基于视频图像的隧道行车速度检测

Tunnel Vehicle Speed Detection Based on Video Image

基于视频图像测量隧道行车的速度,重点是检测车辆在视频图像序列中的位移,现有的位移测量方法大多采用特征匹配法和车牌定位法,两种方法都是提取车辆运动的特征向量来判断车辆的总体位移,有一定的局限性。本文考虑到傅里叶变换具有全局特性和相移特性,提出了一种将图像虚拟检测线灰度均值测量、傅里叶变换、相位相关和数字图像处理相结合的方法,来检测运动车辆及其在视频图像序列中的位移,再将位移量应用于摄像机中标定,计算隧道行车的速度,以提高车速检测的准确性和实时性。

For the speed detection of tunnel driving based on the video image, the most important is to measure the displacement in the video image sequence. At present time, for the approach of the displacement measure- ment, there are two methods called feature matching and license plate location to verify the displacement of the vehicle from the feature vector, which might be boundedness relatively. Many factors were taken into considera- tion including the global and the phase-shift feature of Fourier transform, a method was proposed to detect the vehicle inside the image sequence and to measure its displacement by combining the measurement for the mean value of the gray scale based on the virtual detection line, the Fourier transform, the phase correlation and the digital image processing, beside this, and then these result were applied to the parameter of the camera calibra- tion which was settled to calculate the speed of the vehicle that make it function as the optimization and real-time of the speed detection.