Research And Implementation Of Real Time Dehaze Algorithm For Vehicle Video

In the hazy weather,vehicles are prone to traffic accidents while driving due to the reduced visibility of the road.The real-time video dehazing system can recover and display the status of the road to help drivers to make their correct decision and avoid traffic accidents.At the same time,sine most of the vision-based intelligent assisted driving system assume that the input video is clear,the blurry video recorded in haze weather will affect the accuracy of the systems or even make them out of use.The real-time video defogging system can also be used as a front system for other intelligent assisted driving system to improve their robustness.Therefore,the research on real-time video dehazing algorithm has very important significance.In this paper,we first deduce the atmospheric scattering model detailly,which is commonly used in image dehaze from why haze can reduce image quality.Then,we introduce the image dehaze algorithm based on dark channel prior,and improve the algorithm because it has disadvantages such as inaccurate estimation of atmospheric light intensity,not suitable for bright areas and so on.Finally,gamma correction is used to improve the visual effect of the recovered image.In order to improve the speed of the algorithm,we estimate the atmospheric light intensity by the correlation of the video and filter the atmospheric light to reduce the flicker effect.After that,we use the fast guided filter base on gray image to refine transmission.In order to verify the validity of the algorithm,we implement the algorithm on Android platform,which is widely used in vehicle monitoring system,and test its performance and effect.The test results show that the video dehazing algorithm proposed in this paper can process haze video in real time on the resource-constrained Android platform,and the quality of the recovered video is quite good.