Image Dehazing Algorithm Based On Dark Channel Priori And Retinex Research

Fog,a severe weather with severe consequences,occurs frequently,lasts for a long time and is widely distributed in most areas.The picture is a visual carrier.In the hazy weather,due to the presence of fog,the contrast of the picture or video captured by the imaging system becomes low,the color becomes dim,and the feature is attenuated.Therefore,how to achieve fast and effective Image dehazing,make the collected images more clear and natural,improve the robustness and stability of the imaging system in bad weather,and has a very important significance in image processing.This paper proposes a dark channel prior and Retinex for the dehazing effect and the degree of distortion in the image dehazing algorithm based on dark channel prior,in terms of computation cost,detail processing and image dehazing algorithm based on Retinex theory.The theory combines the dehazing algorithm.The algorithm first performs an average filtering preprocessing on the input image to eliminate noise,improve image quality and make the transmittance of the depth of field mutation more accurate,and then obtain the global atmospheric light value by obtaining the minimum channel map and improving the median filtering.Then,according to the dark channel prior and the steering filtering(the preprocessed image obtained by the mean filtering is used as the guiding image),the transmittance is estimated and optimized,thereby obtaining an initial dehazing image.On this basis,the structural similarity SSIM value of the initial dehazing map and the original fog-containing map is calculated and compared with a preset threshold to determine whether to continue the single-scale Retinex algorithm processing on the initial dehazing map,when the SSIM value When the threshold is greater than or equal to the threshold,the initial de-fogging diagram is processed by the single-scale Retinex algorithm to obtain the intermediate dehazing pattern,and finally the intermediate dehazing pattern is adaptively color-adjusted to obtain the final de-fog image;when the SSIM value is less than the threshold,skip Retinex algorithm processing,directly adaptive color adjustment to get the final dehaze map.The experimental results show that the proposed algorithm not only captures the advantages of image de-fogging algorithm such as image restoration and image enhancement,but also suppresses the distortion effect caused by the combination of the two,which improves the computational efficiency and dehazing effect of the dehazing algorithm.In addition,the evaluation of the dehazing effect introduces the mean structure similarity and the visible edge contrast evaluation index.The comparison results show that the evaluation scheme is more objective than the previous evaluation scheme.