Research And Hardware Implementation Of Image And Video Defogging Algorithm

With the wide application of outdoor vision acquisition system in outdoor production and life such as video photography,traffic monitoring and production detection,we can obtain more and more high-resolution images and videos.According to these images and videos,we can extract a lot of useful information which brings great convenience to our production,life and the information management of city.However,the outdoor vision acquisition system is susceptible to the interference of fog,haze,rain,snow and other bad weather,resulting in the final acquired images and videos with low contrast,color distortion,texture loss and other defects,which will interfere with subsequent image processing and useful information extraction.Therefore,it is of great scientific value and practical significance to study how to effectively reduce the negative effects caused by the bad weather mainly caused by fog and haze,and to realize the rapid and efficient defogging of degraded images and videos.In this dissertation,we studied the key technology of real-time defogging system,and improved the single image defogging algorithm based on the dark channel piror(DCP)theory.besides,a video defogging algorithm based on the pyramid model was proposed,and a set of hd video defogging system was designed and realized.The main contents of this study are as follows:(1)Aiming at the single image defogging algorithm based on the DCP theory and the partially improved defogging algorithm,there are some deficiencies in the sky region color distortion,overestimation of transmittance,texture loss and so on,this dissertation used the OSTU algorithm to effectively segment the sky region of the degraded image to obtain the accurate estimation of the atmospheric light in the non-sky region,so as to reduce the color distortion in the sky region of the defog-free image;And the method of obtaining coarse transmittance is improved by taking the confidence of dark channel and the pseudo-fog removal image as correction factors;And the texture refinement of coarse transmittance was carried out by using jointbilateral filtering to reduce the texture loss of defogging image.The experimental results show that the algorithm has less color distortion,better texture,and defogging effect.(2)Aiming at the timeliness of video defogging algorithm.In this dissertation,the method of sampling under the pyramid is adopted,and the dark channel confidence degree and pseudo-fog removal image are used as correction factors to obtain the coarse transmittance of the sampled degraded layer at 1/16,and then it is refined layer by layer by means of joint bilateral filtering.This method improved the processing speed of reduced quality layers.And Inter-frame modulation light is introduced to solve inter-frame flicker.The experimental results show that the algorithm can meet the requirements of real time defogging of fog video with resolution of 1280×720 and frame rate of 30 fps.(3)The video defogging algorithm is hardware adapted,and a real-time defogging system for hd video is realized.The experimental results show that the system can meet the requirements of real time defogging of fog video with resolution of 1920×1080 and frame rate of 60 fps,and the defogging effect is better and the color is not distorted.