技术领域technical field
本发明涉及车辆安全辅助驾驶技术领域，特别是涉及一种盲区车辆检测方法、系统、车辆变道预警方法及系统。The invention relates to the technical field of vehicle safety assisted driving, in particular to a blind spot vehicle detection method and system, and a vehicle lane change early warning method and system.
背景技术Background technique
随着汽车的不断增多，因变道造成的交通事故不断增多，并且此类事故通常会造成交通堵塞引起交通延误。研究表明，若潜在的交通事故发生前1秒钟给驾驶员预警，则可以避免90%的类似交通事故。因此，快速准确的检测变道盲区内的车辆，识别车辆与本车的距离以及车辆的速度，在驾驶员进行变道操作时及时预警，能够大大提高变道的安全性。With the continuous increase of automobiles, traffic accidents caused by changing lanes continue to increase, and such accidents usually cause traffic jams and cause traffic delays. Studies have shown that if the driver is given a warning 1 second before a potential traffic accident occurs, 90% of similar traffic accidents can be avoided. Therefore, fast and accurate detection of vehicles in the lane-changing blind zone, identification of the distance between the vehicle and the vehicle and the speed of the vehicle, and timely warning when the driver performs a lane-changing operation can greatly improve the safety of lane-changing.
一般而言，造成驾驶员侧方视觉盲点存在的原因有两个，第一是由于人类本身视觉的特性以及视野的限制，第二是车辆在设计时，驾驶员和车辆后视镜的距离以及车辆左右两侧外后视镜本身的成像原理所造成的结果。Generally speaking, there are two reasons for the driver's side visual blind spot. The first is due to the characteristics of human vision and the limitation of the field of vision. The second is the distance between the driver and the vehicle rearview mirror and the It is the result of the imaging principle of the exterior rearview mirrors on the left and right sides of the vehicle.
为了减少车辆侧方视觉盲区的区域，通常有两种比较常见的方法来增加侧方后视镜的视野范围，第一种方式是增加驾驶员和车辆后视镜的距离，但这种方法得到的效果是有限的，因为车辆本身是固定的，所以驾驶员和车辆后视镜间能够调整的距离是相当有限的；第二种方法，就是以各种各样的曲率面镜（双曲率面镜或是变曲率面镜）来取代传统的平面镜，因为曲率面镜中所反射的图像将会变形，并且随着面镜的曲率增加，相对应的变形也就越严重，如果图像变形过度严重，驾驶员也无法利用车辆后视镜判断后方来车和本车的距离。In order to reduce the area of the blind spot on the side of the vehicle, there are usually two common ways to increase the field of view of the side rearview mirror. The first way is to increase the distance between the driver and the vehicle rearview mirror, but this method will get The effect is limited, because the vehicle itself is fixed, so the distance that can be adjusted between the driver and the vehicle rearview mirror is quite limited; the second method is to use various curvature mirrors (double curvature surface mirror or variable curvature mirror) to replace the traditional flat mirror, because the image reflected in the curvature mirror will be deformed, and as the curvature of the mirror increases, the corresponding deformation will be more serious, if the image deformation is too serious , the driver cannot use the rearview mirror of the vehicle to judge the distance between the car coming from behind and the car.
目前，通常使用Adaboost算法对摄像头采集到的整幅盲区图像进行扫描匹配以准确检测出车辆，由此带来了计算量大、计算耗时大等问题，最终导致系统的实时性差，安全性较差。At present, the Adaboost algorithm is usually used to scan and match the entire blind spot image collected by the camera to accurately detect the vehicle, which brings problems such as large amount of calculation and time-consuming calculation, which ultimately leads to poor real-time performance of the system and low security. Difference.
发明内容Contents of the invention
本发明所要解决的技术问题是针对现有的盲区车辆检测方法计算量大所导致的系统反应实时性差、安全性较差的问题，提供一种盲区车辆检测方法。The technical problem to be solved by the present invention is to provide a blind spot vehicle detection method for the problems of poor real-time system response and poor security caused by the large calculation amount of the existing blind spot vehicle detection method.
本发明解决上述技术问题所采用的技术方案为，提供一种盲区车辆检测方法，包括以下步骤：The technical solution adopted by the present invention to solve the above technical problems is to provide a blind spot vehicle detection method, comprising the following steps:
S1、采集包含车辆后视镜盲区的图像；S1. Collect images including blind spots of vehicle rearview mirrors;
S2、使用基于特征的车辆识别方法在上述图像中选取可能存在车辆的图像感兴趣区域；S2. Using a feature-based vehicle recognition method to select an image region of interest where a vehicle may exist in the above image;
S3、使用Adaboost算法在上述图像感兴趣区域内进行扫描匹配以检测车辆。S3. Using the Adaboost algorithm to perform scan matching in the region of interest of the above image to detect the vehicle.
进一步地，步骤S1之后，步骤S2之前还包括图像预处理步骤：Further, after step S1, an image preprocessing step is also included before step S2:
利用高斯滤波器对上述图像进行去噪和平滑处理。The above images were denoised and smoothed using a Gaussian filter.
进一步地，步骤S1具体为：Further, step S1 is specifically:
通过左右两侧摄像头拍摄包含车辆两侧后视镜盲区的图像，并将图像输入给视频解码器；Capture images including the blind spots of the rearview mirrors on both sides of the vehicle through the cameras on the left and right sides, and input the images to the video decoder;
视频解码器将摄像头输入的图像信号解码后输入DSP芯片，DSP芯片通过其上的视频输入接口采集上述图像信号，并将采集到的图像信号存储在存储器中。The video decoder decodes the image signal input by the camera and inputs it into the DSP chip, and the DSP chip collects the above image signal through the video input interface on it, and stores the collected image signal in the memory.
进一步地，步骤S2具体为：Further, step S2 is specifically:
获取上述采集的包含车辆后视镜盲区的图像的灰度直方图，依据车辆灰度值和道路灰度值的不连续性初步选取可能存在车辆的矩形的图像感兴趣区域；Obtain the grayscale histogram of the image collected above that includes the blind area of the vehicle rearview mirror, and initially select a rectangular image region of interest that may exist in the vehicle according to the discontinuity of the grayscale value of the vehicle and the grayscale value of the road;
若选取的图像感兴趣区域远离本车，则进一步利用车辆灰度水平对称特性对上述矩形的图像感兴趣区域进行验证；If the selected image region of interest is far away from the vehicle, further verify the above-mentioned rectangular image region of interest by using the gray level symmetry characteristics of the vehicle;
若验证通过，则选取得到图像感兴趣区域；若验证不通过，则重新依据车辆灰度值和道路灰度值的不连续性初步选取可能存在车辆的矩形的图像感兴趣区域。If the verification is passed, the region of interest in the image is selected; if the verification fails, a rectangular region of interest in the image where vehicles may exist is initially selected based on the discontinuity of the gray value of the vehicle and the gray value of the road.
进一步地，所述“利用车辆灰度水平对称特性对上述矩形的图像感兴趣区域进行验证”所采用的对称性测度公式为：Further, the symmetry measurement formula adopted in the "verification of the above-mentioned rectangular image region of interest by using the symmetry characteristics of the gray level of the vehicle" is:
当S(XS)＝1时，表示上述图像感兴趣区域的水平对称特性为完全对称；当S(XS)＝-1时，表示上述图像感兴趣区域的水平对称特性为完全不对称；当S(XS)＞0时，则表明上述矩形的图像感兴趣区域中存在车辆的可能性较大，即验证通过；当S(XS)≤0时，则表明上述矩形的图像感兴趣区域中存在车辆的可能性较小，即验证不通过；When S(XS )=1, it means that the horizontal symmetry characteristic of the above-mentioned image region of interest is completely symmetrical; when S(XS )=-1, it represents that the horizontal symmetry characteristic of the above-mentioned image region of interest is completely asymmetric; When S(XS )>0, it indicates that there is a greater possibility of vehicles in the region of interest of the above-mentioned rectangular image, that is, the verification is passed; when S(XS )≤0, it indicates that the above-mentioned rectangular image is of interest There is less possibility of vehicles in the area, that is, the verification fails;
上述公式中，E(u)、O(u)分别是函数g(x)＝g(XS+u)的偶函数分量和奇函数分量；In the above formula, E(u), O(u) are respectively the even function component and the odd function component of function g(x)=g(XS +u);
-W/2≤u≤W/2；-W/2≤u≤W/2;
偶函数比重越大，则说明其对称度越高，将偶函数分量E(u)归一化，使其均值为零，则有：The larger the proportion of an even function, the higher its symmetry. Normalize the even function component E(u) to make its mean value zero, then:
上述中，g(x)为将上述矩形的图像感兴趣区域前后方向中间轴所在行的灰度数据视为横坐标的一维函数，该一维函数g(x)的对称轴取为上述图像感兴趣区域的前后方向中间轴XS，W为上述图像感兴趣区域的宽度。In the above, g(x) is a one-dimensional function that regards the grayscale data of the line where the middle axis of the region of interest of the above-mentioned rectangular image is located in the front-back direction as the abscissa, and the symmetry axis of the one-dimensional function g(x) is taken as the above-mentioned image The middle axis XS of the front-back direction of the region of interest, W is the width of the region of interest in the image above.
根据本发明的盲区车辆检测方法，在采集包含车辆后视镜盲区的图像后，先使用基于特征的车辆识别方法在上述图像中选取可能存在车辆的图像感兴趣区域，然后使用Adaboost算法对感兴趣区域进行扫描匹配以检测车辆，这样，盲区图像上除图像感兴趣区域之外的区域无需进行扫描匹配检测，大大减小匹配次数，降低了系统的计算量、计算时间，提高了系统实时性，最终提升了车辆的安全性。According to the blind spot vehicle detection method of the present invention, after collecting images that include the blind spots of vehicle rearview mirrors, first use a feature-based vehicle recognition method to select an image region of interest that may have a vehicle in the above image, and then use the Adaboost algorithm to detect the region of interest. The area is scanned and matched to detect vehicles, so that the areas on the blind area image other than the area of interest in the image do not need to be scanned and matched, which greatly reduces the number of matching times, reduces the calculation amount and calculation time of the system, and improves the real-time performance of the system. Finally, the safety of the vehicle is improved.
另外，本发明还提供了一种盲区车辆检测系统，包括图像拍摄模块及图像处理模块，所述图像处理模块包括控制单元、视频解码器及存储器；In addition, the present invention also provides a blind spot vehicle detection system, including an image capturing module and an image processing module, and the image processing module includes a control unit, a video decoder and a memory;
所述图像拍摄模块，用于拍摄包含车辆后视镜盲区的图像；The image capturing module is used to capture images containing blind spots of vehicle rearview mirrors;
所述图像处理模块，包括图像采集模块及车辆检测模块；所述图像采集模块，用于通过视频解码器将摄像头输入的图像信号解码后输入控制单元，控制单元通过其上的视频输入接口采集上述的图像信号，并将采集到的图像信号存储在存储器中；所述车辆检测模块，使用基于特征的车辆识别方法在上述图像中选取可能存在车辆的图像感兴趣区域，并使用Adaboost算法在上述图像感兴趣区域内进行扫描匹配以检测车辆。The image processing module includes an image acquisition module and a vehicle detection module; the image acquisition module is used to decode the image signal input by the camera through a video decoder and input it to the control unit, and the control unit collects the above-mentioned image signal, and the collected image signal is stored in the memory; the vehicle detection module uses a feature-based vehicle recognition method to select an image region of interest that may have a vehicle in the above image, and uses the Adaboost algorithm to select an image region of interest in the above image Scan matching is performed within the region of interest to detect vehicles.
进一步地，所述图像处理模块还包括连接在所述图像采集模块及车辆检测模块之间的图像预处理模块，所述图像预处理模块利用高斯滤波器对上述图像进行去噪和平滑处理。Further, the image processing module further includes an image preprocessing module connected between the image acquisition module and the vehicle detection module, and the image preprocessing module uses a Gaussian filter to denoise and smooth the above image.
进一步地，所述图像拍摄模块为车辆环视系统的左右侧两个摄像头，所述左右侧两个摄像头分别安装在车辆的左右外后视镜上，所述左右侧两个摄像头的镜头倾斜向下并朝向后方。Further, the image capture module is two cameras on the left and right sides of the vehicle surround view system, the two cameras on the left and right sides are installed on the left and right exterior mirrors of the vehicle respectively, and the lenses of the two cameras on the left and right sides are tilted downward and towards the rear.
另外，本发明还提供了一种车辆变道预警方法，包括如下步骤：In addition, the present invention also provides a vehicle lane change early warning method, comprising the following steps:
根据上述的盲区车辆检测方法检测车辆后视镜盲区内的车辆；Detect vehicles in the blind spot of the vehicle rearview mirror according to the above-mentioned blind spot vehicle detection method;
实时跟踪上述的盲区车辆检测方法检测到的车辆，以检测跟踪车辆的形状、尺寸、移动速度、移动方向及当前位置；Real-time tracking of the vehicle detected by the above blind spot vehicle detection method to detect and track the shape, size, moving speed, moving direction and current position of the vehicle;
根据检测到的车辆的形状、尺寸、移动速度、移动方向及当前位置进行逻辑判断，确定驾驶员当前变道行为是否需要预警；According to the shape, size, moving speed, moving direction and current position of the detected vehicle, logical judgment is made to determine whether the driver's current lane-changing behavior needs to be warned;
在确定需要预警的情况下，以声和/或光的形式预警。In the event that an early warning is determined to be necessary, an early warning is given by sound and/or light.
根据本发明的车辆变道预警方法，在采集包含车辆后视镜盲区的图像后，先使用基于特征的车辆识别方法在上述图像中选取可能存在车辆的图像感兴趣区域，然后使用Adaboost算法对感兴趣区域进行扫描匹配以检测车辆，这样，盲区图像上除图像感兴趣区域之外的区域无需进行扫描匹配检测，大大减小匹配次数，降低了系统的计算量、计算时间，提高了系统实时性，最终提升了车辆的安全性。According to the vehicle lane change early warning method of the present invention, after collecting images that include the blind spots of vehicle rearview mirrors, first use a feature-based vehicle recognition method to select image regions of interest that may have vehicles in the above images, and then use the Adaboost algorithm to detect The area of interest is scanned and matched to detect vehicles, so that the areas on the blind area image other than the area of interest in the image do not need to be scanned and matched, which greatly reduces the number of matching times, reduces the calculation amount and time of the system, and improves the real-time performance of the system , ultimately improving the safety of the vehicle.
另外，本发明还提供了一种车辆变道预警系统，包括权上述的盲区车辆检测系统、车辆跟踪模块、预警逻辑判断模块及预警模块；In addition, the present invention also provides a vehicle lane change early warning system, including the above blind spot vehicle detection system, vehicle tracking module, early warning logic judgment module and early warning module;
所述车辆跟踪模块，用于实时跟踪上述的盲区车辆检测系统检测到的车辆，以检测跟踪车辆的形状、尺寸、移动速度、移动方向及当前位置；The vehicle tracking module is used to track the vehicle detected by the above-mentioned blind spot vehicle detection system in real time, so as to detect the shape, size, moving speed, moving direction and current position of the tracking vehicle;
所述预警逻辑判断模块，根据所述车辆跟踪模块检测到的车辆的形状、尺寸、移动速度、移动方向及当前位置进行逻辑判断，确定驾驶员当前变道行为是否需要预警；The pre-warning logical judgment module performs logical judgment according to the shape, size, moving speed, moving direction and current position of the vehicle detected by the vehicle tracking module, to determine whether the driver's current lane-changing behavior needs a pre-warning;
所述预警模块，用于在所述预警逻辑判断模块确定需要预警的情况下，以声和/或光的形式预警。The early warning module is configured to issue an early warning in the form of sound and/or light when the early warning logic judging module determines that an early warning is required.
附图说明Description of drawings
图1是本发明一实施例提供的盲区车辆检测方法的示意图；Fig. 1 is a schematic diagram of a blind spot vehicle detection method provided by an embodiment of the present invention;
图2是本发明一实施例提供的盲区车辆检测方法的盲区车辆检测示意图；Fig. 2 is a schematic diagram of blind spot vehicle detection according to a blind spot vehicle detection method provided by an embodiment of the present invention;
图3是使用Adaboost算法在图像感兴趣区域内进行扫描匹配以检测车辆步骤的示意图；Fig. 3 is a schematic diagram of the step of using the Adaboost algorithm to perform scan matching in the region of interest of the image to detect the vehicle;
图4是本发明一实施例提供的盲区车辆检测系统的示意图；Fig. 4 is a schematic diagram of a blind spot vehicle detection system provided by an embodiment of the present invention;
图5是本发明一实施例提供的车辆变道预警方法的示意图；Fig. 5 is a schematic diagram of a vehicle lane change warning method provided by an embodiment of the present invention;
图6是本发明一实施例提供的车辆变道预警方法实时跟踪盲区车辆方法步骤的示意图；Fig. 6 is a schematic diagram of the steps of the method for tracking vehicles in blind spots in real time according to the vehicle lane change warning method provided by an embodiment of the present invention;
图7是本发明一实施例提供的车辆变道预警系统的示意图。Fig. 7 is a schematic diagram of a vehicle lane change warning system provided by an embodiment of the present invention.
附图中的标记如下：The marks in the accompanying drawings are as follows:
1、本车辆；2、后视镜盲区；3、盲区内车辆；10、图像拍摄模块；20、图像处理模块；21、图像采集模块；22、车辆检测模块；23、图像预处理模块；24、车辆跟踪模块；25、预警逻辑判断模块；30、预警模块；SL、左外后视镜可视区域；SR、右外后视镜可视区域；SI、内后视镜区域。1. The vehicle; 2. Rearview mirror blind spot; 3. Vehicles in the blind spot; 10. Image capture module; 20. Image processing module; 21. Image acquisition module; 22. Vehicle detection module; 23. Image preprocessing module; 24 . Vehicle tracking module; 25. Early warning logic judgment module; 30. Early warning module; SL, visible area of left outer rearview mirror; SR, visible area of right outer rearview mirror; SI, inner rearview mirror area.
具体实施方式Detailed ways
为了使本发明所解决的技术问题、技术方案及有益效果更加清楚明白，以下结合附图及实施例，对本发明进行进一步的详细说明。应当理解，此处所描述的具体实施例仅仅用以解释本发明，并不用于限定本发明。In order to make the technical problems, technical solutions and beneficial effects solved by the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.
如图1所示，本发明一实施例提供了一种盲区车辆检测方法，包括以下步骤：As shown in Figure 1, an embodiment of the present invention provides a blind spot vehicle detection method, including the following steps:
S1、采集包含车辆后视镜盲区的图像；S1. Collect images including blind spots of vehicle rearview mirrors;
S2、使用基于特征的车辆识别方法在上述图像中选取可能存在车辆的图像感兴趣区域；基于特征的方法又称为基于知识的方法（Knowledge-BasedMethods）。公路上行驶的车辆在灰度图像中具有一些明显的特征：形状特征，大体为矩形，而且满足特殊的形状特征；灰度特征，一般情况下车辆在图像中与背景灰度有显著差异，车辆底部存在灰度值较小的阴影区域；对称特征，车辆的对称特征包括灰度对称、水平边缘和垂直边缘对称。S2. Using a feature-based vehicle recognition method to select an image region of interest where a vehicle may exist in the above image; feature-based methods are also called knowledge-based methods (Knowledge-Based Methods). Vehicles driving on the road have some obvious features in grayscale images: shape features, which are generally rectangular, and meet special shape features; There is a shadow area with a small gray value at the bottom; symmetrical features, the symmetrical features of the vehicle include gray scale symmetry, horizontal edge symmetry, and vertical edge symmetry.
S3、使用Adaboost算法在上述图像感兴趣区域内进行扫描匹配以检测车辆。S3. Using the Adaboost algorithm to perform scan matching in the region of interest of the above image to detect the vehicle.
本实施例中，步骤S3检测盲区车辆的形状、尺寸、移动速度、移动方向及当前位置等特征。In this embodiment, step S3 detects features such as the shape, size, moving speed, moving direction, and current position of the vehicle in the blind spot.
本实施例中，步骤S1之后，步骤S2之前还包括图像预处理步骤。所述图像预处理步骤为利用高斯滤波器对上述图像进行去噪和平滑处理，以提高图像质量。In this embodiment, an image preprocessing step is also included after step S1 and before step S2. The image preprocessing step is to use a Gaussian filter to perform denoising and smoothing processing on the above image, so as to improve image quality.
本实施例中，步骤S1具体为：In this embodiment, step S1 is specifically:
如图2所示，通过左右两侧摄像头拍摄包含车辆两侧后视镜盲区的图像，并将图像输入给视频解码器；图2中，本车辆1两侧的后视镜盲区2用剖面线表示，后视镜盲区2中示出了盲区内车辆3，另外SL表示左外后视镜可视区域，SR表示右外后视镜可视区域，SI表示内后视镜区域。As shown in Figure 2, the images containing the blind spots of the rearview mirrors on both sides of the vehicle are captured by the cameras on the left and right sides, and the images are input to the video decoder; in Figure 2, the blind spots 2 of the rearview mirrors on both sides of the vehicle 1 are hatched Indicates that the rearview mirror blind zone 2 shows the vehicle 3 in the blind zone, and SL indicates the visible area of the left outer rearview mirror, SR indicates the visible area of the right outer rearview mirror, and SI indicates the area of the inner rearview mirror.
视频解码器将摄像头输入的图像信号（模拟信号）解码为YUV的数字信号后输入DSP（Digital Signal Processing，数字信号处理器）芯片，DSP芯片通过其上的视频输入接口采集上述图像信号，并将采集到的图像信号存储在闪存Flash和/或内存DDR类型的存储器中。The video decoder decodes the image signal (analog signal) input by the camera into a YUV digital signal and then inputs it into the DSP (Digital Signal Processing, digital signal processor) chip. The DSP chip collects the above image signal through the video input interface on it, and sends The captured image signals are stored in Flash and/or DDR type memory.
本实施例中，步骤S2具体为：In this embodiment, step S2 is specifically:
获取上述采集的包含车辆后视镜盲区的图像的灰度直方图，依据车辆灰度值和道路灰度值的不连续性初步选取可能存在车辆的矩形的图像感兴趣区域。Obtain the gray histogram of the image collected above that includes the blind area of the vehicle rearview mirror, and initially select a rectangular image region of interest that may contain vehicles based on the discontinuity of the gray value of the vehicle and the gray value of the road.
若选取的图像感兴趣区域远离本车，则进一步利用车辆灰度水平对称特性对上述矩形的图像感兴趣区域进行验证。若验证通过，则选取得到图像感兴趣区域；若验证不通过，则重新依据车辆灰度值和道路灰度值的不连续性初步选取可能存在车辆的矩形的图像感兴趣区域，再次验证。在侧方道路中，因为拍摄的角度原因，当侧方车辆离本车较近时，车辆前部的对称性不高。因此本方法只在当侧方车辆离本车较远的情况下（即图像感兴趣区域远离本车的情况下）利用车辆灰度水平对称性对侧方的图像感兴趣区域进行验证。这样有利于减少验证的时间，并且选取的图像感兴趣区域中存在车辆的可能性增加，有利于减少计算量，提高系统的实时性。If the selected image ROI is far away from the vehicle, then further use the vehicle gray level symmetry characteristics to verify the above rectangular image ROI. If the verification is passed, the region of interest in the image is selected; if the verification fails, a rectangular region of interest in the image that may contain vehicles is initially selected based on the discontinuity of the gray value of the vehicle and the gray value of the road, and the verification is performed again. On the side road, due to the shooting angle, when the side vehicle is close to the vehicle, the symmetry of the front of the vehicle is not high. Therefore, this method only uses the symmetry of the gray level of the vehicle to verify the side image ROI when the side vehicle is far away from the vehicle (that is, the image ROI is far away from the vehicle). This helps to reduce the verification time, and increases the possibility of vehicles in the selected image interest area, which is beneficial to reduce the amount of calculation and improve the real-time performance of the system.
本实施例中，所述“利用车辆灰度水平对称特性对上述矩形的图像感兴趣区域进行验证”所采用的对称性测度公式为：In this embodiment, the symmetry measurement formula adopted in the "verification of the above-mentioned rectangular image region of interest by using the symmetry characteristic of the gray level of the vehicle" is:
当S(XS)＝1时，表示上述图像感兴趣区域的水平对称特性为完全对称；当S(XS)＝-1时，表示上述图像感兴趣区域的水平对称特性为完全不对称；当S(XS)＞0时，则表明上述矩形的图像感兴趣区域中存在车辆的可能性较大，即验证通过；当S(XS)≤0时，则表明上述矩形的图像感兴趣区域中存在车辆的可能性较小，即验证不通过；When S(XS )=1, it means that the horizontal symmetry characteristic of the above-mentioned image region of interest is completely symmetrical; when S(XS )=-1, it represents that the horizontal symmetry characteristic of the above-mentioned image region of interest is completely asymmetric; When S(XS )>0, it indicates that there is a greater possibility of vehicles in the region of interest of the above-mentioned rectangular image, that is, the verification is passed; when S(XS )≤0, it indicates that the above-mentioned rectangular image is of interest There is less possibility of vehicles in the area, that is, the verification fails;
上述公式中，E(u)、O(u)分别是函数g(x)＝g(XS+u)的偶函数分量和奇函数分量；In the above formula, E(u), O(u) are respectively the even function component and the odd function component of function g(x)=g(XS +u);
-W/2≤u≤W/2；-W/2≤u≤W/2;
偶函数比重越大，则说明其对称度越高，将偶函数分量E(u)归一化，使其均值为零，则有：The larger the proportion of an even function, the higher its symmetry. Normalize the even function component E(u) to make its mean value zero, then:
上述中，g(x)为将上述矩形的图像感兴趣区域前后方向中间轴所在行的灰度数据视为横坐标的一维函数，该一维函数g(x)的对称轴取为上述图像感兴趣区域的前后方向中间轴XS，W为上述图像感兴趣区域的宽度。In the above, g(x) is a one-dimensional function that regards the grayscale data of the line where the middle axis of the region of interest of the above-mentioned rectangular image is located in the front-back direction as the abscissa, and the symmetry axis of the one-dimensional function g(x) is taken as the above-mentioned image The middle axis XS of the front-back direction of the region of interest, W is the width of the region of interest in the image above.
在使用Adaboost算法在上述图像感兴趣区域内进行扫描匹配以检测车辆之前，利用车辆灰度水平对称特性对上述矩形的图像感兴趣区域进行验证，这样只有验证通过了才进入后续的步骤，大大提高了车辆识别率，也避免了过多的计算，有利于减少计算量，进一步提高系统的实时性。Before using the Adaboost algorithm to perform scan matching in the region of interest of the above image to detect the vehicle, the gray level symmetry characteristics of the vehicle are used to verify the region of interest of the above rectangular image, so that only after the verification is passed can the subsequent steps be entered, greatly improving It improves the vehicle recognition rate and avoids excessive calculation, which is beneficial to reduce the amount of calculation and further improve the real-time performance of the system.
本实施例中，步骤S3具体为：In this embodiment, step S3 is specifically:
如图3所示，包括左边的离线训练和右边的在线检测，离线训练为对车辆样本和非车辆样本进行预处理，通过对车辆样本和非车辆样本的训练得到若干分类器；在线检测为利用离线训练得到的分类器对每一帧输入图像的感兴趣区域进行扫描匹配，检测图像感兴趣区域是否存在车辆。此处的预处理为采用直方图归一化处理所有的样本，这样可以减小图像本身由于灰度分布造成的影响。离线训练中对分类器的训练算法是本领域通用的技术手段，本发明不再详述。As shown in Figure 3, it includes offline training on the left and online detection on the right. Offline training is to preprocess vehicle samples and non-vehicle samples, and several classifiers are obtained by training vehicle samples and non-vehicle samples; online detection is to use The classifier obtained by offline training scans and matches the ROI of each frame input image to detect whether there is a vehicle in the ROI of the image. The preprocessing here is to use histogram normalization to process all samples, which can reduce the influence of the image itself due to the gray distribution. The training algorithm for the classifier in offline training is a common technical means in the field, and will not be described in detail in the present invention.
根据本发明上述实施例的盲区车辆检测方法，在采集包含车辆后视镜盲区的图像后，先使用基于特征的车辆识别方法在上述图像中选取可能存在车辆的图像感兴趣区域，然后使用Adaboost算法对感兴趣区域进行扫描匹配以检测车辆，这样，盲区图像上除图像感兴趣区域之外的区域无需进行扫描匹配检测，大大减小匹配次数，降低了系统的计算量、计算时间，提高了系统实时性，最终提升了车辆的安全性。According to the vehicle detection method in the blind spot of the above-mentioned embodiment of the present invention, after collecting the image comprising the blind spot of the rearview mirror of the vehicle, first use a feature-based vehicle recognition method to select an image region of interest that may have a vehicle in the above image, and then use the Adaboost algorithm Scan and match the area of interest to detect the vehicle, so that the area on the blind spot image other than the area of interest does not need to be scanned and matched, which greatly reduces the number of matches, reduces the amount of calculation and calculation time of the system, and improves the system performance. Real-time, and ultimately enhance the safety of the vehicle.
另外，如图4所示，本发明一实施例还提供了一种盲区车辆检测系统，包括图像拍摄模块10及图像处理模块20，所述图像处理模块包括控制单元、视频解码器及存储器；In addition, as shown in FIG. 4 , an embodiment of the present invention also provides a vehicle detection system in a blind spot, including an image capturing module 10 and an image processing module 20, and the image processing module includes a control unit, a video decoder and a memory;
所述图像拍摄模块10，用于拍摄包含车辆后视镜盲区的图像；The image capture module 10 is used to capture an image comprising the blind area of the rearview mirror of the vehicle;
所述图像处理模块20，包括图像采集模块21及车辆检测模块22；The image processing module 20 includes an image acquisition module 21 and a vehicle detection module 22;
所述图像采集模块，用于通过视频解码器将摄像头输入的图像信号解码后输入控制单元，控制单元通过其上的视频输入接口采集上述的图像信号，并将采集到的图像信号存储在存储器中；存储器可以是DDR内存或FLASH闪存。优选地，所述控制单元为DSP芯片。The image acquisition module is used to decode the image signal input by the camera through the video decoder and input it to the control unit, and the control unit collects the above-mentioned image signal through the video input interface on it, and stores the collected image signal in the memory ; The memory can be DDR memory or FLASH flash memory. Preferably, the control unit is a DSP chip.
所述车辆检测模块，使用基于特征的车辆识别方法在上述图像中选取可能存在车辆的图像感兴趣区域，并使用Adaboost算法在上述图像感兴趣区域内进行扫描匹配以检测车辆。车辆检测模块22集成在DSP芯片中，通过在DSP芯片中写入相应软件来实现相应功能。The vehicle detection module uses a feature-based vehicle recognition method to select an image region of interest where a vehicle may exist in the above-mentioned image, and uses the Adaboost algorithm to perform scan matching in the above-mentioned image region of interest to detect the vehicle. The vehicle detection module 22 is integrated in the DSP chip, and corresponding functions are realized by writing corresponding software in the DSP chip.
本实施例中，所述图像处理模块20还包括连接在所述图像采集模块及车辆检测模块之间的图像预处理模块23，所述图像预处理模块23利用高斯滤波器对上述图像进行去噪和平滑处理。图像预处理模块23集成在DSP芯片中。In this embodiment, the image processing module 20 also includes an image preprocessing module 23 connected between the image acquisition module and the vehicle detection module, and the image preprocessing module 23 uses a Gaussian filter to denoise the above image and smoothing. The image preprocessing module 23 is integrated in the DSP chip.
本实施例中，所述图像拍摄模块10为车辆环视系统的左右侧两个摄像头，所述左右侧两个摄像头分别安装在车辆的左右外后视镜上，所述左右侧两个摄像头的镜头倾斜向下并朝向后方。这样可以得到后视镜盲区内的实时状况，具体的安装倾斜角度根据不同的车型和车身大小而有所不同，这个可以通过实测得到较优的安装倾斜角度。另外，利用车辆现有的环视系统来实现盲区车辆检测，而无需增加其它设备，有利于减少零部件及降低生产成本。In this embodiment, the image capture module 10 is two cameras on the left and right sides of the vehicle surround view system, and the two cameras on the left and right sides are installed on the left and right exterior rearview mirrors of the vehicle respectively, and the lenses of the two cameras on the left and right sides Tilt down and towards the rear. In this way, the real-time situation in the blind area of the rearview mirror can be obtained. The specific installation inclination angle varies according to different models and body sizes. This can obtain a better installation inclination angle through actual measurement. In addition, the existing surround view system of the vehicle is used to detect vehicles in blind spots without adding other equipment, which is conducive to reducing parts and production costs.
另外，如图5所示，本发明一实施例还提供了一种车辆变道预警方法，包括如下步骤：In addition, as shown in FIG. 5 , an embodiment of the present invention also provides a vehicle lane change warning method, including the following steps:
根据上述的盲区车辆检测方法检测车辆后视镜盲区内的车辆；此步骤包括：Detect vehicles in the blind spot of the vehicle rearview mirror according to the above-mentioned blind spot vehicle detection method; this step includes:
S1、采集包含车辆后视镜盲区的图像；S1. Collect images including blind spots of vehicle rearview mirrors;
S2、使用基于特征的车辆识别方法在上述图像中选取可能存在车辆的图像感兴趣区域；S2. Using a feature-based vehicle recognition method to select an image region of interest where a vehicle may exist in the above image;
S3、使用Adaboost算法在上述图像感兴趣区域内进行扫描匹配以检测车辆。S3. Using the Adaboost algorithm to perform scan matching in the region of interest of the above image to detect the vehicle.
然后，实时跟踪上述的盲区车辆检测方法检测到的车辆，以检测跟踪车辆的形状、尺寸、移动速度、移动方向及当前位置。Then, the vehicle detected by the above-mentioned blind spot vehicle detection method is tracked in real time to detect the shape, size, moving speed, moving direction and current position of the tracked vehicle.
根据检测到的车辆的形状、尺寸、移动速度、移动方向及当前位置进行逻辑判断，确定驾驶员当前变道行为是否需要预警；此方法步骤为公知的技术手段，本发明不再不详细描述。Carry out logical judgment based on the detected shape, size, moving speed, moving direction and current position of the vehicle to determine whether the driver's current lane-changing behavior needs early warning; the steps of this method are known technical means, and will not be described in detail in the present invention.
在确定需要预警的情况下，以声和/或光的形式预警。此方法步骤为公知的技术手段，本发明不再不详细描述。In the event that an early warning is determined to be necessary, an early warning is given by sound and/or light. The method steps are known technical means, and the present invention will not describe them in detail.
本实施例中，如图6所示，所述“实时跟踪上述的盲区车辆检测系统检测到的车辆”具体为：In this embodiment, as shown in FIG. 6, the "real-time tracking of vehicles detected by the above-mentioned blind spot vehicle detection system" is specifically:
通过上述盲区车辆检测方法前一帧的检测结果预测车辆在下一帧图像中的位置和尺寸，然后使用Adaboost算法对感兴趣区域进行扫描匹配以准确检测车辆（即检测得到跟踪车辆的形状、尺寸、移动速度、移动方向及当前位置），比较检测出的车辆和预测车辆的位置和尺寸，判断两者的误差是否在合理范围内，若是则以同样的方式继续跟踪车辆；若否，则跟踪结束。“判断两者的误差是否在合理范围内”具体为，当生成车辆的新的检测区域后（新的检测区域是指根据当前帧检测到得车辆的位置预测的下一帧图像中可能存在车辆的位置），系统将在下一帧图像中对这个区域进行车辆的检测及验证。验证成功后，计算整个矩形框（图像感兴趣区域）的灰度差异程度，判断新的车辆边界（预测的车辆边界）与前一帧检测到的车辆边界的重合程度，以此判断新的检测区域中的车辆与前一帧检测到的车辆是否是相同的车辆。The position and size of the vehicle in the next frame of image are predicted by the detection result of the previous frame of the blind spot vehicle detection method, and then the Adaboost algorithm is used to scan and match the region of interest to accurately detect the vehicle (that is, the shape, size, and shape of the tracked vehicle are detected by detection). moving speed, moving direction, and current position), compare the detected position and size of the vehicle with the predicted vehicle, and judge whether the error between the two is within a reasonable range. If so, continue to track the vehicle in the same way; if not, the tracking ends . "Judging whether the error between the two is within a reasonable range" is specifically, when a new detection area of the vehicle is generated (the new detection area refers to the possibility that there may be a vehicle in the next frame image predicted based on the position of the vehicle detected in the current frame position), the system will detect and verify the vehicle in this area in the next frame of image. After the verification is successful, calculate the gray difference degree of the entire rectangular frame (image region of interest), and judge the coincidence degree of the new vehicle boundary (predicted vehicle boundary) with the vehicle boundary detected in the previous frame, so as to judge the new detection Whether the vehicle in the area is the same vehicle as detected in the previous frame.
车辆跟踪算法具体为，假定侧方车辆的速度或者加速度不变，定义车辆边界框的中心点为(x,y)，Pt为时刻t车辆边界框中心点的位置，Pt-1和Pt-2分别为t-1和t-2时刻车辆边界框中心点的位置，Pt+1为下一帧所预测的车辆位置，则有：Specifically, the vehicle tracking algorithm is assuming that the speed or acceleration of the side vehicle remains unchanged, and the center point of the vehicle bounding box is defined as (x, y), Pt is the position of the center point of the vehicle bounding box at time t, Pt-1 and Pt-2 is the position of the center point of the vehicle bounding box at time t-1 and t-2 respectively, and Pt+1 is the predicted vehicle position in the next frame, then:
由上式可得，下一帧所预测的车辆位置Pt+1为：From the above formula, the predicted vehicle position Pt+1 in the next frame is:
根据得到的预测位置Pt+1，就可以在它周围继续进行车辆的跟踪。According to the obtained predicted position Pt+1 , the tracking of the vehicle can be continued around it.
当侧方跟踪车辆正在进行超车行为并且快速接近本车时，被跟踪车辆在短时间内速度提升，因此相邻帧图像中车辆边界框的大小会有较大变化，此时如果继续采用上述算法对车辆边界框位置进行预测，将会产生误差，算法的鲁棒性降低。因此需要扩大检测区域，扩大检测区域可以通过下式得出：When the side tracking vehicle is overtaking and rapidly approaching the vehicle, the tracked vehicle speeds up in a short period of time, so the size of the vehicle bounding box in the adjacent frame image will change greatly. At this time, if we continue to use the above algorithm Predicting the position of the vehicle's bounding box will generate errors and reduce the robustness of the algorithm. Therefore, it is necessary to expand the detection area, which can be obtained by the following formula:
Range(Pt+1,aWt,aht)；Range(Pt+1 ,aWt ,aht );
式中Range(P,W,h)为侧方跟踪车辆的检测区域，Pt+1为所预测的检测范围中心，Wt和ht分别为前一帧车辆边界框的宽度和高度，a为下一帧图像车辆边界扩展的倍数。Range(Pt+1,aWt,aht)即表示扩大后的检测区域。In the formula, Range(P,W,h) is the detection area of the side tracking vehicle, Pt+1 is the center of the predicted detection range, Wt and ht are the width and height of the vehicle bounding box in the previous frame, respectively, a The multiplier of the vehicle boundary expansion for the next frame image. Range(Pt+1 , aWt , aht ) means the enlarged detection area.
根据本发明上述实施例的车辆变道预警方法，在采集包含车辆后视镜盲区的图像后，先使用基于特征的车辆识别方法在上述图像中选取可能存在车辆的图像感兴趣区域，然后使用Adaboost算法对感兴趣区域进行扫描匹配以检测车辆，这样，盲区图像上除图像感兴趣区域之外的区域无需进行扫描匹配检测，大大减小匹配次数，降低了系统的计算量、计算时间，提高了系统实时性，最终提升了车辆的安全性。According to the vehicle lane change warning method of the above-mentioned embodiment of the present invention, after collecting the image that includes the blind area of the vehicle rearview mirror, first use the feature-based vehicle recognition method to select the image region of interest that may have a vehicle in the above image, and then use Adaboost The algorithm scans and matches the region of interest to detect the vehicle. In this way, the area on the blind spot image other than the region of interest does not need to be scanned and matched, which greatly reduces the number of matches, reduces the amount of calculation and calculation time of the system, and improves the efficiency of the vehicle. The real-time performance of the system ultimately improves the safety of the vehicle.
另外，如图7所示，本发明一实施例还提供了一种车辆变道预警系统，包括上述的盲区车辆检测系统、车辆跟踪模块24、预警逻辑判断模块25及预警模块30；所述车辆跟踪模块、预警逻辑判断模块均集成在DSP芯片中，通过在DSP芯片中写入相应软件来实现相应功能，即车辆跟踪模块24及预警逻辑判断模块25为图像处理模块的一部分。In addition, as shown in Figure 7, an embodiment of the present invention also provides a vehicle lane change early warning system, including the above-mentioned blind spot vehicle detection system, vehicle tracking module 24, early warning logic judgment module 25 and early warning module 30; The tracking module and the early warning logic judgment module are all integrated in the DSP chip, and the corresponding functions are realized by writing corresponding software in the DSP chip, that is, the vehicle tracking module 24 and the early warning logic judgment module 25 are part of the image processing module.
所述车辆跟踪模块24，用于实时跟踪上述的盲区车辆检测系统检测到的车辆，以检测跟踪车辆的形状、尺寸、移动速度、移动方向及当前位置。The vehicle tracking module 24 is used to track the vehicles detected by the blind spot vehicle detection system in real time, so as to detect the shape, size, moving speed, moving direction and current position of the tracked vehicles.
所述预警逻辑判断模块25，根据所述车辆跟踪模块24检测到的车辆的形状、尺寸、移动速度、移动方向及当前位置进行逻辑判断，确定驾驶员当前变道行为是否需要预警；此为公知的技术手段，本发明不再不详细描述。The early warning logic judgment module 25 performs logical judgment according to the shape, size, moving speed, moving direction and current position of the vehicle detected by the vehicle tracking module 24 to determine whether the driver's current lane-changing behavior needs early warning; this is known technical means, the present invention is no longer described in detail.
所述预警模块30，用于在所述预警逻辑判断模块确定需要预警的情况下，以声和/或光的形式预警。例如通过蜂鸣器发出预警，或者是在车辆DVD上显示预警信息，或者是在仪表盘液晶显示屏上显示预警信息。此为公知的技术手段，本发明不再不详细描述。The early warning module 30 is configured to issue an early warning in the form of sound and/or light when the early warning logic judging module determines that an early warning is required. For example, an early warning is issued through a buzzer, or an early warning information is displayed on a vehicle DVD, or an early warning information is displayed on an instrument panel LCD. This is a known technical means, and will not be described in detail in the present invention.
以上所述仅为本发明的较佳实施例而已，并不用以限制本发明，凡在本发明的精神和原则之内所作的任何修改、等同替换和改进等，均应包含在本发明的保护范围之内。The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present invention should be included in the protection of the present invention. within range.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310450273.1ACN104512327B (en) | 2013-09-27 | 2013-09-27 | Blind area vehicle checking method, system, vehicle lane change method for early warning and system |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310450273.1ACN104512327B (en) | 2013-09-27 | 2013-09-27 | Blind area vehicle checking method, system, vehicle lane change method for early warning and system |
| Publication Number | Publication Date |
|---|---|
| CN104512327Atrue CN104512327A (en) | 2015-04-15 |
| CN104512327B CN104512327B (en) | 2017-11-21 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310450273.1AActiveCN104512327B (en) | 2013-09-27 | 2013-09-27 | Blind area vehicle checking method, system, vehicle lane change method for early warning and system |
| Country | Link |
|---|---|
| CN (1) | CN104512327B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104960523A (en)* | 2015-06-25 | 2015-10-07 | 奇瑞汽车股份有限公司 | Intelligent lane changing assisting system for intelligent vehicle and control method thereof |
| CN105438072A (en)* | 2015-12-28 | 2016-03-30 | 新昌县楷翔机电有限公司 | Automobile with wide-angle rear view field |
| CN106864370A (en)* | 2017-03-15 | 2017-06-20 | 上汽通用汽车有限公司 | The backsight method for displaying image of vehicle, device and system |
| CN107031623A (en)* | 2017-03-16 | 2017-08-11 | 浙江零跑科技有限公司 | A kind of road method for early warning based on vehicle-mounted blind area camera |
| CN107031661A (en)* | 2017-03-16 | 2017-08-11 | 浙江零跑科技有限公司 | A kind of lane change method for early warning and system based on blind area camera input |
| CN108674313A (en)* | 2018-06-05 | 2018-10-19 | 浙江零跑科技有限公司 | A kind of blind area early warning system and method based on vehicle-mounted back vision wide angle camera |
| CN109177911A (en)* | 2018-09-19 | 2019-01-11 | 同致电子科技（昆山）有限公司 | One kind being based on the jamproof blind area detection system of ultrasonic wave |
| CN109552173A (en)* | 2017-09-26 | 2019-04-02 | 纵目科技（上海）股份有限公司 | Full visual field camera engine control system |
| CN109552315A (en)* | 2017-09-26 | 2019-04-02 | 纵目科技（上海）股份有限公司 | Full visual field camera engine control system |
| CN111038390A (en)* | 2019-12-23 | 2020-04-21 | 天津布尔科技有限公司 | Car blind area monitoring devices based on multisensor fuses |
| CN111038389A (en)* | 2019-12-23 | 2020-04-21 | 天津布尔科技有限公司 | Large-scale vehicle blind area monitoring devices based on vision |
| CN111083362A (en)* | 2019-12-13 | 2020-04-28 | 山东浪潮通软信息科技有限公司 | Method for realizing automatic focusing of vehicle entering and exiting warehouse |
| CN111077506A (en)* | 2019-12-12 | 2020-04-28 | 苏州智加科技有限公司 | Method, device and system for calibrating millimeter wave radar |
| CN113239912A (en)* | 2021-07-13 | 2021-08-10 | 天津所托瑞安汽车科技有限公司 | Method, device and storage medium for determining BSD image effective area |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101016053A (en)* | 2007-01-25 | 2007-08-15 | 吉林大学 | Warning method and system for preventing collision for vehicle on high standard highway |
| CN102456226B (en)* | 2010-10-22 | 2014-06-04 | 财团法人工业技术研究院 | Tracking methods for regions of interest |
| CN102632839B (en)* | 2011-02-15 | 2015-04-01 | 香港生产力促进局 | A vehicle blind spot early warning system and method based on rear view image cognition |
| CN102201059A (en)* | 2011-05-20 | 2011-09-28 | 北京大学深圳研究生院 | Pedestrian detection method and device |
| CN102592114B (en)* | 2011-12-26 | 2013-07-31 | 河南工业大学 | Method for extracting and recognizing lane line features of complex road conditions |
| CN203157846U (en)* | 2013-01-16 | 2013-08-28 | 郑州宇通客车股份有限公司 | Vehicle driving auxiliary system based on DSP |
| Title |
|---|
| 李建科: ""基于DSP的车辆目标检测系统设计与研究"", 《中国优秀硕士学位论文全文数据库信息科技辑》* |
| 赵宇峰: ""汽车后视镜盲区检测及预警关键技术研究"", 《中国优秀硕士学位论文全文数据库信息科技辑》* |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104960523A (en)* | 2015-06-25 | 2015-10-07 | 奇瑞汽车股份有限公司 | Intelligent lane changing assisting system for intelligent vehicle and control method thereof |
| CN105438072A (en)* | 2015-12-28 | 2016-03-30 | 新昌县楷翔机电有限公司 | Automobile with wide-angle rear view field |
| CN105438072B (en)* | 2015-12-28 | 2018-10-26 | 新昌县楷翔机电有限公司 | A kind of automobile with the wide-angle backsight visual field |
| CN106864370A (en)* | 2017-03-15 | 2017-06-20 | 上汽通用汽车有限公司 | The backsight method for displaying image of vehicle, device and system |
| CN107031623A (en)* | 2017-03-16 | 2017-08-11 | 浙江零跑科技有限公司 | A kind of road method for early warning based on vehicle-mounted blind area camera |
| CN107031661A (en)* | 2017-03-16 | 2017-08-11 | 浙江零跑科技有限公司 | A kind of lane change method for early warning and system based on blind area camera input |
| CN107031623B (en)* | 2017-03-16 | 2019-09-20 | 浙江零跑科技有限公司 | A kind of road method for early warning based on vehicle-mounted blind area camera |
| CN109552173A (en)* | 2017-09-26 | 2019-04-02 | 纵目科技（上海）股份有限公司 | Full visual field camera engine control system |
| CN109552315A (en)* | 2017-09-26 | 2019-04-02 | 纵目科技（上海）股份有限公司 | Full visual field camera engine control system |
| CN108674313A (en)* | 2018-06-05 | 2018-10-19 | 浙江零跑科技有限公司 | A kind of blind area early warning system and method based on vehicle-mounted back vision wide angle camera |
| CN109177911A (en)* | 2018-09-19 | 2019-01-11 | 同致电子科技（昆山）有限公司 | One kind being based on the jamproof blind area detection system of ultrasonic wave |
| CN111077506A (en)* | 2019-12-12 | 2020-04-28 | 苏州智加科技有限公司 | Method, device and system for calibrating millimeter wave radar |
| CN111083362A (en)* | 2019-12-13 | 2020-04-28 | 山东浪潮通软信息科技有限公司 | Method for realizing automatic focusing of vehicle entering and exiting warehouse |
| CN111038390A (en)* | 2019-12-23 | 2020-04-21 | 天津布尔科技有限公司 | Car blind area monitoring devices based on multisensor fuses |
| CN111038389A (en)* | 2019-12-23 | 2020-04-21 | 天津布尔科技有限公司 | Large-scale vehicle blind area monitoring devices based on vision |
| CN113239912A (en)* | 2021-07-13 | 2021-08-10 | 天津所托瑞安汽车科技有限公司 | Method, device and storage medium for determining BSD image effective area |
| Publication number | Publication date |
|---|---|
| CN104512327B (en) | 2017-11-21 |
| Publication | Publication Date | Title |
|---|---|---|
| CN104512327B (en) | Blind area vehicle checking method, system, vehicle lane change method for early warning and system | |
| CN112349144A (en) | Monocular vision-based vehicle collision early warning method and system | |
| US8040227B2 (en) | Method for detecting moving objects in a blind spot region of a vehicle and blind spot detection device | |
| US8232872B2 (en) | Cross traffic collision alert system | |
| US8041079B2 (en) | Apparatus and method for detecting obstacle through stereovision | |
| US20040234136A1 (en) | System and method for vehicle detection and tracking | |
| CN107066986A (en) | A kind of lane line based on monocular vision and preceding object object detecting method | |
| CN108638999A (en) | A kind of collision early warning system and method for looking around input based on 360 degree | |
| JP3424808B2 (en) | Rear side monitoring device for vehicles | |
| CN102629326A (en) | Lane line detection method based on monocular vision | |
| US20060171562A1 (en) | Mobile body surrounding surveillance apparatus, mobile body surrounding surveillance method, control program and computer-readable recording medium | |
| KR101478258B1 (en) | Lane detection method and system | |
| CN102682455A (en) | Front vehicle detection method based on monocular vision | |
| CN107985189B (en) | Depth warning method for driver's lane change in high-speed driving environment | |
| CN108162858B (en) | Vehicle monitoring device and method thereof | |
| CN104517111A (en) | Lane line detection method and system, and lane deviation early warning method and system | |
| CN107229906A (en) | A kind of automobile overtaking's method for early warning based on units of variance model algorithm | |
| JP2006527427A (en) | Method and apparatus for locating and tracking an object from a vehicle | |
| CN102303563B (en) | Front vehicle collision early warning system and method | |
| CN104616502A (en) | License plate identification and positioning system based on combined type vehicle-road video network | |
| CN110991264A (en) | Front vehicle detection method and device | |
| CN115240170B (en) | A road pedestrian detection and tracking method and system based on event camera | |
| US20050189471A1 (en) | Mobile body surrounding surveillance apparatus, mobile body surrounding surveillance method, control program, and readable recording medium | |
| CN114663859A (en) | Sensitive and accurate complex road condition lane deviation real-time early warning system | |
| CN118991606B (en) | Blind spot collision warning method and system |
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |