Research On Video-based Traffic Information Collection System

The Intelligent Transportation system (ITS) is the only way to achieve sustainable traffic development. Traffic information collection and processing technique, which is a key technique to realize information collection, real-time analyzing and processing, is one of the cores of the ITS.Video-based traffic information collection technique has demonstrated incomparable advantages during the development of traffic information collection techniques, and surely will lead the future trend of real-time traffic information collection technique.This paper adopted a traffic information collection technique which was based on moving object extraction and model following, identified and tracked the vehicle as a whole as the object, extracted the information of traffic flow, vehicle type, vehicle speed, vehicle motion trajectory and etc.. The study this paper conducted is as follows:After the writer's study and analysis of many vehicle detection algorithms, this paper adopted the background-subtraction-based video segmentation method to conduct vehicle segmentation, and achieved comparatively satisfying results by introducing supplementary means to help with the segmentation results judgment, while applying Otsu's method to obtain the adaptive thresholds.This paper, on the basis of obtaining the characteristics of a moving object, adopted the vehicle tracking algorithm of Kalman filter, accomplished the object tracking by firstly predicting the location of the object in the next moment, then conducting a object feature matching search in an small area around the predicted location, and finally establishing the corresponding relationship between the object and its counterparts in consecutive frames.In the aspect of traffic parameter extraction, this paper studied the methods of extracting and computing important traffic parameters such as traffic flow, vehicle type, vehicle speed and etc..