Research And Implementation, Video-based Vehicle Detection Prohibited Isobar

With the rapid development of society and economy and the enlargement of city scale, city highway traffic system has become more and more complex. In recent years, traffic increase has speeded up much faster than that of highway and other transportation facilities,which result in traffic jam and accident, environment pollution and energy waste more severe. By means of automatic supervisory to traffic can decrease traffic accidents efficently. Therefore, it has been an important task to design video supervisory control system for the transportation prohibition area by making use of computer technology, multimedia technology and communication comprehensively.Through the deep analysis of Intelligence Transportation System, this thesis introduces domestic status of ITS technology. After objective comparison and analysis of the existing video supervisory control system, the key technologies such as image processing and recognition will be improved. Local match algorithm based on image masking will be proposed so as to realize detection of incumbency of the yellow line of the vehicle, and the system of vehicle pressing yellow line detection will be developed.The thesis contains several tasks as following:(1) We propose image segmentation method based on example images.According to the feature of transportation path yellow line, we adopt segmentation method based on wavelet transformation.We use first and second order derivatives of Guassian function as wavelet base,and use Mallet algorithm to decompose image. Mar operator is adopted to detect image boundary multi scalely,.We also design a simple and efficient method to initialize yello line location manually,which can get rid of background,reduce detect area.(2) According to pressing line recognition of the single frame image, we propose local match algorithm based on image masking, taking one frame image of non-vehicles line pressing as template. It compares timely snapshot image with template yellow line region, to judge whether the vehicles press the line or not. The result indicates that this algorithm can obtain a better recognition result and quicker speed. (3) We propose renewing method of the dynamic template in order to avoiding misjudge caused by other factors. It can deal with illumination, fallen leaf, pedestrian, weather and so on better, so it improves the testing accuracy and can work in full day.(4) We use object-oriented method to design and develop a prototye system based on video. This prototype system integrates some modules including video collection,detection of vehicle pressing line,data manage,data transmission and so on, and it has been put into use. When vehicles press the center yellow line,the system can make real-time recognition.