Visibility detection based on traffic camera imagery
Ming-Wei AN, Guo Zong-liang, Jibin Li, Tao Zhou · 2010
To make full use of traffic video along the highway, to achieve low-visibility weather monitoring, a meteorological visibility detection method is proposed on the traffic video. Fog causes the disappearance of edges in the frame, which is well reflected by the change in high frequency information. In order to extract out the edge feature that reflects the changes in visibility of the image feature parameters, the algorithm uses least squares method and the inverse transform to draw the fitting function, and analyzes the relations of the image characteristics and the distance to determine the visibility values. Algorithm satisfies the human visual system, and effectively improves the accuracy of visibility detection. Video visibility detection system has good real-time, high detection accuracy, a small error, and better robustness. System has function of realtime statistics and analysis.