Research On Traffic Flow Measuring Algorithm By Video Detection

Intelligent Transportation System (ITS) has been considered as the most effective way to solve the urban traffic problems. Meanwhile, it is the main stream direction of urban traffic development. Acquiring the real-time information of traffic, such as traffic flow, vehicle speed, is the foundation of ITS. In this paper, we do research on the approach of traffic flow using video detection algorithms in ITS, such as moving vehicle detection, background extraction, shadow elimination and vehicle counting.The main contents include:1) Vehicle flow measuring algorithms are deeply studied. For vehicle detection algorithms, they include optical flow method, frame difference method and background method. For background extraction algorithms, they include means method, statistical mean value method, Single Gaussian Model and Gaussian Mixture Model. For shadow elimination algorithms, they include the method based on HSV color space, the method based on invariants of the color characteristics and the method based on texture features. At the same time, the paper analyzes the advantages and disadvantages of the above algorithms through experimental simulation.2) An improved Gaussian Mixture Model algorithm is proposed. When reading a number of images, background reaches a steady state. Reading a new frame, if the difference between the pixel and the stable background image’s pixel is greater than the threshold, then the current pixel is updated. Otherwise, it is unchanged.3) An improved vehicle counting algorithm based on the virtual local area is proposed. First, setting the detection area and the test line. Then, tracking the distance between the centroid in the detection and the test line. If the distance is less than Th2, one vehicle is considered to pass through the area.4) A Traffic Flow counting software system is implemented. The system includes video player module, background updating module, foreground building module and vehicle counting module. The function of video player module is to implement video playback and display. The function of background updating module is to implement the paper’s background updating algorithm. The function of foreground building module is to extract the foreground through shadow elimination and morphological operations. The function of vehicle counting module is to achieve the paper’s vehicle flow counting algorithm. The precision of the algorithm can reach 90%In this paper, vehicle flow measuring algorithms are deeply studied. The paper gives an improved Gaussian Mixture Model algorithm and an improved vehicle counting algorithm based on the virtual local area. A Traffic Flow counting software system is implemented. Experimental result shows that the system can realize the precise target detection and traffic flow measuring.