Research On Road Traffic Status Detection Based On Video Analysis

With the rapid economic development and continuous expansion of city scale,the amount of urban population and road vehicles are increasing rapidly,which lead to some problems such as traffic accidents,traffic congestion more and more serious,Furthermore,environmental pollution and waste of resources have cropped up,these problems badly restrict urban development and affect our living quality.Therefore,road traffic state detection becomes one of the important content of research in the traffic video monitoring.A road traffic states detection method based on video analysis is proposed in this paper,through the analysis of road monitoring video,it focuses on running vehicles extraction,traffic feature parameters extraction and traffic state identification.The main work has the following several aspects:(1)The Gaussian mixture model is improved,which enhanced the adaptability of background model to changes in light and moving vehicle is extracted,meanwhile the background difference method is used to obtain the vehicle movement area;(2)A road traffic density detection method based on edge feature is proposed in this paper,which avoids the segmentation of the vehicle area and reducing the difficulty of parameter extraction;(3)Road traffic state is seen as a macroscopic description of road traffic flow,block matching method is used for estimating the speed of traffic flow.On this basis,K-means clustering algorithm is applied to two layers of classifier design,three kinds of traffic state(smooth,slow,and congested)will be classified and recognition.(4)The suggested algorithm is integrated system,and the Guiyang actual road traffic surveillance video has carried on the experiment,the experimental results showed the proposed algorithm is effective and feasible,which lays a good foundation for the subsequent research of traffic state recognition.