Research on the Method for Fog Detection and Removal Based on Traffic Video
Song Hong · Control Engineering of China · 2013
Fog causes traffic video images blurred and is not conducive to the extraction of various traffic parameters. The paper proposes a traffic-video based foggy weather identification method and a fast restoration approach for traffic image blurred by foggy. Firstly,the current background is generated from the real-time traffic video. Difference image is obtained by subtracting the current background image from the sunny background image which is generated in advance. And then,texture analysis is executed on the difference image to recognize the current weather. Finally,if it is foggy,the current traffic video images are recovered by the simplified dark channel prior principle. To verify the effectiveness of our algorithm,traffic videos about the same traffic scene,at the same position and within different weather conditions are collected and stitched together randomly. Experimental results show that the proposed algorithm not only can accurately identify the current weather and improve traffic image quality when the current weather is fog,but also decrease time complexity when comparing with various other algorithms and satisfy real-time requirement.