Research On Single Image Dehazing Algorithm Based On Dark Channel Prior

The image collected in the haze weather is prone to blur and low contrast.This will bring huge challenges to subsequent image processing tasks such as face recognition,target detection,and target segmentation.Therefore,th e study of dehazing algorithm is a matter of extremely application value and practical significance.However,the existing image dehazing algorithms are still affected by some factors that cause color distortion and even partial loss of details.These factors mainly include deviations in the estimation of atmospheric light values and transmission maps,and low image intensity.Aiming at the problem of color distortion and dark brightness in the dehazing results of the dark channel prior algorithm,this pape r proposes an image dehazing algorithm based on quadtree search and HSI color space.Specifically,when the algorithm estimates the atmospheric light candidate region,the quadtree search algorithm is preferred.This not only improves the accuracy of estimating the atmospheric light value,but also improves the color deviation of the dehazed image caused by the positioning of the atmospheric light candidate region on a bright white object.Secondly,the preliminary dehazing result is converted to the HSI color space,and only the brightness I is subjected to the adaptive histogram equalization process with the limited contrast ratio.It effectively solves the problem of low brightness presented by the dehazing result,and highlights the details of the image more.Finally,the experimental results on real hazy images show that the algorithm is superior to several well-known image dehazing algorithms in both qualitative and quantitative evaluation.Aiming at the halo effect and color distortion in the dehazed r esults of the dark channel prior algorithm,this paper proposes a sky region segmentation and an improved dark channel image dehazing algorithm.Specifically,the algorithm first proposes a sky region segmentation scheme based on the minimum channel histogram of a hazy image,which can be considered as the first step to accurately estimate atmospheric light.An accurate atmospheric light value can improve the color distortion in the dehazed results.In addition,in order to further reduce the influence of the halo effect and color distortion of the sky region on the image,the dark channel scheme is improved by fusing the minimum channel and performing the minimum filtering operation.Experimental results show that this method has a better competitive advantage than several latest image dehazing methods on the RESIDE synthetic test sets and some real hazy images.