Research On Brightness Balance Defogging Algorithm Based On Fusion Of Dark Channel Prior And Retinex Algorithm

In China,the smog is more and more serious,the image's quality of the smog weather condition is also worse and worse,and the degraded image will cause many image-based vision system can not use well.Therefore,research on good,fast,widely applicable fog removal algorithm has high theoretical significance and application value.In recent years,with the appearance of many kinds of defogging algorithms,people's demand for image quality is getting higher and higher.In order to get higher quality image,this paper presents a new algorithm of defogging based on brightness balance which combines dark primary color prior and Retinex algorithm.The main work is as follows:(1)The basic principle of the algorithm based on Dark Channel Prior is analyzed,it is found that the defogging effect of the Algorithm is obviously enhanced after the interference of the sky,white building and other areas with high-brightness pixels is eliminated The fog removal theory of Retinex algorithm is analyzed,and it is found that this algorithm is easy to produce Halo artifacts where the brightness gradient changes greatly.Therefore,the idea of combining the two algorithms and carrying out brightness balance to achieve defogging is determined.(2)To extract and process the region recognition of the image with fog: Firstly,the luminance information of the image is obtained by converting the image from RGB to HSV model,and then the luminance image is binarized,the images with fog are divided into highlight regions(including sky,fog,white buildings and reflective water surface)and low highlight regions;and then the Dark Channel Prior Algorithm is used to process the low highlight regions without the interference of highlight pixels;finally,in order to improve the visual effect of the defogged image,the final defogged image is obtained by adjusting the brightness balance ofthe defogged image.(3)Based on Matlab2016 a,the image defogging prototype system is designed,which includes three modules: Image defogging display,image defogging process and image quality evaluation.(4)The image quality evaluation method is analyzed and studied and the experimental results of the Algorithm are evaluated.Based on the analysis of the image quality evaluation data from the prototype system,the effectiveness of the improved algorithm is evaluated and verified by subjective and objective evaluation methods.From the subjective point of view,the fog removal image obtained by this algorithm greatly reduces the halo phenomenon,makes the brightness more balanced,and the color contrast more clear.In this paper,the structural similarity(SSIM),mean square error(MSE)and running time are used to evaluate the defogging images.The experimental results show that the proposed algorithm makes full use of the advantages of Dark Channel Prior and Retinex Algorithm,improves the defogging effect and efficiency,and increases the simplicity and effectiveness of the Algorithm.