Research On Dodging Methods Of Optical Remotely Sensed Image

The uneven illumination phenomenon of optical remote sensing image is the importantproblem which needs to be solved in the preprocessing of remote sensing image. The dodgingtechnique is the best method to solve the problem and it is also the key technique affecting thequality of image productions as well as the further processing. In order to solve the problemswith the unevenness both in brightness and contrast, the involved key techniques are researchedbased on the studyof existing dodging methods, and major works are listed below:1,In this dissertation, the actuality of dodging methods both in home and abroad areintroduced for optical remote sensing image. Two models of dodging processing are concluded,and the key and difficult points are point out to the dodging processing of optical remote sensingimage.2,The principle of several kinds of common dodging method is expounded, and theircharacters and applicability are concluded, which prepares theoretical foundation for furtherresearch.3,The disadvantages of MASK dodging method are point out to the phenomenon ofunevenness in brightness of an image. Aiming at these disadvantages, an improved MASKdodging method is proposed based on wavelet transform. According to the analysis of thecoefficients after wavelet transform, the relationship is concluded between the space ofluminance and the wavelet coefficients of the image with unevenness in brightness. Theproposed method deals with the approximate coefficient and the detailed coefficient respectively.Finally the experiment shows that the improved dodging method has advantages over thetraditional MASK dodging in the aspect of the equilibrium of brightness and contrast, theenhancement of local image and the efficiencyof calculation.4,Aiming at the disadvantage of the existing methods which can only solve theunevenness in brightness, but not solve the unevenness in contrast, this dissertation usesLRM(Local Range Modification) theory and its application in image processing for reference.And an unevenness correction method is proposed based on LRM and mean-variancespecification for remote sensing image. The experiment shows that the proposed method cansolve the unevenness both in brightness and contrast commendably. It can make the result imageconsistent in brightness and contrast, which makes the visual effect better.