Research On Remote Sensing Image Haze Removal Methods

As a result of atmospheric scattering and absorption,remote sensing images shoot at badweather conditions was unclarity and characteristiced by low contrast ratio.Using someprocessing methods,we can get a lot of useful information through the images with low contentsof information,and make those images useful.Therefore,the haze removal processing of thosedegradation images has great application meaning.This paper is aimed at researching forpowerful and automatical processing methods that are applicable for remote sensing images.Themajor works has done are listed as follow:1.Researched the classical dehaze methods,summarized their characteristic.And didexperiments using some of those methods.2.Researched different colour spaces for get avoid of colour distortion. Compared the lineartransformation processing results in RGB colour space and HIS colour space.3.Using dark channel prior for remote sensing image haze removal,and simplified thedehaze method ueing dark channel prior in two points according to the characteristic ofremote sensing image:Firstly,give atmosphere an approximation value during the processof estimating the value of transmissivity.Secondly,escape from the processing of thetransmissivity image.After the simplication,we summarized the characteristic and theapplicability of this method.4.Using Retinex theory for remote sensing image haze removal. Focus on Single-scaleRetinex, Multi-scale Retinex and color rendition Multi-scale Retinex.For get betterprocessing effects and avoid manual work,this paper did linear Contrast Stretch and setautomatic choice of the scale.5.Selected several quality evaluation parameters for impersonality evaluate the results ofthose methods.