基于计算机视觉的实时车辆运动速度检测算法研究  

Study on real-time vehicle motion-speed detection algorithm based on computer vision

作　　者：谭呈祥[1] 

机构地区：[1]广西民族师范学院数学与计算机科学系,广西崇左532200

出　　处：《现代电子技术》2016年第9期164-166,170,共4页Modern Electronics Technique

摘　　要：在对比了传统运动车辆实时检测方法的基础上,提出了一种新的G-C二重差分法,将视频流中的某一当前帧与背景帧进行边缘检测,将两帧的结果相减,得到梯度差分,再将此结果与背景颜色差分结果进行"或"运算,提高了车辆存在判断的准确度,此算法可以准确检测出路面运动车辆存在同时有效消除车辆阴影的影响,为后续车辆运动速度的实时检测提供了有力保证。基于运动车辆检测的结果,在VC环境下进行了编程实现,自动检测出运动车辆的实时速度,测试结果表明该算法效果很好。On the basis of the comparison of the traditional moving vehicle real-time detection methods,a new G-C double difference method is proposed. The edge detection of a certain current frame and background frame in video stream is conducted,and then the results are subtracted from the two frames to get the gradient difference. After that the ″or″ operation for the difference result and background color difference result is performed to greatly improve the accuracy of the vehicle judgment. This algorithm can accurately detect the existing moving vehicle on the road and effectively eliminate the effect of vehicle shadow,which provides a strong guarantee for the real-time detection of the following vehicle moving velocity. Based on the results of moving vehicle detection,the programming realization was conducted in VC environment to automatically detect the real-time speed of the moving vehicle. The test results show that the effect of the algorithm is perfect.

关 键 词：计算机视觉 车辆检测 速度检测 消除车辆阴影 

分 类 号：TN911.73[电子电信—通信与信息系统]