A New Algorithm of Traffic Monitoring Based on Real-Time Video

During the past 10 years, different approaches of traffic monitoring have been reported. Most of these methods either rely on a large knowledge-base or are subject to many restrictions. That means they are time-consuming and difficult be carried out in real time. This paper tries to present a new algorithm to solve this problem. We divide the process into two main procedures. Firstly, an improved method is put forward to perform background subtraction, which segments slow moving objects from an illumination-invariant background. Secondly, object tracking is performed with an improved particle filter, which can avoid the difficult problem of matching. The proposed approach has been tested on several highways and satisfactory results have been obtained.