Research Of Video-Based Vehicle Detection Algorithms Through Background Subtraction

The issue of vehicle detection is the basic and key problem in Intelligent Traffic Systems technology. Compared with traditional vehicle detectors, video sensor has lots of advantages, i.e., easy installation and maintenance, wide monitoring areas, obtaining more kinds of traffic parameters and etc, so it has been widely used in Intelligent Traffic Systems (ITS).The background subtraction is an important approach in video-based vehicle detection. The approach segments moving vehicles in the camera's field-of-view through the difference between a reference frame, called referenced background image, and the current input frame. The key of background subtraction is to extract and update the background image adaptively in changeable road environment affected by all kinds of factors such as weather and illumination condition.This thesis mainly discusses video-based vehicle detection algorithms through background subtraction. Two background extraction and updating algorithms based on different Gaussian Mixture Model (GMM) are proposed. They are also applied in daytime vehicle detection and traffic parameters calculation under the virtual loops based framework. In addition, a video-based vehicle detection system which can work in day and night usually...