Research And Implementation Of Traffic Video Detecting And Tracking System Based On Wavelet Analysis De-noising

There are more and more concerns on the application of computer technology in the field of traffic video monitoring, with its priority of convenience of monitoring, processing data in time and accurately, the administrators of transportation can acquire the monitored scene's traffic information parameters in real-time, and obtain the references. Therefore, traffic video detection and tracking is not only becoming the research focus of ITS (Intelligent Transportation System), but also has the significant application value. The research work is supported by the Ministry of Public Security of P.R.China (No. 20036255201).This dissertation, through the research and implementation of video capture in the system of traffic video detection and tracking, the image preprocess of the video frame, collection of the traffic information, has accomplished the work as follow:1) In video frame collection and capture aspects, the toolkit of DirectShow is used. Meanwhile, it is improved by the method of adjusting the system parameters, so that it can capture the video frame both in single and in sequence, and set the saved format of the captured image etc. It is beneficial to the following work which is detecting the moving vehicles.2) A method for de-noising based on the two level haar wavelet analysis has been proposed. During the collection and transformation of the video frame, the noises are often brought into the image which will have bad effect on the following processing. Wavelet-based de-noising method presented in this thesis can eliminate the effect obviously and make preparation for the following work.3) By detecting and tracking with the moving vehicle, the speed, auto size, and the amount of the traffic flow can be acquired. In the speed calibration aspect, a method using combination of the Kalman filter and the extraction of the region barycenter has been proposed in the thesis. With the pre-estimate characteristic of the Kalman filter, it can shorten the time to acquire the speed, and improve the ability of the system in real-time.4) A system of traffic video detecting and tracking has been designed and implemented on the research above. It has accomplished the function of detecting and tracking on moving vehicles automatically, acquire the traffic parameters information in real-time, such as: speed, auto size, and the amount of the traffic flow.