Research On The Dectection Of Vehicles And Road Surface Condition In Highway Video Surveillance System

As the main line of urban transportation, highway has become the backbone of economic development. The safety levels of highway road surface condition are greatly influenced by natural factors. Rain, fog and snow would affect the operating condition of highway.The purpose of the proposed paper is to realize the highway intelligent video surveillance system application. It mainly focuses on the moving object recognition and tracking in video surveillance. Meanwhile, the paper does research on the road surface condition and then develops the system. The main work and contributions include:1) The paper proposes a new target shadow elimination method. In order to obtain the moving vehicle target accurately, shadow elimination must be used. While dealing with shadow elimination, for the shortcomings of the traditional method based on HSV color space, the paper proposes an efficient method of shadow elimination based on HSV color space and image region information. It can avoid the regions whose color features are close to shadow inside the target object to be mistakenly eliminated.2) The paper proposes a tracking strategy which could apply to the situation of multi-target occlusion. For vehicle target tracking, it uses the tracking method based on target geometry feature and Kalman predictor. While dealing with the situation of multi-target occlusion, the paper proposes a tracking strategy based on centroid information and area information, and does discussions according the number of the predicted target and the actual target.3) For the road surface condition detection problems under rainy and snowy weather, the paper uses native Bayesian classifier to recognize road condition. First, it extracts the road region of the image. Then, it extracts the road image texture and color features, and chooses the entropy, energy, contrast, average brightness and saturation as features. At last, it uses the naive bayes classifier to finish road state classification, and will do alarm for the slippery or snowy road.4) Highway intelligent video surveillance system is designed. The system has the functions of road background extraction, moving vehicle detection, moving vehicle tracking, vehicle information statistics and road state classification. Besides, it can do alarm for abnormal traffic situations and bad road conditions.