Research On Visibility Detection And Polarization Restoration Based On Pitch Angle Compensation

With the popularization of vehicles,more and more traffic accidents,traffic safety problem is increasingly concerned by people.In recent years,the air pollution is becoming more and more serious,and the frequency of bad weather is increasing year by year.In particular,foggy days lead to low visibility on the road surface,resulting in frequent major traffic accidents.It is of great significance for traffic safety to realize real-time detection of road visibility in foggy environment and obtain clear visibility images.In this paper,the visibility detection and image restoration methods based on dark channel prior are deeply studied.Firstly,the atmospheric scattering model is established,and the influencing factors of atmospheric visibility and the characteristics of fog images under different concentrations are analyzed.The detection method of atmospheric visibility is described and the advantages of image information obtained by polarization acquisition and imaging system are analyzed.It lays a theoretical foundation for the follow-up work of atmospheric visibility detection and target image restoration.Secondly,A pitch Angle compensation visibility detection method based on dark channel prior is proposed.In this method,an atmospheric scattering model is established first.After the image is preprocessed,the atmospheric transmittance in the image is obtained by using the dark channel theory,and the atmospheric transmittance is refined by guiding filtering method.Then the improved Canny edge operator is used for edge detection,and the Hough algorithm is used to extract the lane lines in the image after edge detection.Finally,the pitch Angle compensation model is established to reduce the solving error of the distance between the camera and the object in the model,and then the extinction coefficient of the image is obtained,and the visibility of the current detection image is derived.Thirdly,an object image restoration method based on polarization information and dark channel prior method is proposed.Firstly,the polarization degree and polarization Angle of the multi-angle polarization image are obtained by Stocks decomposition.Then,the improved median filtering method is used to denoise the polarization Angle image,and the atmospheric light intensity is obtained.Then,the total atmospheric transmittance is calculated and refined by multistage median bilateral filtering.The brightness of pixel points in the adjacent nine neighborhoods is selected to calculate the mean value,and the maximum value of the mean value is found as the atmospheric light value at infinity.Finally,the restored image of the object can be obtained by atmospheric scattering model inversion.The experimental results show that the visibility detection accuracy and image restoration effect in this paper are greatly improved compared with the traditional methods.