Low-Quality Video Enhancement Method Research For Intelligent Traffic Monitoring

Traffic video surveillance system is the window to understand the traffic situation and the law order situation, and it is an important part of the public security traffic control system. However, traffic monitoring system is greatly affected by the bad weather and other environmental factors. In the foggy weather, the image quality will decline greatly because of the low visibility. It will then affected the function of the system largely. In the condition of dust, it is difficult to distinguish the outline of the object and color information, and it is concerned that the contours of the object in the monitoring and color details of fixed scene. Using infrared cameras can only get gray videos, however, it is unable to meet the demand for color and detail information.In this paper, we design and realize a low-quality video enhancement algorithm for intelligent traffic monitoring system in the fog and dust weather using ordinary camera through color estimation, deblurring, color constancy and edge enhancement. Improvements of fog removal based on atmospheric physical model for fog video, color estimate and edge enhancement processing algorithm for dust video. A large number of experiments in different scenes show that users can directly put fog or dust weather video into our system, and the normal clear videos can be obtained finally. The proposed system can be applied to vehicle detection and human detection applications for high accuracy detection rate.