Research On Methods For Road Traffic Flow Parameters’ Automatic Detection Based On Video Surveillance

The road traffic flow parameters automatic detection technology is used toexamine the target on the road by sensing equipment, and is a kind of technology toacquire relevant traffic parameters automatically, and it is the basement of road trafficmanagement intelligent for examine and deal with road traffic condition and informalaffairs automatically. Compared with the traditional methods of vehicle flow detection,the vehicle flow detection based on video is informative, low-cost and easily-controlled. So the traffic automatic measurement technology based on the video is animportant direction for the ITS in the future and is of great value in research andapplication. This paper emphasizes some key questions in the road traffic flowparameters automatic detection based on the video surveillance. The main researchcontents are as follows:(1) The paper proposed algorithm for based on the region growing method problemsbrought by the complex scene in which there is a large scale of the similar color andfeatures and fault localization of the previous algorithm based on the color ormodeling. This new algorithm brought recently gets the general position of the laneby means of background differencing firstly, then it acquires the region growing bytaking the vehicles’ coordinate points in the frames as the growing points, finally itgets the accurate road position.(2) The paper proposed a new lane line detection algorithm based on boundarydetection. Firstly it does the boundary detection of the frames using the Canny edgedetector, then it detects all the minimum tangent rectangles using the algorithm whichextracts the minimum tangent rectangle based on vertex chain code. Finally, it screensthe set of the detected minimum tangent rectangles by analyzing the aspect ratio andthe area of the lane lines in reality and gets the lane lines’ location. The experimentsshowed that this method can accurately locate the lane lines.(3)In the setting of the virtual loops, most of the existed detection technologiesdepend on the virtual loops setting. To solve the problem, the paper proposed amethod which sets the loops automatically based on the lane lines localization. It firstly gets a set of horizontal rectangles according to relationships of the tangentrectangles’ vertexes, then calculates the loops’ width and height using the lane lines’aspect ratio. The result of the calculation is used as the reference of the vehicles’running distance.(4) The paper proposed an algorithm of the speed measurement based on the graymatching. It firstly takes a base of the virtual loop as a benchmark, counts the graychanges when vehicles pass, and records the scale of the gray changes using thehistogram. Then do the same to the opposite side of the loop so as to solve theproblem of checking whether the recorded vehicle is the same one or not. And theexperimental data shows that this method has a high accuracy.(5) Accomplish the design of an automatic traffic flow parameters detection systembased on the video surveillance according to the research achievements.