An Optimized Video-based Traffic Congestion Monitoring System

Traffic congestion monitoring is being more and important with the enlarging of urban scale and increasing number of vehicles. Video-based traffic congestion monitoring systems are widely used now but it relies on the high performance of hardware. We analyze the procedure of video-based traffic congestion system and divide it into graying, binarization, denosing and moving target detection. The system first reads real time monitoring video and converts them into grayscale images. Through experiment, we find simple global threshold is the most cost efficient for monitoring system. Then we perform noise reduction with different algorithms and find the fittest one for the system. We also put forward a background subtraction method with noise reduction for post-treatment to identify the moving objects, improving identification rate. The system determines whether the congestion occurs by comparison result of the total movement and predefined threshold. We propose an integrated optimized solution for traffic congestion monitoring by making a tradeoff between cost and effect, which uses simple global threshold for binarization, does denoising with IGF (sigma=0.05) and conducts background subtraction with median filter (3x1). The solution achieves real-time response and improves performance without adding more computation.