Research Of Image Dehazing Algorithm Based On Physical Model

As a kind of common weather phenomenon,the haze weather often brings a lot of trouble to people’s study and life.The quality of image is often not very high.The fog on the image will affect the information on the image,and bring great trouble to the analysis and processing of images.Therefore,it is of great practical significance to images dehazing.With the development of the computer field,the relevant algorithms of image dehazing are also developing and becoming more and more mature.There are two kinds of algorithms on traditional image dehazing: one is based on image enhancement,the other is based on image restoration.The algorithm of image dehazing based on image enhancement can enhance the visual effect of the image,and increase the contrast of the image.But there are also problems,such as image distortion and incomplete defogging.The algorithm based on image restoration often depends on the atmospheric scattering model.According to the mechanism of the image,we can get better results.But this method needs to establish the correct physical model or scientific prior knowledge.In this paper,the algorithm of image dehazing based on physical model is studied.The main work are as follows:(1)An algorithm of image dehazing based on boundary constraint is proposed.The main problem to be solved of this algorithm is,that the existing algorithm based on TV regularization model will lose the edge information of some images after processing,and high frequency information similar to the sharp change of depth value in the smooth region of the image,resulting in halo.The proposed algorithm adds boundary constraint to the existing model of estimating transmission,and constructs a new algorithm based on that similar color always have similar depth value.A weighted function is used to estimate the transmission better,and the effectiveness is verified by experiments.(2)An algorithm of image dehazing based on dehazing control factor is proposed.Based on the prior knowledge that there is a positive correlation between the factor and the difference of the maximum and minimum color channel,a mathematical model is proposed.The parameters are estimated by the particle swarm optimization technique.In order to prevent the distortion of in the local area,the support vector data description is used to estimate the atmospheric light.By experiments,the proposed algorithm achieves good results.