基于车辆轮廓凹陷区域的分割算法

Segmentation algorithm based on depressed area of vehicle contour

基于视频检测车流量,由于车辆间距过近容易造成图像上的粘帖而不能准确检测目标车辆,因此需要建立一个准确率高、抗干扰能力强的车辆分割算法。根据车辆轮廓凹陷区域提出一种分割算法。首先提取出二值化图像,在图像上检测并标记出目标遮挡区域,根据面积比来判断是否发生遮挡;然后提出了一种基于七宫格的凹陷区域检测算法,遍历每个区域找到符合条件域的凹陷区域,再通过共同特征来匹配到所需的遮挡区域,最后用Canny算法得到一条合适的分割曲线作为最后的分割结果。从实验结果分析,该分割方法能够适应复杂的环境,分多个遮挡区域的车辆,该算法精度高、鲁棒性好。

Vehicle flow detection based on video is difficult to detect the target vehicle accurately because the distance between vehicles is too close to each other,so it is necessary to establish a vehicle segmentation algorithm with high accuracy and strong anti-interference ability.A segmentation algorithm based on the depressed area of vehicle contour is proposed.Firstly,the binary image is extracted,and the target occlusion area is detected and marked on the image to judge whether occlusion occurs or not according to the area ratio.Then,an algorithm for detecting the depression area based on the sevenpalace grid is proposed,which traverses each region to find the depression area of the eligible region,matches the required occlusion area through common features,and finally obtains an appropriate occlusion area using the canny algorithm.The segmentation curve is used as the final segmentation result.The experimental results show that the segmentation method can adapt to complex environment and segment vehicles in multiple occlusion areas.The algorithm has high accuracy and good robustness.