技术领域technical field
本发明属于遥感与摄影测量中新型遥感数据应用领域，特别是涉及到无人机系统中搭载的激光扫描仪在获取地形、地貌等信息时的快速提取过程中，海量点云数据的自适应滤波处理以及滤波质量的评定方法。The invention belongs to the application field of new remote sensing data in remote sensing and photogrammetry, and in particular relates to the adaptive filtering of massive point cloud data during the rapid extraction process of the laser scanner carried in the unmanned aerial vehicle system when acquiring terrain, landform and other information Methods for evaluating processing and filtering quality.
背景技术Background technique
自上世纪80年代开始，激光扫描技术(Light Detection And Ranging，LIDAR)，作为一种新型的主动式遥感技术，在三维空间信息获取方面取得了重大突破[1]。随着商用机载LiDAR系统的普及和推广，机载LiDAR已逐渐应用于三维城市建模[2]，数字地面模型的获取[3]，真正射影像的制作[4,5]等多个领域。近年来，LiDAR系统在硬件和系统集成方面的发展已经十分成熟，而点云数据的后处理及数据挖掘则相对滞后，主要体现在如何高效组织与管理海量不规则离散点云，识别与提取高精度的地物类别。而几乎所有点云数据的后处理算法研究以及应用的关键步骤便是滤波处理，因此，点云滤波算法的研究已成为大多数专家学者关注的热点之一。LiDAR点云数据后处理研究的重要任务是从海量激光点集中提取有效信息，其中，相当一部分是关于数据的滤波处理，即提取出真实地形和地物特征，然后为进一步将机载LiDAR点云分成有意义的点类(道路、裸地、草地、建筑物、树木等)提供良好的条件。目前，大多数滤波算法都是根据判断某点云与周围邻近点的高程差异或几何关系作了假设，然后形成了四种滤波方法:基于数学形态学的滤波方法,基于分割的滤波算法,基于渐进加密的滤波方法和基于地表的滤波方法。Since the 1980s, laser scanning technology (Light Detection And Ranging, LIDAR), as a new type of active remote sensing technology, has made a major breakthrough in the acquisition of three-dimensional spatial information[1] . With the popularization and promotion of commercial airborne LiDAR systems, airborne LiDAR has been gradually applied in 3D city modeling[2] , digital ground model acquisition[3] , true orthophoto production[4,5] and many other fields . In recent years, the development of LiDAR system in terms of hardware and system integration has been very mature, but the post-processing and data mining of point cloud data are relatively lagging behind, mainly reflected in how to efficiently organize and manage massive irregular discrete point clouds, identify and extract high Precision feature category. The key step in the research and application of almost all post-processing algorithms of point cloud data is filtering processing. Therefore, the research on point cloud filtering algorithms has become one of the hot spots that most experts and scholars pay attention to. The important task of LiDAR point cloud data post-processing research is to extract effective information from a large number of laser points. Sorting into meaningful point classes (roads, bare land, grass, buildings, trees, etc.) provides good conditions. At present, most filtering algorithms make assumptions based on judging the elevation difference or geometric relationship between a certain point cloud and surrounding adjacent points, and then form four filtering methods: filtering method based on mathematical morphology, filtering algorithm based on segmentation, based on Progressive encryption filtering method and surface-based filtering method.
发明内容Contents of the invention
本发明要解决的技术问题在于提供一种使算法自动化切尽量保留点云原始信息的角度设计滤波算法，以解决现有滤波算法应用时存在的问题。The technical problem to be solved by the present invention is to provide an angle-designed filtering algorithm that automates the algorithm and preserves the original information of the point cloud as much as possible, so as to solve the problems existing in the application of the existing filtering algorithm.
本发明采用以下技术方案：The present invention adopts following technical scheme:
基于自适应坡度的无人机载LiDAR点云滤波方法，包括：UAV-borne LiDAR point cloud filtering method based on adaptive slope, including:
步骤1：通过无人机载LiDAR获取测区内的点云数据，对点云数据进行滤波，剔除高程异常点；Step 1: Obtain the point cloud data in the survey area through the LiDAR carried by the UAV, filter the point cloud data, and remove the abnormal elevation points;
步骤2：使用虚拟规则格网对点云数据建立索引；Step 2: Index the point cloud data using a virtual regular grid;
步骤3：对步骤2中形成的所有格网，利用每个格网中的所有点，使用均方根误差最小原则分别进行平面拟合，同时保留拟合的平面中保留最小均方误差不大于设定阈值的最优平面；Step 3: For all the grids formed in step 2, use all the points in each grid to carry out plane fitting respectively using the principle of minimum root mean square error, while retaining the minimum mean square error in the fitted plane not greater than set the optimal plane for the threshold;
步骤4：预设第一次选择的窗口尺寸，选择预设尺寸的窗口内的高程最低点作为地面点的初始种子点；Step 4: Preset the window size selected for the first time, and select the lowest elevation point in the window of the preset size as the initial seed point of the ground point;
步骤5：计算步骤3中的最优平面和与其最近的地面点之间的距离，以及该地面点到最优平面中心的坡度，若距离小于设定的距离阈值，且坡度小于设定的坡度阈值，则认为该平面为地面平面，将该地面平面延伸至另一格网，同时判断另一格网内的点到该地面平面的距离和平面中心的坡度，若距离小于设定的距离阈值，且坡度小于设定的坡度阈值，则判断为地面点，依次类推到其它格网，获取地面点集；Step 5: Calculate the distance between the optimal plane in step 3 and its nearest ground point, and the slope from the ground point to the center of the optimal plane, if the distance is less than the set distance threshold, and the slope is less than the set slope Threshold, the plane is considered to be the ground plane, and the ground plane is extended to another grid, and at the same time, the distance from the point in the other grid to the ground plane and the slope of the plane center are judged. If the distance is less than the set distance threshold , and the slope is less than the set slope threshold, it is judged as a ground point, and so on to other grids to obtain the ground point set;
步骤6：将初始地面点集中的点作为基点，采用反距离加权插值法生成数字地面模型，该数字地面模型的格网尺寸大于步骤2中建立索引时的格网尺寸；Step 6: Use the points in the initial ground point set as the base points, and use the inverse distance weighted interpolation method to generate a digital ground model. The grid size of the digital ground model is larger than the grid size when the index is established in step 2;
步骤7：计算新生的数字地面模型的均值，并从该均值数字地面模型中提取出局部坡度最大值，作为坡度阈值的更新值，同时，使用新尺寸的窗口内的高程最低点作为地面点的初始种子点，重复迭代步骤4～步骤6，直到达到设定的迭代结束条件，获取最终的地面点集。Step 7: Calculate the mean value of the new digital ground model, and extract the local slope maximum from the mean value digital ground model, as the update value of the slope threshold, and at the same time, use the lowest point of elevation in the window of the new size as the ground point For the initial seed point, iterative steps 4 to 6 are repeated until the set iteration end condition is reached, and the final ground point set is obtained.
所述步骤1中，通过基于局部临近点拟合的噪声检测方法剔除高程异常点。In the step 1, elevation anomalies are eliminated by a noise detection method based on local adjacent point fitting.
所述步骤7中，迭代结束的条件为：迭代次数达到设定的阈值，或者生成的数字地面模型中的格网窗口的尺寸包含测区内最大建筑物的尺寸。In step 7, the condition for the end of the iteration is: the number of iterations reaches the set threshold, or the size of the grid window in the generated digital terrain model includes the size of the largest building in the survey area.
所述步骤7中，新尺寸的窗口大小为上次迭代过程中窗口大小的5倍。In the step 7, the window size of the new size is 5 times of the window size in the previous iteration process.
本发明的有益效果：Beneficial effects of the present invention:
(1)采用虚拟格网索引，实现点云搜索效率的提高，而且最大程度地保留激光点云的原始信息。(1) The virtual grid index is used to improve the search efficiency of the point cloud, and to preserve the original information of the laser point cloud to the greatest extent.
(2)基于自适应坡度的滤波是一个迭代更新参数的过程，实现了在坡度参数更新的同时，数字地形模型也进行了迭代更新，且无需人工设置阈值参数，提高算法运行效率。(2) The filtering based on adaptive slope is a process of iteratively updating parameters, which realizes that while the slope parameters are updated, the digital terrain model is also iteratively updated, and there is no need to manually set the threshold parameters, which improves the efficiency of the algorithm.
(3)针对不同类型的城区点云数据，该滤波算法能够很好的顾及大尺度数据整体的地形起伏和局部细节，为激光点云后续处理以及应用提供保障。(3) For different types of urban point cloud data, the filtering algorithm can well take into account the overall terrain fluctuation and local details of large-scale data, and provide guarantee for the subsequent processing and application of laser point cloud.
附图说明Description of drawings
图1为基于虚拟规则格网的索引建立示意图。Figure 1 is a schematic diagram of index establishment based on a virtual regular grid.
图2为判别地面点与平面之间的坡度和距离的示意图。Fig. 2 is a schematic diagram of judging the slope and distance between a ground point and a plane.
图3为本发明的工作流程图。Fig. 3 is a working flow diagram of the present invention.
图4为测区1的滤波前效果图，其中(a)畸变纠正后的影像数据，(b)滤波前点云灰度示意图。Fig. 4 is the effect diagram before filtering of measurement area 1, in which (a) the image data after distortion correction, (b) the schematic diagram of point cloud grayscale before filtering.
图5为测区2的影像与点云数据，(a)原始影像数据，(b)对应的原始点云灰度示意图。Figure 5 is the image and point cloud data of survey area 2, (a) the original image data, (b) the corresponding original point cloud gray scale diagram.
图6为测区1的滤波前后效果图，(a)滤波后的地面点云，(b)地面点云生成的DEM.Figure 6 is the effect diagram before and after filtering of survey area 1, (a) ground point cloud after filtering, (b) DEM generated by ground point cloud.
图7为测区2两种滤波算法对比效果，(a)本发明滤波后地面点局部细节，(b)渐进TIN滤波后的地面点局部细节，c)数据块3影像数据局部效果图。Fig. 7 is the comparison effect of two filtering algorithms in survey area 2, (a) local details of ground points after filtering by the present invention, (b) local details of ground points after progressive TIN filtering, and c) local effect diagram of image data of data block 3.
具体实施方式detailed description
下面结合附图和具体实施方式对本发明作进一步详细说明。The present invention will be described in further detail below in conjunction with the accompanying drawings and specific embodiments.
本发明通过借鉴基于坡度滤波的基本原理，充分考虑点云数据组织形式以及地表形态，提出了适合城区分类的自适应坡度滤波算法。基于坡度的滤波算法最早由Vosselman提出来，通过比较相邻两点间的高程变化确定最优滤波函数。对于给定的高差阈值Δhmax(d)，两点间距离的递减，直接降低高程值大的激光点属于地面的可能性。分析相邻两点间高程突变的原因可能由于两激光点位分别位于地面和植被、或房屋之上，因此以比较两点间的高差值的大小来判别拒绝或接受所选的点是基于坡度滤波算法的依据。此外，假设某一测区内地形坡度不会大于30％，考虑到地面点观测值的误差情况，设置5％的置信区间，则滤波函数可定义为：The invention proposes an adaptive gradient filtering algorithm suitable for urban classification by referring to the basic principle based on gradient filtering and fully considering point cloud data organization forms and surface forms. The slope-based filtering algorithm was first proposed by Vosselman to determine the optimal filtering function by comparing the elevation changes between two adjacent points. For a given height difference threshold Δhmax (d), the decreasing distance between two points directly reduces the possibility that the laser point with a large elevation value belongs to the ground. The reason for analyzing the sudden change in elevation between two adjacent points may be that the two laser points are respectively located on the ground, vegetation, or houses, so it is based on comparing the height difference between the two points to judge whether to reject or accept the selected point The basis of the slope filtering algorithm. In addition, assuming that the terrain slope in a survey area will not be greater than 30%, considering the error of ground point observations and setting a 5% confidence interval, the filter function can be defined as:
其中N为激光点集，DEM为地面点，d为两点间距离，若在点pi的邻域内没有临近点pj满足式(1)，则pi就被划分为地面点。式(2)中，σ为地面点的标准差。Among them, N is the laser point set, DEM is the ground point, and d is the distance between two points. If there is no adjacent point pj in the neighborhood of point pi that satisfies formula (1), then pi is classified as a ground point. In formula (2), σ is the standard deviation of ground points.
本发明提供一种基于自适应坡度的无人机载LiDAR点云滤波方法，并对该滤波方法的滤波质量进行评定。该方法包括两部分内容，一是激光点云数据的预处理，二是滤波过程中的坡度参数的自动更新。本发明能够针对不同类型的城区点云数据进行滤波，并且可以很好的顾及大尺度数据整体的地形起伏和局部细节，为激光点云后续处理及应用提供保障。The invention provides a method for filtering point clouds of UAV-borne LiDAR based on adaptive slope, and evaluates the filtering quality of the filtering method. The method includes two parts, one is the preprocessing of the laser point cloud data, and the other is the automatic update of the slope parameters in the filtering process. The invention can filter different types of urban point cloud data, and can well take into account the overall terrain fluctuation and local details of large-scale data, and provide guarantee for subsequent processing and application of laser point cloud.
机载LiDAR系统获取三维空间中的点云呈随机分布，分别记录了来自裸露地面和地物表面的反射脉冲信号。尤其是城区点云数据，包含不同类型的地物信息，分别是真实地形点、人工地物(建筑物、桥梁等)、自然地物(草地、灌木、高植被等)。一般情况下，影响城区滤波质量的因素可归纳为两个方面：①城市建筑物侧面点云被错误分类为地面点；②将树木靠近地表的多次回波信号误分为地面点。The airborne LiDAR system acquires point clouds in three-dimensional space that are randomly distributed, and the reflected pulse signals from the bare ground and the surface of the object are recorded respectively. In particular, urban point cloud data contains different types of object information, including real terrain points, artificial objects (buildings, bridges, etc.), and natural objects (grassland, shrubs, tall vegetation, etc.). In general, the factors that affect the filtering quality of urban areas can be summarized into two aspects: ① The side point clouds of urban buildings are misclassified as ground points; ② The multiple echo signals of trees close to the surface are misclassified as ground points.
本发明的方法包括以下步骤：Method of the present invention comprises the following steps:
步骤1：通过无人机载LiDAR获取需要的点云数据；该点云数据为获取的测区内的地形数据。同时，点云数据获取以后，需要对点云数据的噪声进行剔除，即对点云数据进行滤波，剔除高程异常点；本发明中，利用K近邻查询的方式搜索某点周围的激光点，剔除一些高程值异常的激光点。Step 1: Obtain the required point cloud data through the UAV-borne LiDAR; the point cloud data is the terrain data in the acquired survey area. At the same time, after the point cloud data is acquired, the noise of the point cloud data needs to be removed, that is, the point cloud data is filtered to remove elevation abnormal points; Some laser points with unusual elevation values.
步骤2：使用虚拟规则格网对点云数据建立索引。由于通过激光获取的激光点集的各点之间是离散分布，没有一定的拓扑关系，致使对于激光点云数据的处理方式和算法都与传统的基于栅格的影像数据有很大不同。目前使用较为广泛的为基于二维分块的格网索引和基于几何的树形索引。Step 2: Index the point cloud data using a virtual regular grid. Since the points of the laser point set obtained by laser are discretely distributed and have no certain topological relationship, the processing methods and algorithms for laser point cloud data are very different from traditional raster-based image data. Currently, two-dimensional block-based grid index and geometry-based tree index are widely used.
针对具体的数据处理需求采用不同的空间索引，本发明为了后续滤波算法设计，采用虚拟格网的点云组织方式。具体建立过程为：从原始点云入手，通过虚拟规则格网来管理三维点云。结合激光点的平面坐标，与网格建立索引机制，既可以判断某一激光点落在哪个网格，也可以快速查询出一个网格包含哪些脚点。图1则表示了基于虚拟规则格网的索引建立示意图。其中图1(a)、(b)中的圆点代表LiDAR脚点(x，y，z)，长方体块代表虚拟格网。Different spatial indexes are adopted for specific data processing requirements, and the present invention adopts a virtual grid point cloud organization method for subsequent filtering algorithm design. The specific establishment process is as follows: start from the original point cloud, and manage the 3D point cloud through a virtual regular grid. Combining the plane coordinates of the laser point and establishing an index mechanism with the grid, it can not only determine which grid a certain laser point falls on, but also quickly query which foot points a grid contains. Figure 1 shows a schematic diagram of index establishment based on a virtual regular grid. The dots in Figure 1(a) and (b) represent the LiDAR foot points (x, y, z), and the cuboid blocks represent the virtual grid.
由图1(a)到图2(b)实现虚拟格网建立的规则是按照下式进行的，其中n_grid代表格网大小，一般大于或者等于点云平均点间距。该类索引的优点便是不仅实现了搜索效率的提高，而且能够保留激光点云的原始信息。From Figure 1(a) to Figure 2(b), the rules for establishing the virtual grid are carried out according to the following formula, where n_grid represents the grid size, which is generally greater than or equal to the average point spacing of the point cloud. The advantage of this type of index is that it not only improves the search efficiency, but also preserves the original information of the laser point cloud.
X＝INT(x-xmin)/n_gridX＝INT(xxmin )/n_grid
Y＝INT(y-ymin)/n_gridY=INT(yymin )/n_grid
本步中，也可采用其余的索引方法，例如不规则三角网索引，八叉树索引机制等现有的点云索引方法，此时，需要修改后面的步骤。In this step, other indexing methods can also be used, such as the existing point cloud indexing methods such as irregular triangulation indexing and octree indexing mechanism. In this case, the following steps need to be modified.
步骤3：选择预设尺寸的窗口内的高程最低点作为地面点的初始种子点。Step 3: Select the lowest elevation point within the window of preset size as the initial seed point of the ground point.
每次迭代循环的最后，都要改变预设窗口大小w_size，即如果选择新的迭代窗口为5倍的上次计算时采用的窗口，则新的迭代窗口＝5*上次迭代窗口w_size。而窗口的移动是根据建立的规则格网进行的，即从数据的整体格网的左上角(x0，y0)，依w_size*w_size形成的窗口，按照行—列的方式(从左至右，从上至下)逐一激光点的进行遍历整个激光点云数据。At the end of each iterative cycle, the preset window size w_size must be changed, that is, if the new iterative window is selected to be 5 times the window used in the last calculation, then the new iterative window=5*w_size of the last iterative window. The movement of the window is carried out according to the established regular grid, that is, from the upper left corner (x0 , y0 ) of the overall grid of the data, the window formed according to w_size*w_size, according to the row-column method (from left to Right, from top to bottom) traverse the entire laser point cloud data point by point.
步骤4：对步骤2中形成的所有格网，利用每个格网中的所有点，使用均方根误差(RMSE)最小原则分别进行平面拟合，同时剔除最小均方根误差大于设定阈值Tp1的平面，保留拟合的平面中保留最小均方误差小于设定阈值的最优平面。其中，拟合平面方程如下式。Step 4: For all grids formed in step 2, use all points in each grid to perform plane fitting using the minimum root mean square error (RMSE) principle, and eliminate the minimum root mean square error greater than the set threshold For the plane of Tp1, keep the optimal plane with the minimum mean square error smaller than the set threshold in the fitted plane. Among them, the fitting plane equation is as follows.
ax+by+cz+d＝0ax+by+cz+d=0
步骤5：计算步骤4中的最优平面和与其最近的地面点之间的距离，以及该地面点到平面中心的坡度，若距离小于设定的距离阈值，且坡度小于设定的坡度阈值，则认为该平面为地面平面，将该地面平面延伸至另一格网，同时判断另一格网内的点到该地面平面的距离和平面中心的坡度，若距离小于设定的距离阈值，且坡度小于设定的坡度阈值，则判断为地面点，依次类推到其它格网，获取地面点集；Step 5: Calculate the distance between the optimal plane in step 4 and its nearest ground point, and the slope from the ground point to the center of the plane. If the distance is less than the set distance threshold and the slope is less than the set slope threshold, Then the plane is considered to be the ground plane, and the ground plane is extended to another grid, and at the same time, the distance from the point in the other grid to the ground plane and the slope of the plane center are judged. If the distance is less than the set distance threshold, and If the slope is less than the set slope threshold, it is judged as a ground point, and so on to other grids to obtain the ground point set;
如图2图3所示，对步骤(4)中保留的平面，搜索与其最近的地面点，计算地面点到平面点的距离，以及地面点到平面中心的坡度。若距离小于阈值Td1，坡度小于阈值Ts1，则认为该平面作为地面平面，图2中，Point1和point2分别代表在判定条件内的平面点，通过与种子地面点的比较(距离阈值和坡度阈值)来判断是否能成为初始地面点集内的元素。As shown in Figure 2 and Figure 3, for the plane retained in step (4), search for its nearest ground point, calculate the distance from the ground point to the plane point, and the slope from the ground point to the center of the plane. If the distance is less than the threshold Td1 and the slope is less than the threshold Ts1, the plane is considered as the ground plane. In Figure 2, Point1 and point2 respectively represent the plane points within the judgment conditions, and are compared with the seed ground points (distance threshold and slope threshold) To judge whether it can become an element in the initial ground point set.
dist＝distance是平面点point1或者point2到种子地面点的距离。dist=distance is the distance from the plane point point1 or point2 to the seed ground point.
对位于地面上的平面，计算该平面格网窗口内的点到该平面的距离，若小于距离阈值Td2，则认为该点为平面点；并对当前格网进行区域增长，判断其邻域格网内的点到该平面的距离和平面中心的坡度，若距离小于阈值Td2，坡度小于阈值Ts2，则判为地面点，其标准依然采用图2进行判断，得到初始地面点集。For a plane located on the ground, calculate the distance from the point in the grid window of the plane to the plane, if it is less than the distance threshold Td2, consider the point as a plane point; and perform region growth on the current grid to judge its neighbor grid The distance from the point in the network to the plane and the slope of the plane center, if the distance is less than the threshold Td2, and the slope is less than the threshold Ts2, it is judged as a ground point, and the standard is still judged by Figure 2 to obtain the initial ground point set.
步骤6：将初始地面点集中的点作为基点，采用反距离加权插值法生成数字地面模型，该数字地面模型的格网尺寸大于步骤2中建立索引时的格网尺寸，例如可设置为原来的格网尺寸的五倍。Step 6: Use the points in the initial ground point set as the base points, and use the inverse distance weighted interpolation method to generate a digital ground model. The grid size of the digital ground model is larger than the grid size when the index is established in step 2. For example, it can be set to the original Five times the grid size.
步骤7：计算新生的数字地面模型的均值，并从该均值数字地面模型中提取出局部坡度最大值，作为坡度阈值的更新值，重复步骤3～步骤6，直到生成的数字地面模型中的格网尺寸包含测区内最大建筑物的尺寸，获取最终的比较可靠的地面点集。Step 7: Calculate the mean value of the newly born digital terrain model, and extract the local maximum slope value from the mean value digital terrain model as the update value of the slope threshold, repeat steps 3 to 6 until the grid in the generated digital terrain model The mesh size includes the size of the largest building in the survey area to obtain the final and reliable ground point set.
本发明采用基于地表坡度的滤波思想，从改进算法自动化和尽量保留点云原始信息的角度设计滤波算法，使用时，①鉴于本文实验是大多以城区为基准，其地形相对平坦，为了提高处理效率，采用小尺度的规则格网压缩和组织点云数据。②应对大尺度数据整体的地形起伏同时顾及局部真实的地形细节，采用迭代更新参数的方法，在坡度参数更新的同时，数字地形模型也进行了迭代更新。The present invention adopts the filtering idea based on the slope of the ground surface, and designs the filtering algorithm from the perspective of improving the automation of the algorithm and retaining the original information of the point cloud as much as possible. , using a small-scale regular grid to compress and organize point cloud data. ② In response to the overall terrain fluctuation of large-scale data while taking into account the local real terrain details, the method of iteratively updating parameters is adopted. When the slope parameters are updated, the digital terrain model is also iteratively updated.
本发明还提供对最后生成的滤波结果(地面点集和非地面点集)进行定性和定量评定的方法。The present invention also provides a qualitative and quantitative evaluation method for the finally generated filtering results (ground point set and non-ground point set).
即本发明基于自适应坡度的无人机载LiDAR点云滤波方法系统中，至少应包括以下模块：That is, in the UAV-borne LiDAR point cloud filtering method system based on adaptive slope of the present invention, at least the following modules should be included:
点云数据预处理模块：用于海量无人机LiDAR点云的噪声剔除和建立点云索引机制，仅实现了搜索效率的提高，而且能够保留激光点云的原始信息。Point cloud data preprocessing module: It is used for noise removal and point cloud indexing mechanism of massive UAV LiDAR point cloud, which not only improves the search efficiency, but also retains the original information of laser point cloud.
自适应坡度滤波模块：考虑到应对大尺度数据整体的地形起伏同时顾及局部真实的地形细节，本发明提出的基于自适应坡度的滤波算法，不仅能够自动更新坡度阈值，而且可以同时更新数字地面模型(DTM)，此外，在应对大尺度数据整体的地形起伏和局部真实的地形细节方面具有较好的优势。Adaptive slope filtering module: Considering the overall terrain fluctuation of large-scale data while taking into account the local real terrain details, the filtering algorithm based on adaptive slope proposed by the present invention can not only automatically update the slope threshold, but also update the digital ground model at the same time (DTM), in addition, has better advantages in dealing with the overall terrain fluctuations and local real terrain details of large-scale data.
无人机载点云滤波性能评定模块：为了验证本发明在实现基于无人机载海量点云数据的滤波处理方案的可适用性，分别从定性(DEM点集平滑度)和定量(I类误差、II类误差、总体误差和Kappa系数)两个方面进行性能评定，并输出最终评定结果。UAV-borne point cloud filtering performance evaluation module: in order to verify the applicability of the present invention in realizing the filtering processing scheme based on unmanned aerial vehicle-borne massive point cloud data, from qualitative (DEM point set smoothness) and quantitative (Class I) respectively Error, type II error, overall error and Kappa coefficient) for performance evaluation, and output the final evaluation results.
与传统无人机载点云滤波算法相比较，本发明的优势体现在：Compared with the traditional UAV-borne point cloud filtering algorithm, the advantages of the present invention are reflected in:
(1)采用虚拟格网索引，实现点云搜索效率的提高，而且最大程度地保留激光点云的原始信息。(1) The virtual grid index is used to improve the search efficiency of the point cloud, and to preserve the original information of the laser point cloud to the greatest extent.
(2)基于自适应坡度的滤波是一个迭代更新参数的过程，实现了在坡度参数更新的同时，数字地形模型也进行了迭代更新，且无需人工设置阈值参数，提高算法运行效率。(2) The filtering based on adaptive slope is a process of iteratively updating parameters, which realizes that while the slope parameters are updated, the digital terrain model is also iteratively updated, and there is no need to manually set the threshold parameters, which improves the efficiency of the algorithm.
(3)针对不同类型的城区点云数据，该滤波算法能够很好的顾及大尺度数据整体的地形起伏和局部细节，为激光点云后续处理以及应用提供保障。(3) For different types of urban point cloud data, the filtering algorithm can well take into account the overall terrain fluctuation and local details of large-scale data, and provide guarantee for the subsequent processing and application of laser point cloud.
算法验证与分析Algorithm verification and analysis
测区1数据：该数据来自我国的沙市地区，该测区共包含1,954,659个激光点，X轴向的跨度是904.55米，Y轴向的跨度是789.47米，由于原始数据存在噪声，所以最小高程值是-160.60米，最大高程值是192.33，经过粗差剔除后的最大高程差是38.48米。图4(a)表示的是经过纠正的该测区影像数据，图3(b)是对应的点云数据滤波前构成的数字表面模型(DSM).Survey area 1 data: The data comes from the Shashi area of my country. This survey area contains 1,954,659 laser points in total. The span of the X-axis is 904.55 meters, and the span of the Y-axis is 789.47 meters. Due to the noise in the original data, the minimum elevation The value is -160.60 meters, the maximum elevation value is 192.33, and the maximum elevation difference after gross error elimination is 38.48 meters. Figure 4(a) shows the corrected image data of the survey area, and Figure 3(b) shows the digital surface model (DSM) formed before the corresponding point cloud data is filtered.
测区2数据：该实验数据块选取的是我国湖北老河口地区的一部分，老河口测区的点云数据是采用LSM-Q160激光仪于2013年4月采集的，整个测区共有7条点云条带，共20,141,032个激光点，以及利用DMC-II-140航摄仪获取的12条航带456张影像数据。本测试数据—数据块3的点云个数是443,415，X轴向跨度是545.35米，Y轴向的跨度是505.71米，最小高程值是63.73米，高程最大值是125.64米，图5(a)是数据块3的影像数据，图5(b)则是对应的原始点云DSM。Survey area 2 data: The experimental data block is selected from a part of the Laohekou area in Hubei, my country. The point cloud data of the Laohekou survey area was collected in April 2013 with the LSM-Q160 laser instrument. There are 7 points in the entire survey area Cloud strips, a total of 20,141,032 laser points, and 456 image data of 12 strips acquired by the DMC-II-140 aerial camera. The test data—data block 3 has 443,415 point clouds, the X-axis span is 545.35 meters, the Y-axis span is 505.71 meters, the minimum elevation value is 63.73 meters, and the maximum elevation value is 125.64 meters. Figure 5(a ) is the image data of data block 3, and Fig. 5(b) is the corresponding original point cloud DSM.
算法评价—定性：Algorithm Evaluation - Qualitative:
测区1：如图6所示，从滤波后生成DEM可知，建筑物和树木均被有效地滤除，并且整体效果较为平滑，没有异常的凸起或者凹陷的现象(除了沿河流的两侧的变化)，与实际的地形特征比较吻合。Survey area 1: As shown in Figure 6, from the DEM generated after filtering, it can be seen that buildings and trees are effectively filtered out, and the overall effect is relatively smooth, with no abnormal protrusions or depressions (except for two side change), which is more consistent with the actual topographical features.
算法评价—定量：Algorithm Evaluation - Quantitative:
①混淆矩阵①Confusion matrix
总分类精度(Overall Accuracy):Overall Accuracy:
变化一致性(Change agreement):Change agreement:
Kappa系数：Kappa coefficient:
其中，n表示原始点云总数，g是滤波后的地面点，o是滤波后的地物点，T_g表示实际地面点(人工或经验得出的样本数据)，T_o则是实际地物点个数。而a表示地面点中被正确判断为地面点的数量，b是地面点中被误判为地物点的数量；c则表示地物点中被误判为地面点的数量，d为地物点中正确判断为地物点的个数。Among them, n represents the total number of original point clouds, g is the ground point after filtering, o is the ground object point after filtering, T_g represents the actual ground point (sample data obtained manually or empirically), T_o is the actual ground point number. And a represents the number of ground points that are correctly judged as ground points, b is the number of ground points that are misjudged as ground points; c represents the number of ground points that are misjudged as ground points, and d is the number of ground points The number of points that are correctly judged as feature points.
对于测区2除了采用本发明滤波算法外，还采用传统的渐进TIN滤波算法，图7(a)和(b)表示的两种方法滤波后地面点的细节图，为了更好地进行定性对比，图7(c)则给出了该测区某一部分局部方法的影像数据。表1则是给出了两种方法滤波的混淆矩阵以及相关精度指标，其中样本点中地面点个数是82750，地物点个数是360665。For survey area 2, in addition to the filtering algorithm of the present invention, the traditional progressive TIN filtering algorithm is also used. Figure 7 (a) and (b) show the detailed maps of ground points filtered by the two methods, in order to better perform qualitative comparison , and Fig. 7(c) shows the image data of a certain part of the survey area by the local method. Table 1 shows the confusion matrix filtered by the two methods and related accuracy indicators. The number of ground points in the sample points is 82750, and the number of ground object points is 360665.
表1 测区2两种滤波算法的误差统计Table 1 Error statistics of the two filtering algorithms in survey area 2
由表1可知，两种方法的总体精度都能达到75％以上，即可视为都能得到不错的滤波结果。然而本文的滤波算法是一个迭代更新的过程，不仅会在迭代过程中更新坡度值，还可以更新DTM，能够很好的顾及大尺度数据整体的地形起伏和局部细节，这主要体现在II类误差较之渐近TIN要大幅降低，减少了非地面点误分为地面点的几率，得到比较好的滤波效果，为后续点云数据处理及应用(点云分类、地物识别)提供条件。It can be seen from Table 1 that the overall accuracy of the two methods can reach more than 75%, which can be regarded as obtaining good filtering results. However, the filtering algorithm in this paper is an iterative update process. It not only updates the slope value during the iterative process, but also updates the DTM. It can take into account the overall terrain fluctuation and local details of the large-scale data, which is mainly reflected in the type II error. Compared with asymptotic TIN, it is greatly reduced, which reduces the probability of non-ground points being misclassified as ground points, and obtains a better filtering effect, which provides conditions for subsequent point cloud data processing and applications (point cloud classification, ground object recognition).
以上所述的仅是本发明的优选实施方式，应当指出，对于本领域的技术人员来说，在不脱离本发明整体构思前提下，还可以作出若干改变和改进，这些也应该视为本发明的保护范围。The above are only preferred embodiments of the present invention. It should be pointed out that for those skilled in the art, some changes and improvements can be made without departing from the overall concept of the present invention, and these should also be regarded as the present invention. scope of protection.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610981407.6ACN106529469B (en) | 2016-11-08 | 2016-11-08 | Unmanned aerial vehicle-mounted LiDAR point cloud filtering method based on self-adaptive gradient |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610981407.6ACN106529469B (en) | 2016-11-08 | 2016-11-08 | Unmanned aerial vehicle-mounted LiDAR point cloud filtering method based on self-adaptive gradient |
| Publication Number | Publication Date |
|---|---|
| CN106529469Atrue CN106529469A (en) | 2017-03-22 |
| CN106529469B CN106529469B (en) | 2020-01-03 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610981407.6AActiveCN106529469B (en) | 2016-11-08 | 2016-11-08 | Unmanned aerial vehicle-mounted LiDAR point cloud filtering method based on self-adaptive gradient |
| Country | Link |
|---|---|
| CN (1) | CN106529469B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107506724A (en)* | 2017-08-21 | 2017-12-22 | 中清能绿洲科技股份有限公司 | Inclination angle detection method and detection terminal |
| CN107818550A (en)* | 2017-10-27 | 2018-03-20 | 广东电网有限责任公司机巡作业中心 | A kind of point cloud top portion noise elimination method based on LiDAR |
| CN108919295A (en)* | 2018-05-15 | 2018-11-30 | 国网通用航空有限公司 | Airborne LiDAR point cloud road information extracting method and device |
| CN109283936A (en)* | 2018-08-15 | 2019-01-29 | 广州极飞科技有限公司 | Mobile device control method, device and terminal |
| CN109345638A (en)* | 2018-09-21 | 2019-02-15 | 东华理工大学 | A point cloud filtering method based on Snake model multi-element fusion |
| CN109785261A (en)* | 2019-01-11 | 2019-05-21 | 辽宁工程技术大学 | A kind of airborne LIDAR three-dimensional filtering method based on gray scale volume element model |
| CN110109142A (en)* | 2019-04-04 | 2019-08-09 | 深圳市速腾聚创科技有限公司 | Point cloud filtering method, device, computer equipment and storage medium |
| CN110335352A (en)* | 2019-07-04 | 2019-10-15 | 山东科技大学 | A dual-element multi-resolution hierarchical filtering method for airborne lidar point clouds |
| CN110443786A (en)* | 2019-07-25 | 2019-11-12 | 深圳一清创新科技有限公司 | Laser radar point cloud filtering method, device, computer equipment and storage medium |
| CN110660027A (en)* | 2019-08-28 | 2020-01-07 | 青岛秀山移动测量有限公司 | Laser point cloud continuous profile ground filtering method for complex terrain |
| CN110703227A (en)* | 2019-09-20 | 2020-01-17 | 中国铁路设计集团有限公司 | Ground point cloud filtering method based on Terrasolide parameter threshold automatic selection |
| CN111080536A (en)* | 2019-11-13 | 2020-04-28 | 武汉华中天经通视科技有限公司 | Self-adaptive filtering method for airborne laser radar point cloud |
| CN111337932A (en)* | 2020-03-23 | 2020-06-26 | 广东电网有限责任公司广州供电局 | Power grid infrastructure construction verification method based on airborne lidar system |
| CN111445472A (en)* | 2020-03-26 | 2020-07-24 | 达闼科技成都有限公司 | Laser point cloud ground segmentation method and device, computing equipment and storage medium |
| CN112215823A (en)* | 2020-10-16 | 2021-01-12 | 中铁上海设计院集团有限公司 | Point cloud-based police buoy plane position measuring method |
| CN113421209A (en)* | 2021-06-21 | 2021-09-21 | 安谋科技（中国）有限公司 | Image processing method, system on chip, electronic device, and medium |
| CN113436336A (en)* | 2021-06-22 | 2021-09-24 | 京东鲲鹏(江苏)科技有限公司 | Ground point cloud segmentation method and device and automatic driving vehicle |
| CN114063616A (en)* | 2021-11-11 | 2022-02-18 | 深圳市城市公共安全技术研究院有限公司 | Method and device for planning forest area path based on three-dimensional laser scanning detection |
| CN114485560A (en)* | 2021-12-22 | 2022-05-13 | 绍兴市特种设备检测院 | Road slope rapid detection method for off-highway tourist and sightseeing vehicle |
| CN114612806A (en)* | 2022-02-07 | 2022-06-10 | 安徽理工大学 | A method for improving the accuracy of consumer-grade UAV DEM products |
| CN114942726A (en)* | 2022-05-26 | 2022-08-26 | 中国航空工业集团公司沈阳飞机设计研究所 | Data classification unloading method and system |
| CN115792900A (en)* | 2022-10-19 | 2023-03-14 | 国网电力空间技术有限公司 | Deformation detection method and system for geological sensitive area of power transmission channel |
| CN116229006A (en)* | 2023-05-08 | 2023-06-06 | 四川华恒升科技发展有限公司 | Method for eliminating land arrangement aerial survey error elevation points |
| CN116486268A (en)* | 2023-05-05 | 2023-07-25 | 中交信息技术国家工程实验室有限公司 | A laser point cloud ground filtering method and device |
| CN117132486A (en)* | 2023-07-26 | 2023-11-28 | 安徽农业大学 | A multi-constrained connected graph segmentation point cloud filtering method suitable for complex hilly areas |
| CN118247649A (en)* | 2024-01-16 | 2024-06-25 | 广东国地规划科技股份有限公司 | Filtering method and device for terrain point cloud data |
| CN119942133A (en)* | 2025-01-02 | 2025-05-06 | 中国自然资源航空物探遥感中心 | Airborne LiDAR point cloud filtering method, device and storage medium in complex scenes |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102419794A (en)* | 2011-10-31 | 2012-04-18 | 武汉大学 | Rapid filtering method for airborne laser point cloud data |
| CN102663237A (en)* | 2012-03-21 | 2012-09-12 | 武汉大学 | Point cloud data automatic filtering method based on grid segmentation and moving least square |
| CN102930509A (en)* | 2012-09-18 | 2013-02-13 | 广州建通测绘技术开发有限公司 | Intelligent filtering method for airborne laser point cloud data |
| CN103679655A (en)* | 2013-12-02 | 2014-03-26 | 河海大学 | LiDAR point cloud filter method based on gradient and area growth |
| CN103745436A (en)* | 2013-12-23 | 2014-04-23 | 西安电子科技大学 | LiDar point cloud data morphological filtering method based on area prediction |
| US8831354B1 (en)* | 2014-01-08 | 2014-09-09 | Faroudja Enterprises, Inc. | System and method for edge-adaptive and recursive non-linear filtering of ringing effect |
| CN104123730A (en)* | 2014-07-31 | 2014-10-29 | 武汉大学 | Method and system for remote-sensing image and laser point cloud registration based on road features |
| CN104849722A (en)* | 2015-05-08 | 2015-08-19 | 南京林业大学 | Light detection and ranging (LiDAR) waveform comprehensive feature-based individual tree identification method |
| CN105118090A (en)* | 2015-05-19 | 2015-12-02 | 西南交通大学 | Adaptive point-cloud filtering method for complex terrain structure |
| CN105844629A (en)* | 2016-03-21 | 2016-08-10 | 河南理工大学 | Automatic segmentation method for point cloud of facade of large scene city building |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102419794A (en)* | 2011-10-31 | 2012-04-18 | 武汉大学 | Rapid filtering method for airborne laser point cloud data |
| CN102663237A (en)* | 2012-03-21 | 2012-09-12 | 武汉大学 | Point cloud data automatic filtering method based on grid segmentation and moving least square |
| CN102930509A (en)* | 2012-09-18 | 2013-02-13 | 广州建通测绘技术开发有限公司 | Intelligent filtering method for airborne laser point cloud data |
| CN103679655A (en)* | 2013-12-02 | 2014-03-26 | 河海大学 | LiDAR point cloud filter method based on gradient and area growth |
| CN103745436A (en)* | 2013-12-23 | 2014-04-23 | 西安电子科技大学 | LiDar point cloud data morphological filtering method based on area prediction |
| US8831354B1 (en)* | 2014-01-08 | 2014-09-09 | Faroudja Enterprises, Inc. | System and method for edge-adaptive and recursive non-linear filtering of ringing effect |
| CN104123730A (en)* | 2014-07-31 | 2014-10-29 | 武汉大学 | Method and system for remote-sensing image and laser point cloud registration based on road features |
| CN104849722A (en)* | 2015-05-08 | 2015-08-19 | 南京林业大学 | Light detection and ranging (LiDAR) waveform comprehensive feature-based individual tree identification method |
| CN105118090A (en)* | 2015-05-19 | 2015-12-02 | 西南交通大学 | Adaptive point-cloud filtering method for complex terrain structure |
| CN105844629A (en)* | 2016-03-21 | 2016-08-10 | 河南理工大学 | Automatic segmentation method for point cloud of facade of large scene city building |
| Title |
|---|
| JUNICHI SUSAKI: "Adaptive Slope Filtering of Airborne LiDAR Data in Urban Areas for Digital Terrain Model (DTM) Generation", 《REMOTE SENSING》* |
| PEREZ-GARCIA, J. L. 等: "PROGRESSIVE DENSIFICATION AND REGION GROWING METHODS FOR LIDAR DATA CLASSIFICATION", 《INTERNATIONAL ARCHIVES OF THE PHOTOGRAMMETRY REMOTE SENSING AND SPATIAL INFORMATION SCIENCES》* |
| 冯义从: "车载LIDAR点云的建筑物立面信息快速自动提取", 《万方数据知识服务平台》* |
| 吴侃 等著: "《矿山开采沉陷监测及预测新技术》", 31 December 2012* |
| 惠振阳 等: "基于LIDAR数字高程模型构建的数学形态学滤波方法综述", 《激光与光电子学进展》* |
| 罗伊萍 等: "基于自适应和多尺度数学形态学的点云数据滤波方法", 《测绘科学技术学报》* |
| 魏祖帅: "基于坡度信息和平面拟合的机载LiDAR点云滤波方法", 《河南城建学院学报》* |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107506724A (en)* | 2017-08-21 | 2017-12-22 | 中清能绿洲科技股份有限公司 | Inclination angle detection method and detection terminal |
| CN107818550A (en)* | 2017-10-27 | 2018-03-20 | 广东电网有限责任公司机巡作业中心 | A kind of point cloud top portion noise elimination method based on LiDAR |
| CN108919295A (en)* | 2018-05-15 | 2018-11-30 | 国网通用航空有限公司 | Airborne LiDAR point cloud road information extracting method and device |
| CN109283936A (en)* | 2018-08-15 | 2019-01-29 | 广州极飞科技有限公司 | Mobile device control method, device and terminal |
| CN109345638B (en)* | 2018-09-21 | 2023-08-25 | 东华理工大学 | A point cloud filtering method based on multi-element fusion of Snake model |
| CN109345638A (en)* | 2018-09-21 | 2019-02-15 | 东华理工大学 | A point cloud filtering method based on Snake model multi-element fusion |
| CN109785261A (en)* | 2019-01-11 | 2019-05-21 | 辽宁工程技术大学 | A kind of airborne LIDAR three-dimensional filtering method based on gray scale volume element model |
| CN109785261B (en)* | 2019-01-11 | 2023-03-17 | 辽宁工程技术大学 | Airborne LIDAR three-dimensional filtering method based on gray voxel model |
| CN110109142B (en)* | 2019-04-04 | 2021-04-02 | 深圳市速腾聚创科技有限公司 | Point cloud filtering method and device, computer equipment and storage medium |
| CN110109142A (en)* | 2019-04-04 | 2019-08-09 | 深圳市速腾聚创科技有限公司 | Point cloud filtering method, device, computer equipment and storage medium |
| CN110335352A (en)* | 2019-07-04 | 2019-10-15 | 山东科技大学 | A dual-element multi-resolution hierarchical filtering method for airborne lidar point clouds |
| CN110335352B (en)* | 2019-07-04 | 2022-11-29 | 山东科技大学 | Double-element multi-resolution hierarchical filtering method for airborne laser radar point cloud |
| CN110443786A (en)* | 2019-07-25 | 2019-11-12 | 深圳一清创新科技有限公司 | Laser radar point cloud filtering method, device, computer equipment and storage medium |
| CN110443786B (en)* | 2019-07-25 | 2021-12-07 | 深圳一清创新科技有限公司 | Laser radar point cloud filtering method and device, computer equipment and storage medium |
| CN110660027A (en)* | 2019-08-28 | 2020-01-07 | 青岛秀山移动测量有限公司 | Laser point cloud continuous profile ground filtering method for complex terrain |
| CN110660027B (en)* | 2019-08-28 | 2023-03-31 | 青岛秀山移动测量有限公司 | Laser point cloud continuous profile ground filtering method for complex terrain |
| CN110703227A (en)* | 2019-09-20 | 2020-01-17 | 中国铁路设计集团有限公司 | Ground point cloud filtering method based on Terrasolide parameter threshold automatic selection |
| CN110703227B (en)* | 2019-09-20 | 2023-01-24 | 中国铁路设计集团有限公司 | Ground point cloud filtering method based on Terrasolide parameter threshold automatic selection |
| CN111080536A (en)* | 2019-11-13 | 2020-04-28 | 武汉华中天经通视科技有限公司 | Self-adaptive filtering method for airborne laser radar point cloud |
| CN111337932A (en)* | 2020-03-23 | 2020-06-26 | 广东电网有限责任公司广州供电局 | Power grid infrastructure construction verification method based on airborne lidar system |
| CN111445472B (en)* | 2020-03-26 | 2023-05-12 | 达闼机器人股份有限公司 | Laser point cloud ground segmentation method, device, computing equipment and storage medium |
| CN111445472A (en)* | 2020-03-26 | 2020-07-24 | 达闼科技成都有限公司 | Laser point cloud ground segmentation method and device, computing equipment and storage medium |
| CN112215823A (en)* | 2020-10-16 | 2021-01-12 | 中铁上海设计院集团有限公司 | Point cloud-based police buoy plane position measuring method |
| CN113421209B (en)* | 2021-06-21 | 2022-12-30 | 安谋科技（中国）有限公司 | Image processing method, system on chip, electronic device, and medium |
| CN113421209A (en)* | 2021-06-21 | 2021-09-21 | 安谋科技（中国）有限公司 | Image processing method, system on chip, electronic device, and medium |
| CN113436336A (en)* | 2021-06-22 | 2021-09-24 | 京东鲲鹏(江苏)科技有限公司 | Ground point cloud segmentation method and device and automatic driving vehicle |
| CN113436336B (en)* | 2021-06-22 | 2024-01-12 | 京东鲲鹏(江苏)科技有限公司 | Ground point cloud segmentation method and device and automatic driving vehicle |
| WO2022267711A1 (en)* | 2021-06-22 | 2022-12-29 | 京东鲲鹏(江苏)科技有限公司 | Ground point cloud segmentation method and apparatus, and autonomous vehicle |
| CN114063616A (en)* | 2021-11-11 | 2022-02-18 | 深圳市城市公共安全技术研究院有限公司 | Method and device for planning forest area path based on three-dimensional laser scanning detection |
| CN114063616B (en)* | 2021-11-11 | 2024-03-01 | 深圳市城市公共安全技术研究院有限公司 | Method and device for planning forest path based on three-dimensional laser scanning detection |
| CN114485560B (en)* | 2021-12-22 | 2024-03-15 | 绍兴市特种设备检测院 | Road gradient rapid detection method for off-highway sightseeing vehicle |
| CN114485560A (en)* | 2021-12-22 | 2022-05-13 | 绍兴市特种设备检测院 | Road slope rapid detection method for off-highway tourist and sightseeing vehicle |
| CN114612806A (en)* | 2022-02-07 | 2022-06-10 | 安徽理工大学 | A method for improving the accuracy of consumer-grade UAV DEM products |
| CN114942726A (en)* | 2022-05-26 | 2022-08-26 | 中国航空工业集团公司沈阳飞机设计研究所 | Data classification unloading method and system |
| CN115792900A (en)* | 2022-10-19 | 2023-03-14 | 国网电力空间技术有限公司 | Deformation detection method and system for geological sensitive area of power transmission channel |
| CN116486268A (en)* | 2023-05-05 | 2023-07-25 | 中交信息技术国家工程实验室有限公司 | A laser point cloud ground filtering method and device |
| CN116486268B (en)* | 2023-05-05 | 2025-08-29 | 中交信息技术国家工程实验室有限公司 | A laser point cloud ground filtering method and device |
| CN116229006B (en)* | 2023-05-08 | 2023-08-08 | 四川华恒升科技发展有限公司 | Method for eliminating land arrangement aerial survey error elevation points |
| CN116229006A (en)* | 2023-05-08 | 2023-06-06 | 四川华恒升科技发展有限公司 | Method for eliminating land arrangement aerial survey error elevation points |
| CN117132486A (en)* | 2023-07-26 | 2023-11-28 | 安徽农业大学 | A multi-constrained connected graph segmentation point cloud filtering method suitable for complex hilly areas |
| CN118247649A (en)* | 2024-01-16 | 2024-06-25 | 广东国地规划科技股份有限公司 | Filtering method and device for terrain point cloud data |
| CN119942133A (en)* | 2025-01-02 | 2025-05-06 | 中国自然资源航空物探遥感中心 | Airborne LiDAR point cloud filtering method, device and storage medium in complex scenes |
| Publication number | Publication date |
|---|---|
| CN106529469B (en) | 2020-01-03 |
| Publication | Publication Date | Title |
|---|---|---|
| CN106529469B (en) | Unmanned aerial vehicle-mounted LiDAR point cloud filtering method based on self-adaptive gradient | |
| CN116310192B (en) | Urban building three-dimensional model monomer reconstruction method based on point cloud | |
| CN108363983B (en) | An urban vegetation classification method based on UAV images and reconstructed point clouds | |
| CN105513127B (en) | Method and system for regularized three-dimensional modeling of rods based on density peak clustering | |
| CN106815847B (en) | Trees dividing method and single tree extracting method based on laser radar point cloud | |
| CN114332366B (en) | Digital urban single house point cloud elevation 3D feature extraction method | |
| CN105488770B (en) | A kind of airborne laser radar point cloud filtering method of object-oriented | |
| CN111340723B (en) | A terrain-adaptive thin-plate spline interpolation filtering method for airborne LiDAR point cloud regularization | |
| CN111598780B (en) | Terrain adaptive interpolation filtering method suitable for airborne LiDAR point cloud | |
| CN108171720A (en) | A kind of oblique photograph model object frontier probe method based on geometrical statistic information | |
| CN106970375A (en) | A kind of method that building information is automatically extracted in airborne laser radar point cloud | |
| CN109816664A (en) | A kind of three-dimensional point cloud dividing method and device | |
| CN102915560B (en) | Threshold-irrelative point cloud filtering method and device for airborne laser radar | |
| CN108765568A (en) | A kind of multi-level building quick three-dimensional reconstructing method based on laser radar point cloud | |
| CN114387288A (en) | Single standing tree three-dimensional information extraction method based on vehicle-mounted laser radar point cloud data | |
| CN114332291B (en) | Method for extracting outline rule of oblique photography model building | |
| CN115099304A (en) | Building facade point cloud extraction method | |
| CN114862715A (en) | TIN (triangulated irregular network) progressive encryption denoising method fusing terrain feature semantic information | |
| CN116258857A (en) | Outdoor tree-oriented laser point cloud segmentation and extraction method | |
| CN116579949B (en) | Airborne point cloud ground point filtering method suitable for urban multi-noise environment | |
| CN116843693A (en) | A method and system for extracting tree branch skeletons for UAV collection of point cloud data | |
| CN115409969A (en) | Long-distance-measurement ground laser point cloud plane segmentation algorithm | |
| CN117115012A (en) | A road pavement point cloud marking line segmentation and denoising method | |
| CN117132478A (en) | An orbit point cloud denoising method based on normal vector two-norm characteristic parameters | |
| CN118072029B (en) | A vehicle-mounted point cloud single tree segmentation method and system based on improved Thiessen polygon constraints |
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | 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 |