Research And Application Of Haze Removal Algorithm In Remote Sensing Images

With the rapid development of human technology,the damage to nature is beyond the self-healing ability of nature.Leading to the occurrence of global natural disasters.Such as tsunami,haze and other extreme natural phenomena in many places around the world.And other issues.The reduction of effective information will reduce the application value of the image in the military and civilian fields,and it will also increase the economic and time costs.Therefore,it has important practical significance and application value for removing haze in the image.After using the Retinex algorithm and the dark channel prior algorithm to remove haze,there are problems in image restoration in terms of details and color distortion.This article addresses these problems by using Retinex theory and dark channel prior theory to haze in optical remote sensing images.After removing the analysis,an improved Retinex algorithm and an improved dark channel prior algorithm are proposed.The main work of this article is as follows:(1)Analyze the causes of haze formation,and the theoretical basis for the effects of haze on remote sensing image imaging.Research on atmospheric scattering models,haze image degradation models,and image restoration and evaluation to lay a theoretical foundation for removing haze.(2)Detailed analysis of Retinex theory and algorithm.After removing the haze in the remote sensing image from the MSRCR algorithm,there are disadvantages such as unsatisfactory details retention and color distortion.By changing the sigma parameters in the MSRCR algorithm,the high,middle and low frequency information of the image can be more retention,to achieve the effect of improving image details.(3)By analyzing the dark channel prior algorithm,the quality of the dark channel image can affect the image quality after removing the haze.Improve the dark channel algorithm: use different size filter windows to obtain dark channel images,and then use guided filtering to optimize dark channel images and obtain transmittance to improve color distortion after removing haze.(4)In order to solve the problems of haze removal using MATLAB,the software has a long startup time and occupies a large amount of computer memory.This article builds a dedicated image haze operation platform.The image enhancement algorithm,dark channel prior algorithm and the method proposed in this paper are written into the platform,and the function of the platform is tested in detail.The results show that the platform has a fast startup speed and takes up Less memory.