Research Of Traffic Congestion Identifi Cation Based On Multi-feature Fusion

The intelligent detection of traffic flow state plays an important role in solving the traffic jam. Its realization will not only greatly save human resources, but also maximum utilization of the limited traffic resource and increase the utilization rate of road resources. So we turn to research intelligent decision of traffic congestion from the perspective of computer vision.In this paper, the corresponding algorithm is designed to intelligently extract the road area from the video, which reduces the influence of the non-road region on the determination results. A new background reconstruction algorithm based on frame differencing and mode is proposed to better use background difference method to extract moving object design. In order to update the background in real time, the different methods are used to update the foreground and background.This article establishes a traffic flow criterion which is based on extracting the feature and dividing the states of the traffic flow. A new method based on multi-feature fusion of traffic congestion is proposed to judge the congestion level of the current road.The congestion level of the current road is judged by combined the regional density, the time occupancy and the average speed of vehicle flow. In view of space and time dimension, the new model is more suitable to the regularity of traffic flow. The simulation shows that this method has better accuracy and objectivity.