Research On Video Vehicle Detection And Track Arithmetic Based On HSV Color Space

Intelligent transportation system (ITS) is a system for t applying computer vision technology, image processing technology, communication technology in traffic domain. The aim of ITS is to make full use of available road facilities, improve reciprocity of vehicle, road and people, and then enhance the economical efficiency of the whole system. Moving vehicle detection without human intervene have become a problem attracting more attention in the ITS. Aimed at detection and track of moving vehicle in the traffic control system, a series of exploratory researches done in this paper are as follows:1,This paper proposes background abstraction and vehicle detection based on HSV color space. it is a sort of comparatively effective and novel method compared with most background abstraction and vehicles detection based on RGB color space at present. In the background abstraction, it uses the character of background image element point basically according with Gauss Distribution, and sets up mixed Gauss Distribution for HSV weight of every image element. For the judge of background and foreground, take full advantage of color information in order to reach optimal purpose of background abstraction. In the background renovation arithmetic, it considers the matter of learning rate and effectively uses image element of former background, so the arithmetic effectively avoids contingency factor to infect background and enhances stability.2,After vehicle detection, this paper proposes a improved shadow detection arithmetic to process shadow. It introduces first order annihilability arithmetic based on original HSV color space detection arithmetic. Then the new arithmetic enhances stability and validity about original arithmetic. Simultaneously, the arithmetic based on HSV color space is simple, frequent and effective compared whit RGB color space. The color space is consistent with background abstraction and vehicle detection, so it enhances the efficiency of the whole arithmetic.3,In the vehicle track, firstly, this paper counts color histogram of H about chosen ROI area, and enlarges it to color probability distributing of the whole original image, and then uses the mean shift arithmetic to account center of the tracked object. Then it makes the center coordinate and window size as eigenvector of the disperse Kalman filter. Then use the disperse Kalman filter to achieve track of moving object. The experiment indicates the arithmetic of this paper has good acquit about veracity and stability of the background abstraction, vehicle detection and shadow detection. It can be used in the detection and track of the complex scene and many moving vehicles.