Research On Dehazing And Deblocking Methods In Foggy Image Transmission System

Due to the influence of haze weather,the images collected by outdoor imaging equipment have the characteristics of low contrast,dim color,and illegible detail information,which cause the image quality of receiver end to be poor and seriously affect the subsequent processing of images.Due to the limitation of channel bandwidth,the images collected by outdoor imaging equipment need to be compressed before the long distance transmission.However,blocking artifacts in images leads to degradation of images at receiver end.In order to obtain a clear dehazed reconstruction image at receiver end,this thesis carries out the study of image dehazing and deblocking methods.The main work done is as follows:(1)This thesis improves the image dehazing method based on Dark Channel Prior(DCP)principle.This method uses the minimum color component filtered by bilateral filter as the guided image to refine rough transmission.In addition,we modify the transmission of regions which cannot meet the DCP principle,and conduct tone mapping to adjust the tone of whole dehazed images.Experimental results show that the improved dehazing method can obtain better dehazing effect than the previous dehazing method,that is,the details are clearer and the color is more natural.(2)This thesis designs a single image dehazing method called direct denoising dehazing method.This dehazing method obtains the refined transmission by using dehazing contrast-enhancement filters to directly denoise the noisy transmission of foggy images,and achieves image dehazing.Further theoretical analysis gives the effects of dehazing method based on DCP principle and direct denoising dehazing method on compression of dehazed images.We compare the effects of four dehazing contrast-enhancement filters on image quality of dehazed reconstruction image.Experimental results show that the direct denoising dehazing method can meet the requirement of image dehazing and the dehazed image quality after image compression,and simulation results verify the correctness of the theoretical analysis.(3)This thesis designs the image deblocking method via sparse representation and adaptive residual threshold.An adaptive residual threshold for Orthogonal Matching Pursuit(OMP)is designed by combining blind evaluation algorithm for image blocking artifacts.Designed image deblocking method not only achieves good deblocking effect,but also has high image processing speed.