Vehicle Detection Based On Video Image Processing

With the rapid development of society and economy, people's living standard has got great improvement and the number of vehicles has quickly increased. Most countries in the world have devoted a large amount of manpower and material resources to research Intelligent Transportation Systems (ITS). Vehicle detection is the foundational part of ITS and provides basic data for the system. Video-based vehicle detection has become an important method of vehicle detection for its peculiar strongpoint.Through analyzing existing methods of detection, this thesis adopts background subtraction to detect vehicle from image, and this method include background building, background updating and compensation of background. Considering the impact of vehicle shadow upon detecting, this thesis diminishes the influence degree by using inherent characteristic of vehicle shadow. When the color of vehicle is similar with the color of background, the pixels of vehicle are easily divided into background. Aiming at that problem, the thesis puts forward an algorithm which combines threshold segmentation with edge detection to get integrated vehicles object from image and make preparations for vehicle located.Based on detecting vehicle successfully, feature matching method is used to track vehicle in order to gain the traffic parameters such as number of flow and vehicles speed. Object widow is one of primary features during selecting vehicle features. When several vehicles appear in real scene, it is difficult to fix every object window exactly with projective method. In this thesis, firstly use perspective transformation to adjust the original image to be a demarcated image. Then do horizontal and vertical projection in demarcated image to fix the object window exactly. After chose feature of vehicles, introduce feature matching matrix to track vehicle objects.The experimental results show that the method combining threshold segmentation with edge detection can get vehicles object integrally. Through demarcating the image first and then projecting can overcome the influence leaded by vehicle location and fix the object window correctly. After that, gain the traffic parameters by object tracking. The algorithm above has small amount of count, good real time capability and high detecting precision. Using above operation can gain the traffic parameters and provide guarantee for application of ITS.