据还包括：待训练的遥感卫星图像中的每一像素点是否为道路关键点和与作为道路关键点的像素点的相邻的其他道路关键点的个数；与作为道路关键点的像素点的相邻的其他道路关键点的个数和作为道路关键点的像素点、与作为道路关键点的像素点的相邻的其他道路关键点两者之间的距离信息；待训练的遥感卫星图像中的每一像素点是否为道路关键点与作为道路关键点的像素点、与作为道路关键点的像素点的相邻的其他道路关键点两者之间的距离信息。The encoding data also includes: whether each pixel in the remote sensing satellite image to be trained is a road key point and the number of other road key points adjacent to the pixel as a road key point; and the number of other road key points adjacent to the pixel as a road key point; The number of other road key points adjacent to the point and the distance information between the pixel point as the road key point and other adjacent road key points as the pixel point as the road key point; the remote sensing satellite to be trained Whether each pixel in the image is a road key point, the distance information between the pixel as the road key point, and other adjacent road key points of the pixel as the road key point.
编码数据还包括：待训练的遥感卫星图像中的每一像素点是否为道路关键点、与作为道路关键点的像素点的相邻的其他道路关键点的个数和作为道路关键点的像素点、与作为道路关键点的像素点的相邻的其他道路关键点两者之间的距离信息。The encoding data also includes: whether each pixel in the remote sensing satellite image to be trained is a road key point, the number of other road key points adjacent to the pixel that is the road key point, and the pixel that is the road key point. , and the distance information between other road key points adjacent to the pixel point that is the road key point.
本实施例中，待训练的遥感卫星图像中的每一像素点是否为道路关键点可以用两位编码数据位进行表示。例如，该像素点为道路关键点时的编码数据位为10，该像素点不为道路关键点时的编码数据位为01。图7中的道路关键点A、道路关键点B、道路关键点C、道路关键点D和道路关键点E以及道路关键点F的其中两位编码数据位均可为10。In this embodiment, whether each pixel in the remote sensing satellite image to be trained is a road key point can be expressed by two coded data bits. For example, when the pixel is a key point on the road, the coded data bit is 10, and when the pixel is not a key point on the road, the coded data bit is 01. Each of the two encoding data bits of the road key point A, the road key point B, the road key point C, the road key point D, the road key point E and the road key point F in Figure 7 can be 10.
本实施例中，与作为道路关键点的像素点的相邻的其他道路关键点的个数的编码数据位为12位，这样设置的原因是道路关键点的像素点的相邻的其他道路关键点的个数一般最多是6个。本实施例中，与作为道路关键点的像素点的相邻的其他道路关键点的个数的编码数据位为12位仅是示例说明，该编码数据位是可以自行设置的，例如，还可以为20位。从图7中看出，与道路关键点A的像素点的相邻的其他道路关键点的个数为1，与道路关键点B的像素点的相邻的其他道路关键点的个数为2，与道路关键点C的像素点的相邻的其他道路关键点的个数为3，与道路关键点D的像素点的相邻的其他道路关键点的个数为1，与道路关键点E的像素点的相邻的其他道路关键点的个数为2，与道路关键点F的像素点的相邻的其他道路关键点的个数为1。在确定了道路关键点的像素点的相邻的其他道路关键点的个数，在以道路关键点的像素点为中心划分六个区域，具体的划分方式可以参见图8中示出的一种道路关键点的像素点的相邻的其他道路关键点的个数统计示意图。则可以在图8中的六个区域中依次确认，若第一个区域中存在道路关键点，则将编码数据位12位的前两位确认为10，若第二个区域中存在道路关键点，则将编码数据位12位的第三四位确认为10，依次类推。In this embodiment, the encoding data bits of the number of other road key points adjacent to the pixel point that is the road key point are 12 bits. The reason for this setting is that the number of other road key points adjacent to the pixel point of the road key point is The number of points is generally up to 6. In this embodiment, the encoding data bits of the number of other road key points adjacent to the pixel point that is the road key point is 12 bits. This is just an example. The encoding data bits can be set by yourself. For example, you can also for 20 digits. As can be seen from Figure 7, the number of other road key points adjacent to the pixel point of road key point A is 1, and the number of other road key points adjacent to the pixel point of road key point B is 2. , the number of other road key points adjacent to the pixel point of road key point C is 3, the number of other road key points adjacent to the pixel point of road key point D is 1, and the number of other road key points adjacent to the pixel point of road key point D is 1, and the number of other road key points adjacent to the pixel point of road key point D is 1. The number of other road key points adjacent to the pixel of F is 2, and the number of other road key points adjacent to the pixel of road key point F is 1. After determining the number of other road key points adjacent to the pixel of the road key point, six regions are divided with the pixel of the road key point as the center. The specific division method can be seen in Figure 8. Statistical diagram of the number of adjacent pixels of a road key point and other road key points. It can be confirmed in the six areas in Figure 8 in sequence. If there are road key points in the first area, the first two bits of the 12-bit coded data bits will be confirmed as 10. If there are road key points in the second area , then the third and fourth bits of the 12-bit coded data bits are confirmed to be 10, and so on.
作为道路关键点的像素点、与作为道路关键点的像素点的相邻的其他道路关键点两者之间的距离信息，具体为将待训练的遥感卫星图像按照坐标信息划分，确认每一个道路关键点的像素点的坐标信息，由每一个道路关键点的像素点的坐标信息与相邻的其他道路关键点的坐标信息确定两者之间的距离信息。其中，该编码数据位为12位，其中，每两位为一个相邻的其他道路关键点的距离信息。其中，该编码数据位的距离信息与道路关键点的个数的编码数据位是对应关系。The distance information between the pixel point as the key point of the road and other key points of the road adjacent to the pixel point as the key point of the road. Specifically, the remote sensing satellite image to be trained is divided according to the coordinate information, and each road is confirmed. The coordinate information of the pixel points of the key points is determined by the coordinate information of the pixel points of each road key point and the coordinate information of other adjacent road key points to determine the distance information between the two. The coded data bits are 12 bits, and each two bits is the distance information of an adjacent other road key point. The distance information of the encoded data bits is in a corresponding relationship with the encoded data bits of the number of road key points.
具体的，图9中示出了一种待训练的遥感卫星图像的每一像素点的编码数据以及三维张量图。从图9中可以看出，假设从三维张量图中确定一个像素点的编码数据，其中，该像素点为道路关键点C的像素点，该编码数据是由三个部分组成的，第一个部分为待训练的遥感卫星图像中的像素点是否为道路关键点的编码数据位，由于该像素点为道路关键点C的像素点，则第一个部分的编码数据位为10；第二个部分为与作为道路关键点的像素点的相邻的其他道路关键点的个数，由于该像素点为道路关键点C的像素点，与道路关键点C的像素点的相邻的其他道路关键点的个数为3，从图8中可以看出，道路关键点C在一个区域、第三个区域和第四个区域存在道路关键点，则第二个部分的编码数据位为100110100101；第三个部分为作为道路关键点的像素点、与作为道路关键点的像素点的相邻的其他道路关键点两者之间的距离信息，由于该像素点为道路关键点C的像素点，则计算道路关键点B、道路关键点D和道路关键点E分别与道路关键点C的距离信息。Specifically, Figure 9 shows the encoded data and three-dimensional tensor diagram of each pixel of a remote sensing satellite image to be trained. As can be seen from Figure 9, assuming that the encoded data of a pixel is determined from the three-dimensional tensor diagram, where the pixel is the pixel of the road key point C, the encoded data is composed of three parts. First The first part is the coded data bit of whether the pixel in the remote sensing satellite image to be trained is a road key point. Since the pixel is a pixel of the road key point C, the coded data bit of the first part is 10; is the number of other road key points adjacent to the pixel point that is the road key point. Since the pixel is a pixel point of road key point C, other roads adjacent to the pixel point of road key point C The number of key points is 3. As can be seen from Figure 8, road key point C has road key points in one area, the third area and the fourth area, then the encoded data bits of the second part are 100110100101; The third part is the distance information between the pixel point as the road key point and other adjacent road key points adjacent to the pixel point as the road key point. Since the pixel point is the pixel point of the road key point C, Then calculate the distance information between road key point B, road key point D and road key point E and road key point C respectively.
这样设置的好处是能够通过准确的数据形式准确地表征待训练的遥感卫星图像的道路关键点信息。The advantage of this setting is that the road key point information of the remote sensing satellite images to be trained can be accurately represented in an accurate data form.
本公开提供一种应用于地图更新的图编码模型的训练方法，包括：获取多个待训练的遥感卫星图像，待训练的遥感卫星图像具有标准的三维张量图，响应于用户的标注操作，获取待训练的遥感卫星图像中的道路关键点；根据待训练的遥感卫星图像中的道路关键点，以及与道路关键点相邻的其他道路关键点，生成待训练的遥感卫星图像的道路编码信息；其中，道路编码信息包括待训练的遥感卫星图像的每一像素点的编码数据；根据待训练的遥感卫星图像的道路编码信息，生成待训练的遥感卫星图像的标准的三维张量图。采用本技术方案，能够得到准确的图编码模型，进而输出比较准确的三维张量图，使得得到的第二拓扑图的召回率较高，保证地图的更新速度，优化作业过程。The present disclosure provides a training method for a graph coding model applied to map updating, which includes: acquiring a plurality of remote sensing satellite images to be trained, the remote sensing satellite images to be trained having a standard three-dimensional tensor graph, and in response to a user's annotation operation, Obtain the road key points in the remote sensing satellite image to be trained; generate the road coding information of the remote sensing satellite image to be trained based on the road key points in the remote sensing satellite image to be trained and other road key points adjacent to the road key point ; Among them, the road coding information includes the coding data of each pixel of the remote sensing satellite image to be trained; according to the road coding information of the remote sensing satellite image to be trained, a standard three-dimensional tensor diagram of the remote sensing satellite image to be trained is generated. Using this technical solution, an accurate graph coding model can be obtained, and then a relatively accurate three-dimensional tensor map can be output, so that the recall rate of the second topological map obtained is higher, ensuring the update speed of the map, and optimizing the operation process.
图10是根据本公开第五实施例的示意图，如图10所示，图10示出的是一种基于遥感卫星图像的地图更新装置，该装置10包括：Figure 10 is a schematic diagram according to the fifth embodiment of the present disclosure. As shown in Figure 10, Figure 10 shows a map updating device based on remote sensing satellite images. The device 10 includes:
获取单元1001，用于获取遥感卫星图像。Acquisition unit 1001, used to acquire remote sensing satellite images.
第一确定单元1002，用于对遥感卫星图像进行语义分割处理，得到第一拓扑图，第一拓扑图包括遥感卫星图像所对应的位置上的道路；并根据遥感卫星图像，确定三维张量图，三维张量图表征遥感卫星图像所对应的位置上的道路的编码信息。The first determination unit 1002 is used to perform semantic segmentation processing on remote sensing satellite images to obtain a first topological map. The first topological map includes roads at positions corresponding to remote sensing satellite images; and determine a three-dimensional tensor map based on the remote sensing satellite images. , the three-dimensional tensor diagram represents the encoded information of the road at the location corresponding to the remote sensing satellite image.
第二确定单元1003，用于对三维张量图进行解码处理，得到第二拓扑图，其中，第二拓扑图包括遥感卫星图像所对应的位置上的道路；The second determination unit 1003 is used to decode the three-dimensional tensor map to obtain a second topological map, where the second topological map includes roads at positions corresponding to remote sensing satellite images;
第三确定单元1004，用于根据第一拓扑图和第二拓扑图，确定道路拓扑图，并根据道路拓扑图更新地图。The third determination unit 1004 is configured to determine the road topology map based on the first topology map and the second topology map, and update the map based on the road topology map.
所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，上述描述装置的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the above-described device can be referred to the corresponding process in the foregoing method embodiment, and will not be described again here.
图11是根据本公开第六实施例的示意图，如图11所示，图11示出的是一种基于遥感卫星图像的地图更新装置，该装置11包括：Figure 11 is a schematic diagram according to the sixth embodiment of the present disclosure. As shown in Figure 11, Figure 11 shows a map updating device based on remote sensing satellite images. The device 11 includes:
获取单元1101，用于获取遥感卫星图像；Acquisition unit 1101, used to acquire remote sensing satellite images;
第一确定单元1102，用于对遥感卫星图像进行语义分割处理，得到第一拓扑图，第一拓扑图包括遥感卫星图像所对应的位置上的道路；并根据遥感卫星图像，确定三维张量图，三维张量图表征遥感卫星图像所对应的位置上的道路的编码信息。The first determination unit 1102 is used to perform semantic segmentation processing on remote sensing satellite images to obtain a first topological map. The first topological map includes roads at positions corresponding to remote sensing satellite images; and determine a three-dimensional tensor map based on the remote sensing satellite images. , the three-dimensional tensor diagram represents the encoded information of the road at the location corresponding to the remote sensing satellite image.
第二确定单元1103，用于对三维张量图进行解码处理，得到第二拓扑图，其中，第二拓扑图包括遥感卫星图像所对应的位置上的道路。The second determination unit 1103 is used to decode the three-dimensional tensor map to obtain a second topological map, where the second topological map includes roads at positions corresponding to remote sensing satellite images.
第三确定单元1104，用于根据第一拓扑图和第二拓扑图，确定道路拓扑图，并根据道路拓扑图更新地图。The third determination unit 1104 is configured to determine the road topology map based on the first topology map and the second topology map, and update the map based on the road topology map.
示例性地，其中，第一确定单元1102，包括：Illustratively, the first determining unit 1102 includes:
第一确定模块11021，用于将遥感卫星图像输入至图编码模型中，得到三维张量图；其中，图编码模型为基于具有标准的三维张量图的遥感卫星图像进行训练所得到的。The first determination module 11021 is used to input remote sensing satellite images into the graph coding model to obtain a three-dimensional tensor graph; wherein the graph encoding model is obtained by training based on remote sensing satellite images with standard three-dimensional tensor graphs.
示例性地，其中，第一确定模块11021，包括：Illustratively, the first determination module 11021 includes:
提取子模块110211，用于基于图编码模型对遥感卫星图像进行特征提取，得到全局特征图和局部特征图；其中，全局特征图表征遥感卫星图像的全局特征，局部特征图表征遥感卫星图像的道路特征。Extraction sub-module 110211 is used to extract features from remote sensing satellite images based on the graph coding model to obtain global feature maps and local feature maps; among them, the global feature map represents the global features of the remote sensing satellite image, and the local feature map represents the road of the remote sensing satellite image. feature.
融合子模块110212，用于基于图编码模型对全局特征图和局部特征图进行特征融合，得到融合后的特征图。The fusion sub-module 110212 is used to perform feature fusion on the global feature map and the local feature map based on the graph coding model to obtain the fused feature map.
生成子模块110213，用于根据融合后的特征图，生成三维张量图。The generation submodule 110213 is used to generate a three-dimensional tensor map based on the fused feature map.
示例性地，其中，提取子模块110211，包括：Exemplarily, the extraction sub-module 110211 includes:
基于图编码模型对遥感卫星图像进行特征提取，得到全局特征图；Based on the graph coding model, feature extraction is performed on remote sensing satellite images to obtain global feature maps;
对全局特征图进行二值化处理，得到二值化的特征图，二值化的特征图中包括道路的特征。The global feature map is binarized to obtain a binarized feature map, which includes the characteristics of the road.
基于二值化的特征图中的道路点，在全局特征图中确定与道路点对应的道路位置区域；并根据与道路点对应的道路位置区域，生成局部特征图。Based on the road points in the binarized feature map, the road location area corresponding to the road point is determined in the global feature map; and a local feature map is generated based on the road location area corresponding to the road point.
示例性地，其中，融合子模块110212，包括：Illustratively, the fusion sub-module 110212 includes:
对局部特征图进行上采样处理，得到上采样后的局部特征图；上采样后的局部特征图的尺寸与全局特征图的尺寸相同。The local feature map is upsampled to obtain the upsampled local feature map; the size of the upsampled local feature map is the same as the size of the global feature map.
基于图编码模型对全局特征图和上采样后的局部特征图进行特征融合，得到融合后的特征图。Based on the graph coding model, the global feature map and the upsampled local feature map are feature fused to obtain the fused feature map.
示例性地，其中，第三确定单元1104，包括：Illustratively, the third determining unit 1104 includes:
加入模块11041，用于若确定第一拓扑图中的道路像素点，不存在于第二拓扑图中，则将第一拓扑图中的道路像素点，加入到第二拓扑图中，以生成道路拓扑图。Adding module 11041 is used to add the road pixels in the first topology map to the second topology map if it is determined that the road pixels in the first topology map do not exist in the second topology map to generate roads. Topology.
示例性地，其中，装置还包括：Exemplarily, the device further includes:
处理单元1105，用于对道路拓扑图进行图像增强处理，得到增强处理后的道路拓扑图。The processing unit 1105 is configured to perform image enhancement processing on the road topology map to obtain an enhanced road topology map.
示例性地，其中，第一拓扑图为二值图，第二拓扑图为二值图。For example, the first topological graph is a binary graph, and the second topological graph is a binary graph.
所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，上述描述装置的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the above-described device can be referred to the corresponding process in the foregoing method embodiment, and will not be described again here.
图12是根据本公开第七实施例的示意图，如图12所示，图12示出的是一种应用于地图更新的图编码模型的训练装置，该装置12包括：Figure 12 is a schematic diagram according to the seventh embodiment of the present disclosure. As shown in Figure 12, Figure 12 shows a training device for a graph coding model applied to map update. The device 12 includes:
第一获取单元1201，用于获取多个待训练的遥感卫星图像，待训练的遥感卫星图像具有标准的三维张量图；The first acquisition unit 1201 is used to acquire multiple remote sensing satellite images to be trained. The remote sensing satellite images to be trained have a standard three-dimensional tensor map;
第一确定单元1202，用于重复以下各步骤，直至达到预设条件：将待训练的遥感卫星图像输入至图编码模型中，得到预测的三维张量图，预测的三维张量图表征待训练的遥感卫星图像所对应的位置上的道路的编码信息；根据预测的三维张量图和标准的三维张量图对图编码模型进行参数调整；The first determination unit 1202 is used to repeat the following steps until the preset conditions are reached: input the remote sensing satellite image to be trained into the image coding model to obtain a predicted three-dimensional tensor map, and the predicted three-dimensional tensor map represents the to-be-trained The encoding information of the road at the location corresponding to the remote sensing satellite image; adjust the parameters of the graph encoding model based on the predicted three-dimensional tensor graph and the standard three-dimensional tensor graph;
其中，达到预设条件所得到的图编码模型，用于确定权利要求15-22任一项装置中的遥感卫星图像的三维张量图。Among them, the image coding model obtained by meeting the preset conditions is used to determine the three-dimensional tensor image of the remote sensing satellite image in the device of any one of claims 15-22.
所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，上述描述装置的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the above-described device can be referred to the corresponding process in the foregoing method embodiment, and will not be described again here.
图13是根据本公开第八实施例的示意图，如图13所示，图13示出的是一种应用于地图更新的图编码模型的训练装置，该装置13包括：Figure 13 is a schematic diagram according to the eighth embodiment of the present disclosure. As shown in Figure 13, Figure 13 shows a training device for a graph coding model applied to map update. The device 13 includes:
第一获取单元1301，用于获取多个待训练的遥感卫星图像，待训练的遥感卫星图像具有标准的三维张量图。The first acquisition unit 1301 is used to acquire multiple remote sensing satellite images to be trained. The remote sensing satellite images to be trained have a standard three-dimensional tensor map.
第一确定单元1302，用于重复以下各步骤，直至达到预设条件：将待训练的遥感卫星图像输入至图编码模型中，得到预测的三维张量图，预测的三维张量图表征待训练的遥感卫星图像所对应的位置上的道路的编码信息；根据预测的三维张量图和标准的三维张量图对图编码模型进行参数调整。The first determination unit 1302 is used to repeat the following steps until the preset conditions are reached: input the remote sensing satellite image to be trained into the image coding model to obtain a predicted three-dimensional tensor map, and the predicted three-dimensional tensor map represents the to-be-trained The encoding information of the road at the location corresponding to the remote sensing satellite image; adjust the parameters of the graph encoding model based on the predicted three-dimensional tensor graph and the standard three-dimensional tensor graph.
其中，达到预设条件所得到的图编码模型，用于确定本实施例中装置中的遥感卫星图像的三维张量图。Among them, the graph coding model obtained by meeting the preset conditions is used to determine the three-dimensional tensor graph of the remote sensing satellite image in the device in this embodiment.
示例性地，其中，第一确定单元1302，包括：Illustratively, the first determining unit 1302 includes:
提取模块13021，用于基于图编码模型对待训练的遥感卫星图像进行特征提取，得到全局特征图和局部特征图；其中，全局特征图表征待训练的遥感卫星图像的全局特征，局部特征图表征待训练的遥感卫星图像的道路特征。The extraction module 13021 is used to extract features of the remote sensing satellite images to be trained based on the graph coding model to obtain global feature maps and local feature maps; among which, the global feature map represents the global features of the remote sensing satellite images to be trained, and the local feature map represents the global features of the remote sensing satellite images to be trained. Road features from trained remote sensing satellite images.
确定模块13022，用于基于图编码模型对全局特征图和局部特征图进行特征融合，得到融合后的特征图。The determination module 13022 is used to perform feature fusion on the global feature map and the local feature map based on the graph coding model to obtain the fused feature map.
生成模块13023，用于根据融合后的特征图，生成三维张量图。The generation module 13023 is used to generate a three-dimensional tensor map based on the fused feature map.
示例性地，其中，提取模块13021，包括：Exemplarily, the extraction module 13021 includes:
提取子模块130211，用于基于图编码模型对待训练的遥感卫星图像进行特征提取，得到全局特征图。The extraction sub-module 130211 is used to extract features from the remote sensing satellite images to be trained based on the graph coding model to obtain the global feature map.
处理子模块130212，用于对全局特征图进行二值化处理，得到二值化的特征图，二值化的特征图中包括道路的特征。The processing sub-module 130212 is used to perform binarization processing on the global feature map to obtain a binarized feature map. The binarized feature map includes the characteristics of the road.
生成子模块130213，用于基于二值化的特征图中的道路点，在全局特征图中确定与道路点对应的道路位置区域；并根据与道路点对应的道路位置区域，生成局部特征图。The generation submodule 130213 is used to determine the road location area corresponding to the road point in the global feature map based on the road point in the binarized feature map; and generate a local feature map based on the road location area corresponding to the road point.
示例性地，其中，确定模块13022，包括：Illustratively, the determination module 13022 includes:
处理子模块130221，用于对局部特征图进行上采样处理，得到上采样后的局部特征图；上采样后的局部特征图的尺寸与全局特征图的尺寸相同。The processing sub-module 130221 is used to upsample the local feature map to obtain the upsampled local feature map; the size of the upsampled local feature map is the same as the size of the global feature map.
融合子模块130222，用于基于图编码模型对全局特征图和上采样后的局部特征图进行特征融合，得到融合后的特征图。The fusion sub-module 130222 is used to perform feature fusion on the global feature map and the upsampled local feature map based on the graph coding model to obtain the fused feature map.
示例性地，还包括：Examples include:
第二获取单元1303，用于响应于用户的标注操作，获取待训练的遥感卫星图像中的道路关键点。The second acquisition unit 1303 is configured to acquire road key points in the remote sensing satellite images to be trained in response to the user's annotation operation.
第一生成单元1304，用于根据待训练的遥感卫星图像中的道路关键点，以及与道路关键点相邻的其他道路关键点，生成待训练的遥感卫星图像的道路编码信息；其中，道路编码信息包括待训练的遥感卫星图像的每一像素点的编码数据。The first generation unit 1304 is used to generate road coding information of the remote sensing satellite image to be trained based on the road key points in the remote sensing satellite image to be trained and other road key points adjacent to the road key points; wherein, the road coding information The information includes the encoded data of each pixel of the remote sensing satellite image to be trained.
第二生成单元1305，用于根据待训练的遥感卫星图像的道路编码信息，生成待训练的遥感卫星图像的标准的三维张量图。The second generation unit 1305 is configured to generate a standard three-dimensional tensor diagram of the remote sensing satellite image to be trained based on the road coding information of the remote sensing satellite image to be trained.
示例性地，还包括：其中，待训练的遥感卫星图像的每一像素点的编码数据包括以下信息中的一种或多种：待训练的遥感卫星图像中的每一像素点是否为道路关键点；与作为道路关键点的像素点的相邻的其他道路关键点的个数；作为道路关键点的像素点、与作为道路关键点的像素点的相邻的其他道路关键点两者之间的距离信息。Exemplarily, it also includes: wherein the encoded data of each pixel of the remote sensing satellite image to be trained includes one or more of the following information: whether each pixel of the remote sensing satellite image to be trained is a road key. point; the number of other road key points adjacent to the pixel point as the road key point; the number of other road key points adjacent to the pixel point as the road key point and the other adjacent road key points to the pixel point as the road key point distance information.
所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，上述描述装置的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the above-described device can be referred to the corresponding process in the foregoing method embodiment, and will not be described again here.
根据本公开的实施例，本公开还提供了一种计算机设备、一种可读存储介质和一种计算机程序产品。According to embodiments of the present disclosure, the present disclosure also provides a computer device, a readable storage medium, and a computer program product.
根据本公开的实施例，本公开还提供了一种计算机程序产品，计算机程序产品包括：计算机程序，计算机程序存储在可读存储介质中，计算机设备的至少一个处理器可以从可读存储介质读取计算机程序，至少一个处理器执行计算机程序使得计算机设备执行上述任一实施例提供的方案。According to an embodiment of the present disclosure, the present disclosure also provides a computer program product. The computer program product includes: a computer program. The computer program is stored in a readable storage medium. At least one processor of the computer device can read from the readable storage medium. Taking a computer program, at least one processor executes the computer program so that the computer device executes the solution provided by any of the above embodiments.
图14是根据本公开第九实施例的示意图，如图14所示，本公开中的服务器1400可以包括：处理器1401和存储器1402。Figure 14 is a schematic diagram according to the ninth embodiment of the present disclosure. As shown in Figure 14, the server 1400 in the present disclosure may include: a processor 1401 and a memory 1402.
存储器1402，用于存储程序；存储器1402，可以包括易失性存储器(英文：volatilememory)，例如随机存取存储器(英文：random-access memory，缩写：RAM)，如静态随机存取存储器(英文：static random-access memory，缩写：SRAM)，双倍数据率同步动态随机存取存储器(英文：Double Data Rate Synchronous Dynamic Random Access Memory，缩写：DDR SDRAM)等；存储器也可以包括非易失性存储器(英文：non-volatile memory)，例如快闪存储器(英文：flash memory)。存储器1402用于存储计算机程序(如实现上述方法的应用程序、功能模块等)、计算机指令等，上述的计算机程序、计算机指令等可以分区存储在一个或多个存储器1402中。并且上述的计算机程序、计算机指令、数据等可以被处理器1401调用。Memory 1402 is used to store programs; memory 1402 may include volatile memory (English: volatile memory), such as random access memory (English: random-access memory, abbreviation: RAM), such as static random access memory (English: static random access memory) static random-access memory (abbreviation: SRAM), Double Data Rate Synchronous Dynamic Random Access Memory (English: Double Data Rate Synchronous Dynamic Random Access Memory, abbreviation: DDR SDRAM), etc.; the memory can also include non-volatile memory ( English: non-volatile memory), such as flash memory (English: flash memory). The memory 1402 is used to store computer programs (such as application programs, functional modules, etc. that implement the above methods), computer instructions, etc. The above-mentioned computer programs, computer instructions, etc. can be stored in one or more memories 1402 in partitions. And the above-mentioned computer programs, computer instructions, data, etc. can be called by the processor 1401.
上述的计算机程序、计算机指令等可以分区存储在一个或多个存储器1402中。并且上述的计算机程序、计算机指据等可以被处理器1401调用。The above-described computer programs, computer instructions, etc. may be partitioned and stored in one or more memories 1402. And the above-mentioned computer programs, computer instructions, etc. can be called by the processor 1401.
处理器1401，用于执行存储器1402存储的计算机程序，以实现上述实施例涉及的基于遥感卫星图像的地图更新方法中的各个步骤。The processor 1401 is configured to execute the computer program stored in the memory 1402 to implement various steps in the map update method based on remote sensing satellite images involved in the above embodiment.
具体可以参见前面方法实施例中的相关描述。For details, please refer to the relevant descriptions in the previous method embodiments.
处理器1401和存储器1402可以是独立结构，也可以是集成在一起的集成结构。当处理器1401和存储器1402是独立结构时，存储器1402、处理器1401可以通过总线1403耦合连接。The processor 1401 and the memory 1402 may be independent structures or may be an integrated structure integrated together. When the processor 1401 and the memory 1402 are independent structures, the memory 1402 and the processor 1401 can be coupled and connected through the bus 1403.
本实施例的服务器可以执行上述方法中的技术方案，其具体实现过程和技术原理相同，此处不再赘述。The server in this embodiment can execute the technical solution in the above method. The specific implementation process and technical principles are the same and will not be described again here.
根据本公开的实施例，本公开还提供了一种存储有计算机指令的非瞬时计算机可读存储介质，其中，计算机指令用于使计算机执行上述相应的实施例提供的方案。According to an embodiment of the present disclosure, the present disclosure also provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to execute the solutions provided by the above corresponding embodiments.
根据本公开的实施例，本公开还提供了一种计算机程序产品，计算机程序产品包括：计算机程序，计算机程序存储在可读存储介质中，服务器的至少一个处理器可以从可读存储介质读取计算机程序，至少一个处理器执行计算机程序使得服务器执行上述相应的实施例提供的方案。According to an embodiment of the present disclosure, the present disclosure also provides a computer program product. The computer program product includes: a computer program. The computer program is stored in a readable storage medium. At least one processor of the server can read from the readable storage medium. Computer program, at least one processor executes the computer program to cause the server to execute the solutions provided by the above corresponding embodiments.
根据本公开的实施例，本公开还提供了一种计算机程序产品，计算机程序产品包括：计算机程序，计算机程序存储在可读存储介质中，车辆的控制设备的至少一个处理器可以从可读存储介质读取计算机程序，至少一个处理器执行计算机程序使得车辆的控制设备执行上述相应的实施例提供的方案。According to an embodiment of the present disclosure, the present disclosure also provides a computer program product. The computer program product includes: a computer program. The computer program is stored in a readable storage medium. At least one processor of the control device of the vehicle can read from the readable storage medium. The medium reads the computer program, and at least one processor executes the computer program so that the control device of the vehicle executes the solution provided by the above-mentioned corresponding embodiment.
图15是根据本公开第十实施例的示意图，如图15所示，本公开中的服务器1500可以包括：处理器1501和存储器1502。Figure 15 is a schematic diagram according to the tenth embodiment of the present disclosure. As shown in Figure 15, the server 1500 in the present disclosure may include: a processor 1501 and a memory 1502.
存储器1502，用于存储程序；存储器1502，可以包括易失性存储器(英文：volatilememory)，例如随机存取存储器(英文：random-access memory，缩写：RAM)，如静态随机存取存储器(英文：static random-access memory，缩写：SRAM)，双倍数据率同步动态随机存取存储器(英文：Double Data Rate Synchronous Dynamic Random Access Memory，缩写：DDR SDRAM)等；存储器也可以包括非易失性存储器(英文：non-volatile memory)，例如快闪存储器(英文：flash memory)。存储器1502用于存储计算机程序(如实现上述方法的应用程序、功能模块等)、计算机指令等，上述的计算机程序、计算机指令等可以分区存储在一个或多个存储器1502中。并且上述的计算机程序、计算机指令、数据等可以被处理器1501调用。Memory 1502 is used to store programs; memory 1502 may include volatile memory (English: volatile memory), such as random access memory (English: random-access memory, abbreviation: RAM), such as static random access memory (English: static random access memory) static random-access memory (abbreviation: SRAM), Double Data Rate Synchronous Dynamic Random Access Memory (English: Double Data Rate Synchronous Dynamic Random Access Memory, abbreviation: DDR SDRAM), etc.; the memory can also include non-volatile memory ( English: non-volatile memory), such as flash memory (English: flash memory). The memory 1502 is used to store computer programs (such as application programs, functional modules, etc. that implement the above methods), computer instructions, etc. The above-mentioned computer programs, computer instructions, etc. can be stored in one or more memories 1502 in partitions. And the above-mentioned computer programs, computer instructions, data, etc. can be called by the processor 1501.
上述的计算机程序、计算机指令等可以分区存储在一个或多个存储器1502中。并且上述的计算机程序、计算机指据等可以被处理器1501调用。The above-described computer programs, computer instructions, etc. may be partitioned and stored in one or more memories 1502. And the above-mentioned computer programs, computer instructions, etc. can be called by the processor 1501.
处理器1501，用于执行存储器1502存储的计算机程序，以实现上述实施例涉及的应用于地图更新的图编码模型的训练方法中的各个步骤。The processor 1501 is configured to execute the computer program stored in the memory 1502 to implement each step in the training method of the graph coding model applied to map updating involved in the above embodiment.
具体可以参见前面方法实施例中的相关描述。For details, please refer to the relevant descriptions in the previous method embodiments.
处理器1501和存储器1502可以是独立结构，也可以是集成在一起的集成结构。当处理器1501和存储器1502是独立结构时，存储器1502、处理器1501可以通过总线1503耦合连接。The processor 1501 and the memory 1502 may be independent structures or may be an integrated structure integrated together. When the processor 1501 and the memory 1502 are independent structures, the memory 1502 and the processor 1501 can be coupled and connected through the bus 1503.
本实施例的服务器可以执行上述方法中的技术方案，其具体实现过程和技术原理相同，此处不再赘述。The server in this embodiment can execute the technical solution in the above method. The specific implementation process and technical principles are the same and will not be described again here.
根据本公开的实施例，本公开还提供了一种存储有计算机指令的非瞬时计算机可读存储介质，其中，计算机指令用于使计算机执行上述相应的实施例提供的方案。According to an embodiment of the present disclosure, the present disclosure also provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to execute the solutions provided by the above corresponding embodiments.
根据本公开的实施例，本公开还提供了一种计算机程序产品，计算机程序产品包括：计算机程序，计算机程序存储在可读存储介质中，服务器的至少一个处理器可以从可读存储介质读取计算机程序，至少一个处理器执行计算机程序使得服务器执行上述相应的实施例提供的方案。According to an embodiment of the present disclosure, the present disclosure also provides a computer program product. The computer program product includes: a computer program. The computer program is stored in a readable storage medium. At least one processor of the server can read from the readable storage medium. Computer program, at least one processor executes the computer program to cause the server to execute the solutions provided by the above corresponding embodiments.
根据本公开的实施例，本公开还提供了一种计算机程序产品，计算机程序产品包括：计算机程序，计算机程序存储在可读存储介质中，车辆的控制设备的至少一个处理器可以从可读存储介质读取计算机程序，至少一个处理器执行计算机程序使得车辆的控制设备执行上述相应的实施例提供的方案。According to an embodiment of the present disclosure, the present disclosure also provides a computer program product. The computer program product includes: a computer program. The computer program is stored in a readable storage medium. At least one processor of the control device of the vehicle can read from the readable storage medium. The medium reads the computer program, and at least one processor executes the computer program so that the control device of the vehicle executes the solution provided by the above-mentioned corresponding embodiment.
图16示出了可以用来实施本公开的实施例的示例计算机设备1600的示意性框图。计算机设备旨在表示各种形式的数字计算机，诸如，膝上型计算机、台式计算机、工作台、个人数字助理、计算机设备、刀片式计算机设备、大型计算机、和其它适合的计算机。计算机设备还可以表示各种形式的移动装置，诸如，个人数字处理、蜂窝电话、智能电话、可穿戴设备和其它类似的计算装置。本文所示的部件、它们的连接和关系、以及它们的功能仅仅作为示例，并且不意在限制本文中描述的和/或者要求的本公开的实现。16 illustrates a schematic block diagram of an example computer device 1600 that may be used to implement embodiments of the present disclosure. Computer equipment is intended to mean various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, computer equipment, blade computer equipment, mainframe computers, and other suitable computers. Computer devices may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are examples only and are not intended to limit implementations of the disclosure described and/or claimed herein.
如图16所示，计算机设备1600包括计算单元1601，其可以根据存储在只读存储器(ROM)1602中的计算机程序或者从存储单元1608加载到随机访问存储器(RAM)1603中的计算机程序，来执行各种适当的动作和处理。在RAM 1603中，还可存储设备1600操作所需的各种程序和数据。计算单元1601、ROM 1602以及RAM 1603通过总线1604彼此相连。输入/输出(I/O)接口1605也连接至总线1604。As shown in FIG. 16 , the computer device 1600 includes a computing unit 1601 that can perform calculations based on a computer program stored in a read-only memory (ROM) 1602 or loaded from a storage unit 1608 into a random access memory (RAM) 1603 . Perform various appropriate actions and processing. In the RAM 1603, various programs and data required for the operation of the device 1600 may also be stored. Computing unit 1601, ROM 1602 and RAM 1603 are connected to each other via bus 1604. An input/output (I/O) interface 1605 is also connected to bus 1604.
计算机设备1600中的多个部件连接至I/O接口1605，包括：输入单元1606，例如键盘、鼠标等；输出单元1607，例如各种类型的显示器、扬声器等；存储单元1608，例如磁盘、光盘等；以及通信单元1609，例如网卡、调制解调器、无线通信收发机等。通信单元1609允许计算机设备1600通过诸如因特网的计算机网络和/或各种电信网络与其他设备交换信息/数据。Multiple components in the computer device 1600 are connected to the I/O interface 1605, including: input unit 1606, such as keyboard, mouse, etc.; output unit 1607, such as various types of displays, speakers, etc.; storage unit 1608, such as magnetic disk, optical disk etc.; and communication unit 1609, such as network card, modem, wireless communication transceiver, etc. The communication unit 1609 allows the computer device 1600 to exchange information/data with other devices over a computer network such as the Internet and/or various telecommunications networks.
计算单元1601可以是各种具有处理和计算能力的通用和/或专用处理组件。计算单元1601的一些示例包括但不限于中央处理单元(CPU)、图形处理单元(GPU)、各种专用的人工智能(AI)计算芯片、各种运行机器学习模型算法的计算单元、数字信号处理器(DSP)、以及任何适当的处理器、控制器、微控制器等。计算单元1601执行上文所描述的各个方法和处理，例如用于生成高精地图的定位信息的处理方法。例如，在一些实施例中，方法用于图像处理的模型训练可被实现为计算机软件程序，其被有形地包含于机器可读介质，例如存储单元1608。在一些实施例中，计算机程序的部分或者全部可以经由ROM 1602和/或通信单元1609而被载入和/或安装到计算机设备1600上。当计算机程序加载到RAM 1603并由计算单元1601执行时，可以执行上文描述的基于遥感卫星图像的地图更新方法的一个或多个步骤。备选地，在其他实施例中，计算单元1601可以通过其他任何适当的方式(例如，借助于固件)而被配置为执行方法用于图像处理的模型训练。Computing unit 1601 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of computing units 1601 include, but are not limited to, central processing units (CPUs), graphics processing units (GPUs), various dedicated artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, digital signal processing processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 1601 performs various methods and processes described above, such as a processing method for generating positioning information of a high-precision map. For example, in some embodiments, methods for model training for image processing may be implemented as a computer software program that is tangibly embodied in a machine-readable medium, such as storage unit 1608 . In some embodiments, part or all of a computer program may be loaded and/or installed onto computer device 1600 via ROM 1602 and/or communication unit 1609 . When the computer program is loaded into the RAM 1603 and executed by the computing unit 1601, one or more steps of the map updating method based on remote sensing satellite images described above may be performed. Alternatively, in other embodiments, the computing unit 1601 may be configured to perform the method for model training for image processing in any other suitable manner (eg, by means of firmware).
本文中以上描述的系统和技术的各种实施方式可以在数字电子电路系统、集成电路系统、场可编程门阵列(FPGA)、专用集成电路(ASIC)、专用标准产品(ASSP)、芯片上系统的系统(SOC)、负载可编程逻辑设备(CPLD)、计算机硬件、固件、软件、和/或它们的组合中实现。这些各种实施方式可以包括：实施在一个或者多个计算机程序中，该一个或者多个计算机程序可在包括至少一个可编程处理器的可编程系统上执行和/或解释，该可编程处理器可以是专用或者通用可编程处理器，可以从存储系统、至少一个输入装置、和至少一个输出装置接收数据和指令，并且将数据和指令传输至该存储系统、该至少一个输入装置、和该至少一个输出装置。Various implementations of the systems and techniques described above may be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip implemented in a system (SOC), load programmable logic device (CPLD), computer hardware, firmware, software, and/or a combination thereof. These various embodiments may include implementation in one or more computer programs executable and/or interpreted on a programmable system including at least one programmable processor, the programmable processor The processor, which may be a special purpose or general purpose programmable processor, may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device. An output device.
用于实施本公开的方法的程序代码可以采用一个或多个编程语言的任何组合来编写。这些程序代码可以提供给通用计算机、专用计算机或其他可编程数据处理装置的处理器或控制器，使得程序代码当由处理器或控制器执行时使流程图和/或框图中所规定的功能/操作被实施。程序代码可以完全在机器上执行、部分地在机器上执行，作为独立软件包部分地在机器上执行且部分地在远程机器上执行或完全在远程机器或计算机设备上执行。Program code for implementing the methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that the program codes, when executed by the processor or controller, cause the functions specified in the flowcharts and/or block diagrams/ The operation is implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or computer device.
在本公开的上下文中，机器可读介质可以是有形的介质，其可以包含或存储以供指令执行系统、装置或设备使用或与指令执行系统、装置或设备结合地使用的程序。机器可读介质可以是机器可读信号介质或机器可读储存介质。机器可读介质可以包括但不限于电子的、磁性的、光学的、电磁的、红外的、或半导体系统、装置或设备，或者上述内容的任何合适组合。机器可读存储介质的更具体示例会包括基于一个或多个线的电气连接、便携式计算机盘、硬盘、随机存取存储器(RAM)、只读存储器(ROM)、可擦除可编程只读存储器(EPROM或快闪存储器)、光纤、便捷式紧凑盘只读存储器(CD-ROM)、光学储存设备、磁储存设备、或上述内容的任何合适组合。In the context of this disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media may include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media would include one or more wire-based electrical connections, laptop disks, hard drives, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above.
为了提供与用户的交互，可以在计算机上实施此处描述的系统和技术，该计算机具有：用于向用户显示信息的显示装置(例如，CRT(阴极射线管)或者LCD(液晶显示器)监视器)；以及键盘和指向装置(例如，鼠标或者轨迹球)，用户可以通过该键盘和该指向装置来将输入提供给计算机。其它种类的装置还可以用于提供与用户的交互；例如，提供给用户的反馈可以是任何形式的传感反馈(例如，视觉反馈、听觉反馈、或者触觉反馈)；并且可以用任何形式(包括声输入、语音输入或者、触觉输入)来接收来自用户的输入。To provide interaction with a user, the systems and techniques described herein may be implemented on a computer having a display device (eg, a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user ); and a keyboard and pointing device (eg, a mouse or a trackball) through which a user can provide input to the computer. Other kinds of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and may be provided in any form, including Acoustic input, voice input or tactile input) to receive input from the user.
可以将此处描述的系统和技术实施在包括后台部件的计算系统(例如，作为数据计算机设备)、或者包括中间件部件的计算系统(例如，应用计算机设备)、或者包括前端部件的计算系统(例如，具有图形用户界面或者网络浏览器的用户计算机，用户可以通过该图形用户界面或者该网络浏览器来与此处描述的系统和技术的实施方式交互)、或者包括这种后台部件、中间件部件、或者前端部件的任何组合的计算系统中。可以通过任何形式或者介质的数字数据通信(例如，通信网络)来将系统的部件相互连接。通信网络的示例包括：局域网(LAN)、广域网(WAN)和互联网。The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data computing device), or a computing system that includes middleware components (e.g., an application computing device), or a computing system that includes front-end components (e.g., as a data computing device). For example, a user computer having a graphical user interface or a web browser through which the user can interact with implementations of the systems and technologies described herein), or including such backend components, middleware components, or any combination of front-end components in a computing system. The components of the system may be interconnected by any form or medium of digital data communication (eg, a communications network). Examples of communication networks include: local area network (LAN), wide area network (WAN), and the Internet.
计算机系统可以包括客户端和计算机设备。客户端和计算机设备一般远离彼此并且通常通过通信网络进行交互。通过在相应的计算机上运行并且彼此具有客户端-计算机设备关系的计算机程序来产生客户端和计算机设备的关系。计算机设备可以是云计算机设备，又称为云计算计算机设备或云主机，是云计算服务体系中的一项主机产品，以解决了传统物理主机与VPS服务("Virtual Private Server"，或简称"VPS")中，存在的管理难度大，业务扩展性弱的缺陷。计算机设备也可以为分布式系统的计算机设备，或者是结合了区块链的计算机设备。Computer systems may include clients and computer devices. Clients and computer devices are generally remote from each other and often interact through a communications network. The relationship of client and computer device is created by computer programs running on the respective computers and having a client-computer device relationship with each other. Computer equipment can be cloud computer equipment, also known as cloud computing computer equipment or cloud hosts. It is a host product in the cloud computing service system to solve the problem of traditional physical hosts and VPS services ("Virtual Private Server", or "Virtual Private Server" for short) VPS") has the disadvantages of difficult management and weak business scalability. The computer equipment may also be a computer equipment of a distributed system, or a computer equipment combined with a blockchain.
图17示出了可以用来实施本公开的实施例的示例计算机设备1700的示意性框图。计算机设备旨在表示各种形式的数字计算机，诸如，膝上型计算机、台式计算机、工作台、个人数字助理、计算机设备、刀片式计算机设备、大型计算机、和其它适合的计算机。计算机设备还可以表示各种形式的移动装置，诸如，个人数字处理、蜂窝电话、智能电话、可穿戴设备和其它类似的计算装置。本文所示的部件、它们的连接和关系、以及它们的功能仅仅作为示例，并且不意在限制本文中描述的和/或者要求的本公开的实现。17 illustrates a schematic block diagram of an example computer device 1700 that may be used to implement embodiments of the present disclosure. Computer equipment is intended to mean various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, computer equipment, blade computer equipment, mainframe computers, and other suitable computers. Computer devices may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are examples only and are not intended to limit implementations of the disclosure described and/or claimed herein.
如图17所示，计算机设备1700包括计算单元1701，其可以根据存储在只读存储器(ROM)1702中的计算机程序或者从存储单元1708加载到随机访问存储器(RAM)1703中的计算机程序，来执行各种适当的动作和处理。在RAM 1703中，还可存储设备1700操作所需的各种程序和数据。计算单元1701、ROM 1702以及RAM 1703通过总线1704彼此相连。输入/输出(I/O)接口1705也连接至总线1704。As shown in FIG. 17 , the computer device 1700 includes a computing unit 1701 that can perform calculations based on a computer program stored in a read-only memory (ROM) 1702 or loaded from a storage unit 1708 into a random access memory (RAM) 1703 . Perform various appropriate actions and processing. In the RAM 1703, various programs and data required for the operation of the device 1700 may also be stored. The computing unit 1701, the ROM 1702 and the RAM 1703 are connected to each other via a bus 1704. An input/output (I/O) interface 1705 is also connected to bus 1704.
计算机设备1700中的多个部件连接至I/O接口1705，包括：输入单元1706，例如键盘、鼠标等；输出单元1707，例如各种类型的显示器、扬声器等；存储单元1708，例如磁盘、光盘等；以及通信单元1709，例如网卡、调制解调器、无线通信收发机等。通信单元1709允许计算机设备1700通过诸如因特网的计算机网络和/或各种电信网络与其他设备交换信息/数据。Multiple components in the computer device 1700 are connected to the I/O interface 1705, including: input unit 1706, such as keyboard, mouse, etc.; output unit 1707, such as various types of displays, speakers, etc.; storage unit 1708, such as magnetic disk, optical disk etc.; and communication unit 1709, such as network card, modem, wireless communication transceiver, etc. The communication unit 1709 allows the computer device 1700 to exchange information/data with other devices over a computer network such as the Internet and/or various telecommunications networks.
计算单元1701可以是各种具有处理和计算能力的通用和/或专用处理组件。计算单元1701的一些示例包括但不限于中央处理单元(CPU)、图形处理单元(GPU)、各种专用的人工智能(AI)计算芯片、各种运行机器学习模型算法的计算单元、数字信号处理器(DSP)、以及任何适当的处理器、控制器、微控制器等。计算单元1701执行上文所描述的各个方法和处理，例如用于生成高精地图的定位信息的处理方法。例如，在一些实施例中，方法用于图像处理的模型训练可被实现为计算机软件程序，其被有形地包含于机器可读介质，例如存储单元1708。在一些实施例中，计算机程序的部分或者全部可以经由ROM 1702和/或通信单元1709而被载入和/或安装到计算机设备1700上。当计算机程序加载到RAM 1703并由计算单元1701执行时，可以执行上文描述的应用于地图更新的图编码模型的训练方法的一个或多个步骤。备选地，在其他实施例中，计算单元1701可以通过其他任何适当的方式(例如，借助于固件)而被配置为执行方法用于图像处理的模型训练。Computing unit 1701 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of the computing unit 1701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, digital signal processing processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 1701 performs various methods and processes described above, such as a processing method for generating positioning information of a high-precision map. For example, in some embodiments, methods for model training for image processing may be implemented as a computer software program that is tangibly embodied in a machine-readable medium, such as storage unit 1708 . In some embodiments, part or all of a computer program may be loaded and/or installed onto computer device 1700 via ROM 1702 and/or communication unit 1709 . When the computer program is loaded into the RAM 1703 and executed by the computing unit 1701, one or more steps of the above-described training method for a graph encoding model applied to map updating may be performed. Alternatively, in other embodiments, the computing unit 1701 may be configured to perform the method for model training for image processing in any other suitable manner (eg, by means of firmware).
本文中以上描述的系统和技术的各种实施方式可以在数字电子电路系统、集成电路系统、场可编程门阵列(FPGA)、专用集成电路(ASIC)、专用标准产品(ASSP)、芯片上系统的系统(SOC)、复杂可编程逻辑设备(CPLD)、计算机硬件、固件、软件、和/或它们的组合中实现。这些各种实施方式可以包括：实施在一个或者多个计算机程序中，该一个或者多个计算机程序可在包括至少一个可编程处理器的可编程系统上执行和/或解释，该可编程处理器可以是专用或者通用可编程处理器，可以从存储系统、至少一个输入装置、和至少一个输出装置接收数据和指令，并且将数据和指令传输至该存储系统、该至少一个输入装置、和该至少一个输出装置。Various implementations of the systems and techniques described above may be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip implemented in a system (SOC), complex programmable logic device (CPLD), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include implementation in one or more computer programs executable and/or interpreted on a programmable system including at least one programmable processor, the programmable processor The processor, which may be a special purpose or general purpose programmable processor, may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device. An output device.
用于实施本公开的方法的程序代码可以采用一个或多个编程语言的任何组合来编写。这些程序代码可以提供给通用计算机、专用计算机或其他可编程数据处理装置的处理器或控制器，使得程序代码当由处理器或控制器执行时使流程图和/或框图中所规定的功能/操作被实施。程序代码可以完全在机器上执行、部分地在机器上执行，作为独立软件包部分地在机器上执行且部分地在远程机器上执行或完全在远程机器或计算机设备上执行。Program code for implementing the methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that the program codes, when executed by the processor or controller, cause the functions specified in the flowcharts and/or block diagrams/ The operation is implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or computer device.
在本公开的上下文中，机器可读介质可以是有形的介质，其可以包含或存储以供指令执行系统、装置或设备使用或与指令执行系统、装置或设备结合地使用的程序。机器可读介质可以是机器可读信号介质或机器可读储存介质。机器可读介质可以包括但不限于电子的、磁性的、光学的、电磁的、红外的、或半导体系统、装置或设备，或者上述内容的任何合适组合。机器可读存储介质的更具体示例会包括基于一个或多个线的电气连接、便携式计算机盘、硬盘、随机存取存储器(RAM)、只读存储器(ROM)、可擦除可编程只读存储器(EPROM或快闪存储器)、光纤、便捷式紧凑盘只读存储器(CD-ROM)、光学储存设备、磁储存设备、或上述内容的任何合适组合。In the context of this disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media may include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media would include one or more wire-based electrical connections, laptop disks, hard drives, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above.
为了提供与用户的交互，可以在计算机上实施此处描述的系统和技术，该计算机具有：用于向用户显示信息的显示装置(例如，CRT(阴极射线管)或者LCD(液晶显示器)监视器)；以及键盘和指向装置(例如，鼠标或者轨迹球)，用户可以通过该键盘和该指向装置来将输入提供给计算机。其它种类的装置还可以用于提供与用户的交互；例如，提供给用户的反馈可以是任何形式的传感反馈(例如，视觉反馈、听觉反馈、或者触觉反馈)；并且可以用任何形式(包括声输入、语音输入或者、触觉输入)来接收来自用户的输入。To provide interaction with a user, the systems and techniques described herein may be implemented on a computer having a display device (eg, a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user ); and a keyboard and pointing device (eg, a mouse or a trackball) through which a user can provide input to the computer. Other kinds of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and may be provided in any form, including Acoustic input, voice input or tactile input) to receive input from the user.
可以将此处描述的系统和技术实施在包括后台部件的计算系统(例如，作为数据计算机设备)、或者包括中间件部件的计算系统(例如，应用计算机设备)、或者包括前端部件的计算系统(例如，具有图形用户界面或者网络浏览器的用户计算机，用户可以通过该图形用户界面或者该网络浏览器来与此处描述的系统和技术的实施方式交互)、或者包括这种后台部件、中间件部件、或者前端部件的任何组合的计算系统中。可以通过任何形式或者介质的数字数据通信(例如，通信网络)来将系统的部件相互连接。通信网络的示例包括：局域网(LAN)、广域网(WAN)和互联网。The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data computing device), or a computing system that includes middleware components (e.g., an application computing device), or a computing system that includes front-end components (e.g., as a data computing device). For example, a user computer having a graphical user interface or a web browser through which the user can interact with implementations of the systems and technologies described herein), or including such backend components, middleware components, or any combination of front-end components in a computing system. The components of the system may be interconnected by any form or medium of digital data communication (eg, a communications network). Examples of communication networks include: local area network (LAN), wide area network (WAN), and the Internet.
计算机系统可以包括客户端和计算机设备。客户端和计算机设备一般远离彼此并且通常通过通信网络进行交互。通过在相应的计算机上运行并且彼此具有客户端-计算机设备关系的计算机程序来产生客户端和计算机设备的关系。计算机设备可以是云计算机设备，又称为云计算计算机设备或云主机，是云计算服务体系中的一项主机产品，以解决了传统物理主机与VPS服务("Virtual Private Server"，或简称"VPS")中，存在的管理难度大，业务扩展性弱的缺陷。计算机设备也可以为分布式系统的计算机设备，或者是结合了区块链的计算机设备。Computer systems may include clients and computer devices. Clients and computer devices are generally remote from each other and often interact through a communications network. The relationship of client and computer device is created by computer programs running on the respective computers and having a client-computer device relationship with each other. Computer equipment can be cloud computer equipment, also known as cloud computing computer equipment or cloud hosts. It is a host product in the cloud computing service system to solve the problem of traditional physical hosts and VPS services ("Virtual Private Server", or "Virtual Private Server" for short) VPS") has the disadvantages of difficult management and weak business scalability. The computer equipment may also be a computer equipment of a distributed system, or a computer equipment combined with a blockchain.
应该理解，可以使用上面所示的各种形式的流程，重新排序、增加或删除步骤。例如，本发公开中记载的各步骤可以并行地执行也可以顺序地执行也可以不同的次序执行，只要能够实现本公开公开的技术方案所期望的结果，本文在此不进行限制。It should be understood that various forms of the process shown above may be used, with steps reordered, added or deleted. For example, each step described in the present disclosure can be executed in parallel, sequentially, or in a different order. As long as the desired results of the technical solution disclosed in the present disclosure can be achieved, there is no limitation here.
上述具体实施方式，并不构成对本公开保护范围的限制。本领域技术人员应该明白的是，根据设计要求和其他因素，可以进行各种修改、组合、子组合和替代。任何在本公开的精神和原则之内所作的修改、等同替换和改进等，均应包含在本公开保护范围之内。The above-mentioned specific embodiments do not constitute a limitation on the scope of the present disclosure. It will be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions are possible depending on design requirements and other factors. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of this disclosure shall be included in the protection scope of this disclosure.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111567400.7ACN114283343B (en) | 2021-12-20 | 2021-12-20 | Map updating method, training method and device based on remote sensing satellite image |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111567400.7ACN114283343B (en) | 2021-12-20 | 2021-12-20 | Map updating method, training method and device based on remote sensing satellite image |
| Publication Number | Publication Date |
|---|---|
| CN114283343A CN114283343A (en) | 2022-04-05 |
| CN114283343Btrue CN114283343B (en) | 2023-09-26 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111567400.7AActiveCN114283343B (en) | 2021-12-20 | 2021-12-20 | Map updating method, training method and device based on remote sensing satellite image |
| Country | Link |
|---|---|
| CN (1) | CN114283343B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114842107A (en)* | 2022-04-06 | 2022-08-02 | 山东新一代信息产业技术研究院有限公司 | High-precision map generation method based on satellite map |
| CN117612026B (en)* | 2023-11-24 | 2024-06-07 | 北京邮电大学 | Path prediction method and system based on remote sensing image road extraction |
| CN117440135B (en)* | 2023-12-20 | 2024-03-08 | 深圳市宇隆移动互联网有限公司 | Image transmission method and system based on Beidou satellite communication |
| CN120597993A (en)* | 2025-08-11 | 2025-09-05 | 上海博礼智能科技有限公司 | Cross-modal large model fine-tuning training system based on prompt word embedding |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109919051A (en)* | 2019-02-22 | 2019-06-21 | 同济大学 | A Convolutional Neural Network Acceleration Method for Video Image Processing |
| CN109991984A (en)* | 2019-04-22 | 2019-07-09 | 上海蔚来汽车有限公司 | For generating the method, apparatus and computer storage medium of fine map |
| CN110376594A (en)* | 2018-08-17 | 2019-10-25 | 北京京东尚科信息技术有限公司 | A kind of method and system of the intelligent navigation based on topological diagram |
| CN111126166A (en)* | 2019-11-30 | 2020-05-08 | 武汉汉达瑞科技有限公司 | Remote sensing image road extraction method and system |
| CN111815776A (en)* | 2020-02-04 | 2020-10-23 | 山东水利技师学院 | Three-dimensional building fine geometric reconstruction method integrating airborne and vehicle-mounted three-dimensional laser point clouds and streetscape images |
| CN112749578A (en)* | 2019-10-29 | 2021-05-04 | 中科星图股份有限公司 | Remote sensing image automatic road extraction method based on deep convolutional neural network |
| CN113008247A (en)* | 2020-03-24 | 2021-06-22 | 青岛慧拓智能机器有限公司 | High-precision map construction method and device for mining area |
| CN113642585A (en)* | 2021-10-14 | 2021-11-12 | 腾讯科技（深圳）有限公司 | Image processing method, apparatus, device, storage medium, and computer program product |
| WO2021238283A1 (en)* | 2020-05-25 | 2021-12-02 | 华为技术有限公司 | Positioning method and positioning apparatus |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110809784B (en)* | 2017-09-27 | 2021-04-20 | 谷歌有限责任公司 | End-to-end network model for high resolution image segmentation |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110376594A (en)* | 2018-08-17 | 2019-10-25 | 北京京东尚科信息技术有限公司 | A kind of method and system of the intelligent navigation based on topological diagram |
| CN109919051A (en)* | 2019-02-22 | 2019-06-21 | 同济大学 | A Convolutional Neural Network Acceleration Method for Video Image Processing |
| CN109991984A (en)* | 2019-04-22 | 2019-07-09 | 上海蔚来汽车有限公司 | For generating the method, apparatus and computer storage medium of fine map |
| CN112749578A (en)* | 2019-10-29 | 2021-05-04 | 中科星图股份有限公司 | Remote sensing image automatic road extraction method based on deep convolutional neural network |
| CN111126166A (en)* | 2019-11-30 | 2020-05-08 | 武汉汉达瑞科技有限公司 | Remote sensing image road extraction method and system |
| CN111815776A (en)* | 2020-02-04 | 2020-10-23 | 山东水利技师学院 | Three-dimensional building fine geometric reconstruction method integrating airborne and vehicle-mounted three-dimensional laser point clouds and streetscape images |
| CN113008247A (en)* | 2020-03-24 | 2021-06-22 | 青岛慧拓智能机器有限公司 | High-precision map construction method and device for mining area |
| WO2021238283A1 (en)* | 2020-05-25 | 2021-12-02 | 华为技术有限公司 | Positioning method and positioning apparatus |
| CN113642585A (en)* | 2021-10-14 | 2021-11-12 | 腾讯科技（深圳）有限公司 | Image processing method, apparatus, device, storage medium, and computer program product |
| Title |
|---|
| ResUNet-a:A deep learning framework for semantic segmentation of remotely sensed data;Foivos I.Diakogiannis等;ISPRS Journal of Photogrammetry and Remote Sensing;全文* |
| 基于U-Net优化的SAR遥感图像语义分割;王鑫 等;计算机科学;全文* |
| 多尺度信息融合的遥感图像语义分割模型;张静 等;计算机辅助设计与图形学学报;全文* |
| Publication number | Publication date |
|---|---|
| CN114283343A (en) | 2022-04-05 |
| Publication | Publication Date | Title |
|---|---|---|
| CN114283343B (en) | Map updating method, training method and device based on remote sensing satellite image | |
| US10853687B2 (en) | Method and apparatus for determining matching relationship between point cloud data | |
| CN113204615B (en) | Entity extraction methods, devices, equipment and storage media | |
| WO2023015941A1 (en) | Text detection model training method and apparatus, text detection method, and device | |
| CN112560862A (en) | Text recognition method, device and electronic device | |
| CN111291681B (en) | Detection methods, devices and equipment for lane line change information | |
| CN115880555A (en) | Target detection method, model training method, device, equipment and medium | |
| CN112509126B (en) | Method, device, equipment and storage medium for detecting three-dimensional object | |
| CN112862006B (en) | Training method, device and electronic equipment for image depth information acquisition model | |
| CN115620321B (en) | Form recognition method and device, electronic device and storage medium | |
| WO2022237821A1 (en) | Method and device for generating traffic sign line map, and storage medium | |
| CN113724388B (en) | High-precision map generation method, device, equipment and storage medium | |
| CN114627257A (en) | Construction method, device, electronic device and storage medium of three-dimensional road network map | |
| CN113762397A (en) | Detection model training, high-precision map update method, equipment, medium and product | |
| CN117745944A (en) | Pre-training model determining method, device, equipment and storage medium | |
| CN115937546A (en) | Image matching, three-dimensional image reconstruction method, device, electronic device and medium | |
| CN114299242A (en) | Method, device and equipment for processing images in high-precision map and storage medium | |
| CN117132964A (en) | Model training method, point cloud coding method, object processing method and device | |
| CN114463409B (en) | Method, device, electronic equipment and medium for determining image depth information | |
| CN115937449A (en) | High-precision map generation method and device, electronic equipment and storage medium | |
| CN114911811A (en) | Map data updating method and device and electronic equipment | |
| CN115115793A (en) | Image processing method, device, device and storage medium | |
| CN116740498B (en) | Model pre-training method, model training method, object processing method and device | |
| CN111400537B (en) | Road element information acquisition method and device and electronic equipment | |
| CN113205151A (en) | Ship target real-time detection method and terminal based on improved SSD model |
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |