Research On Image Processing Technology Under Fog And Haze

With the rapid economic development,also accompanied by a number of environmental problems,more and more cities appear bad haze weather.Fog haze weather not only causes great distress to people's breathing,but also to people's traffic bringing a lot of inconvenience,such as video surveillance in the transportation sector.Clear and effective images and video are important channels for obtaining various kinds of information in modern society.Therefore,it is necessary to obtain the image with acceptable quality for the fogging processing of the image in the haze weather.In this paper,the fog-forming algorithm is studied by two aspects.The first is the use of image enhancement,and then the physical model of image restoration.Image enhancement is based on the contrast and brightness of an image,regardless of the quality of the image.In this paper,the Retinex algorithm is mainly studied.Aiming at the existing problems of the existing algorithms,an adaptive scale parameter Retinex defog algorithm is proposed.Then we compare the results of several Retinex image enhancement algorithms on the fog and haze weather image processing,and then evaluate the pros and cons of various defogging algorithms.Based on the reason of image degradation under fog and haze,combined with the atmospheric physical model,this paper presents a defog algorithm based on the optimization of guided filtering.At the same time,the algorithm of quad-tree search is used to estimate the atmospheric light accurately,and the sky region is estimated by the difference image obtained by the maximum filter and the minimum filter,and the transmittance is compensated.At the same time,In this paper,a filter window based on weight optimization is used to improve the guide map by using gamma correction.Finally,the guided filter is used to improve the transmittance and the fog-free clear image is reconstructed based on the dark scattering model.Simulation results show that the algorithm proposed in this paper is fast,which satisfies the real-time processing of image.