Summary: | 碩士 === 國立中央大學 === 資訊工程研究所 === 91 === With the advancement of modernized society, the relationship between social security and computer technologies is getting closer and closer. Computer technologies can provide more comfortable and secure environment and improve the living standard of human beings. Tracking and monitoring systems gradually become an indispensable part in current society, which can be realized by utilizing image processing and pattern recognition techniques. They constitute the essential part of a video surveillance system.
In this thesis, an intelligent traffic monitoring system which can track vehicles in multi-lanes highway. The proposed system can also extract many important traffic parameters to reflect the real time traffic situation in highways. In the proposed system, the task of image preprocessing is first performed to construct reliable image background and eliminate shadows caused by the vehicles. Then, the mechanism of active contour is devised to describe the shape of vehicles. Last, the vehicle tracking model is built by utilizing the linear prediction model of dynamic objects.
Experiments were conducted on several image sequences captured in highways. Experimental results verify the validity of the proposed approach in successfully tracking vehicles.
|