Abstract: In order to study vehicle traffic flow under extreme weather conditions such as haze and so on, this paper mainly focus on the application of related video processing technology in vehicle detection, and studies the traffic flow detection method based on video images. Based on the Gaussian mixture model, background detection and video de-haze algorithm is used to model the background, video frame is extracted and HSV model is used to make the video image clearer and then traffic flow is detected. MATLAB simulation results show that the proposed flow detection system can produce high detection accuracy of the vehicle in the haze environment.

Keywords: Gauss model, HSV model, de-haze algorithm, vehicle detection and tracking.