Research On Vehicle Detection And Tracking Algorithm In The Road Video Surveillance System

In recent years, the improvement of people’s living standards, the number of vehicles is increasing, traffic congestion, frequent accidents and other phenomena are followed. In order to solve such transportation problem transportation, intelligent transportation technology arises at the historic moment. As an important technology in intelligent transportation systems(ITS), vehicle detection and tracking technology has become a research hotspot in recent years.This paper puts forward a kind of vehicle detection and tracking algorithm which meets the real-time and accuracy requirements of video monitoring system. The main work is as follows:1. This paper improves ViBe algorithm, which makes ViBe algorithm more robust for image mutation. By detecting foreground point number whether mutations, image mutations can judged. In case of image mutation, ViBe background model will be initialized, to ensure that the background model can quickly adapt to the image mutations. At the same time, this article introduces two update rates, a fast update rate used in the initialization, another update rate put forward by the author of ViBe. To do so can quickly eliminate Ghost area, at the same time, and will not affect the detection effect of ViBe algorithm.2. This paper proposes a vehicle detection algorithm which combines Background modeling algorithm and machine learning algorithm. First, by using the improved Vi Be algorithm and connected domain algorithm, the motion target area can be got. Then, by using Haar classifier to further identify the movement target area, we can confirm whether the movement target area has vehicle. Machine learning algorithms further identify motion target area, which can improve the accuracy of vehicle detection algorithm. At the same time, motion detection algorithm reduces the detection area of machine learning algorithm, which improves the real-time vehicle detection algorithm. The vehicle detection algorithm proposed in this paper, therefore, can also meet the requirements of real-time and accuracy.3. This paper proposes a vehicle detection algorithm which combines simple tracking algorithm and pyramid L-K track algorithm. We use simple tracking algorithm to get vehicle target initial position, and then complete vehicle target follow-up tracking by using pyramid L-K track algorithm. The pyramid L–K tracking algorithm is a high precision algorithm. It can satisfy the precision requirement of video monitoring system.The experimental results show that the detection and tracking algorithm is feasible and can satisfy the requirement of video monitoring system of real-time and accuracy. Therefore, it can be applied to the actual situation.