Single Image Dehazing Using Color Attenuation Prior Based On Haze-Lines

With the development of science and technology,the computer vision system has been widely used in daily life.However,sometimes the image acquired by computer vision system has the problem of decreasing clarity and color saturation,which is due to the influence of bad weather such as haze.This brings certain difficulty to the later image recognition and image processing,so the image dehazing technology has important research value.In this paper,a new single image dehazing method is proposed,which is based on the prior dehazing algorithm of color attenuation.The main work of this paper is as follows:Firstly,aiming at the estimation method of atmospheric light in the prior algorithm of color attenuation,we introduce the prior estimation of the position and value of atmospheric light by the haze-lines in the non-local image.The prior principle of haze-lines is that under the influence of fog,the RGB value of image pixels will form several different lines,and the intersection of these lines is the atmospheric light value we seek.Secondly,in view of the shortcomings of the atmospheric scatteringcoefficient in the prior of color attenuation,we introduce the dynamic scattering coefficient function,which is about the exponential growth of scene depth.According to the scene depth of different images,different values of atmospheric scattering coefficient can be obtained.Finally,this paper compares the image dehazing algorithm with the color attenuation prior algorithm,non-local image dehazing algorithm and other dehazing algorithms to analyze the effectiveness of the image dehazing algorithm.