Video-based Vehicle Detection And Recognition

Traffic is the life vein of national economy, and ITS (Intelligent Transport System) focuses on those information and intelligent techniques which can manage modern traffic and provide services to all the users. Recent years many people concentrate on ITS because of the fast development in traffic. This paper is the main part of the research project named 《Video-based Vehicle Detection and Recognition 》 funded by China National Electronic Information Development Foundation. I apply digital image system to detect the speed and position of motive vehicles, it's very important and potential in ITS.Chapter 1: Discuss the connotation and development of Intelligent Traffic. Why we need a new technique to detect vehicles automatically? The advantages and disadvantages of several ultra-fast detection techniques are compared. At last, I describe my work, new techniques and creative ideas.Chapter 2: Discuss the configuration of my system. Introduce the front-end units, lens, CCD sensors and image acquisition card are introduced in detail.Chapter 3: Introduce the automatic control in image acquisition process, analyze the factors that can affect the image quality and give solutions to these problems. After analyzing why the focal length changes, I give an automatic focus algorithm controlled by computer. But that's not all, I also give an automatic intensity modulation algorithm to acquire good image luminance under different circumstances.Chapter 4: This chapter discusses some preprocessing algorithms such as gray level transformation of image, noise canceling and edge enhancement before object detection.Chapter 5: This part discusses how to detect motive vehicles, especially motive detections base on multi-image sequence. Compare testing algorithms base on frame difference to those base on deductible background. At last, I give an image registration approach to solve the problem caused by camera's shaking during testing process.Chapter 6: Discuss tracing and speed detection of vehicles. Apply tracking gate to address vehicle tracing problem in video image sequence. At the same time I provide an easy and fast tracing approach to identify the relation between sequential frame difference image and vehicle moving track. At last I give the summary of this paper and possible research area in the future .