The system described in this page is a Computer Vision Based Traking System that monitors the traffic of vehicules. It can classify the vehicules and extract information about the speed of those vehicules. Such as police speed radar it can extract the speed of multiple vehicles at the same time but it’s not accurate enough to be used that way. Tracking and recognition methods in video are implemented with Intel’s Open source Computer Vision (OpenCV) Library helped by the cvBlobsLib Library. In this project, we use a perspective warping transformation, then background subtraction and blob detection. Indeed, we split our application into three procedures. First, the transformation into a bird’s eye view which allows us to extract information about the vehicules more easily. Second, automatic background extraction procedure, in which the background is extracted from the successive frames. Then, the vehicules detection procedure based on the blobs area.

The Team

The team was composed by:

And supervised by Zheng-Hua Tan.


Go back to the home page