Research On Traffic Flow Statistics Based On Computer Vision In The Intelligent Transportation System

Establishing a comprehensive and efficient intelligent transportation system, and the effective coordination and positive interaction between operators and the traffic control department, can effectively control the growing city and intercity traffic. With the rapid increase of the number of city vehicles, traffic congestion has become a difficult problem in the field of urban transport. Especially in recent years, the urban planning lack of enough consideration, ignoring the importance of the construction of urban infrastructure, and outdated road traffic management technology, lead to road congestion and traffic accidents frequently, and make the traffic congestion become one of the most difficult problems to be resolved in the large or medium-sized cities in China. Road congestion, traffic congestion, traveling difficult, driving difficult and walking difficult affect people’s travel and city development, and seriously restricted the development of the city. How to achieve efficient traffic scheduling, guide the standardized traffic behavior, and reduce the frequency of traffic accidents is imminent. We urgently need to solve the problems.Traffic flow statistics play an important role in traffic management.Video monitoring system as a subsystem of the intelligent transportation system(ITS), plays an important role in the intelligent traffic management. The traffic flow statistics method based on computer vision is a very important part in the intelligent transportation system.It processes the video captured by a video monitor for traffic flow using a computer, obtains the number of vehicles in a certain period of time on the road, and provids the basic data for subsequent processing in the intelligent transportation system. It can dispatch road scientifically, so as to reduce the occurrence of traffic accidents and congestion and travel conveniently.A traffic flow statistics method based on the target area is proposed in this paper. The method has many advantages. First, the use of artificial line, can be more appropriate and accurate to get the target area. Second, the size of the video frames to be processed is less affected. We just get video in each frame of a piece of sub region for processing and access to the size of the sub regions with strong flexibility, not on the entire video frame processing. Therefore, the size of the video to be processed has little effect. Third, it speeds up the running speed of the system, and it is more real-time. Because we deal with the video in each frame of a sub region, no matter which method we use based on characteristic of accurate object recognition technology or based on background modeling of fuzzy object identification technology, we speed up the speed of object recognition with respect to each whole frame of video for processing. Fourth, we simplify the tracking process. The general process of tracking need long time, and tracking algorithm itself is time-consuming. Therefor, the tracking process is relatively time consuming. Traffic statistical method based on the target area, only needs to associate objects in the consecutive frames under specific circumstances, which not only reduces the tracking process, diminishes the error risk, and improves the system speed, but also simplifies the process of tracking, and makes the tracking algorithm more flexible. The method of the traffic flow statistics based on the target area, is important to have the following points: the acquisition of the target sub area, object recognition technology based on mixed background modeling, counting and simplified tracking process. The experimental results show that the traffic flow statistics method based on the target area has higher real-time performance and higher accuracy under the circumstances of non congestion.