Research on High-Definition Video Detection in ITS
|School||North China University of|
|Course||Detection Technology and Automation|
|Keywords||High-Definition Video Brightness Curve Vehicle Location Occlusion Detection Mean Shift algorithm|
With the development of computer technology and digital image processing technology, intelligent traffic control system based on video is developed fast, and it becomes the mainstream of traffic control. In the first place, this paper introduced the research background of this issue and the present research status of ITS, analyzed the deficiencies existed in modern video surveillance technology, and put forward a new detection method based on HD videos.According to the technical difficulties of HD videos, common detection algorithm used in standard videos can’t adopt to HD videos. HD videos token by HD camera are seen as the research object in this paper, and then we do deep research on vehicle detection based on HD video based on the existing algorithm, put forward the corresponding algorithm and do related applications.This paper’s main work is as follows:(1) Analyze the advantages and disadvantages of commonly used vehicle detection technology, and the put forward a new algorithm of background extraction and update based on HD video. Because of the exiting of static status when the light is red, this paper adds the judgment of red light, do background update according to the state of light.(2) When doing the algorithm of vehicle location, this article using brightness curve to detect the existence of vehicles. Because two-dimensional image is changed to one-dimensional brightness curve in this paper, corresponding processing time is reduces sharply, and this algorithm can meet the requirement of real-time when using in HD video. Then according to the characteristics of brightness curve, we can do occlusion detection and division according to the window location. Then set virtual windows in every lane and do traffic statistics.(3) Analyze the characteristics of kalman filter and the algorithm of Mean Shift, this paper uses the algorithm of Mean Shift based on the forecast of kalman in vehicle tracking, which is used in violation detection.(4) In this paper, violation detection is analyzed, the red light running and the retrograde algorithm is included. In the algorithm of red light running, this paper set virtual windows to judge whether it is running red light. In the algorithm of retrograde, we first judge the traffic flow, and then do comparison with every vehicle’s flow. The algorithins used in this paper are with little computation, which can effectively deal with large amounts of data, it also can effectively reduce the noises, and it can be applied in HD video monitoring.