Improved Single Image Dehazing Algorithm Using Dark Channel Prior

Because of the influence of the haze weather,the images which capture from the image acquisition equipment not only have low contrast and color inconsistency,but also reduce the application value.So,it is significant to improve the quality of the haze images.In this paper,we have studied of the single image dehazing algorithm,discussed the core ideas of different single image algorithm,two improved algorithms are proposed with a large number of theoretical and experimental analyses.(1)An improved algorithm for single image dehazing based on filtering is proposed.Firstly,halo effect is removed by using median filter for dark channel prior function.Secondly,employing modified factors would better raise the accuracy of transmission map at high-brightness and sky areas.Finally,wiener filter is used for smoothing the transmission map,and let the local transmission remain unchanged at the same time reduce the time complexity.Experiments demonstrate that the proposed method would better remove the haze and halo effect,reduce the running time of the algorithm.(2)A novel dehazing method is carried out based on image fusion.First,the dark dehazed image with little color inconsistency is obtained by using improved dark channel prior method.Second,by analyzing the characteristics of haze image and haze-free image in HSI color space,the other dehazed image with high saturation and intensity is derived by dealing with the S and I component adopting contrast limited adaptive histogram equalization algorithm(CLAHE).Finally,the ultimate dehazed image is restored based on image fusion at pixel level using the above two input images.Experiments indicate that the proposed algorithm can enhanced the contrast of dehazing images and better restored color inconsistency.Moreover,the way to combine the physical model and the non-physical model is one of the research directions in the future.