摘要
The bad weather along the rail track has brought many inconveniences for driving, among which the snow weather in winter seriously threatens the safety of driving. However, existing image-based methods are only focus on snow removal, this paper emphasize on snow detection after restore scene. The imaging model of snowflake in video is built, which takes snowflake individuals in video images along the rail track as the research object. The improved guided filtering algorithm is applied to restore the background image, and then the enhanced background subtraction method is employed to extract the snowflake foreground, so as to realize the analysis and judgment of the snow status, which has an attractive detection effect. The comparison experiment for snow detection is carried out to demonstrate the efficiency of the proposed method, which achieves significant improvements over the state-of-the-art methods. Finally, a snow condition judgment standard based on video analysis is established, which matches different snow grades and video feature information. The operation and maintenance department can carry out the train speed limit management plan according to the standard, so as to ensure the safety of track operation in snowy days.