Texture Information Extraction In Remote Sensing Imageries With Gray Level Co-occurrence Matrix And Wavelet Transform

Image auto-reorganization is the pop problem in the fields of graphics, patternreorganization, and artificial intelligence. Remote sensing imagery auto reorganizations havebeen one of the focuses by international scholars due to the complexity of the imagingmechanism. The precision has been restricted to a certain extent when identifying groundtargets by spectrum analysis owing to the complex medium existing between satellite sensorand earth surface, which bring more factors such as the atmosphere, cloud and so on, to affectthe echo and dispersion mechanism of ground targets to electromagnetic wave. The facts thatthe same targets with different spectrums and the different targets with the same spectrumsmake it helpless for spectrums analysis to identify various ground objects, and as a resultpeople begin to seek other techniques to improve the precision. The spectrum and geometryresolutions have been improved in recent years with the development of satellite remotelysensed technique, and at the same time the texture information in those newly remote sensingimageries have become more abundant than before. How to make use of the textureinformation in high resolution satellite imageries, such as Quickbird, IKONOS and so on isone of the key issues confront with scholars all around the world, and of course the extractionof texture information is the basic of this issue.The aim of this article is extraction of texture information in high resolution satelliteimageries with gray level co-occurrence matrix (GLCM) and wavelet transform. Firstly, thisarticle gives an overview of measures of extracting texture information in recent years, andthen takes two experiments to extract texture information from IKONOS imageries withGLCM and wavelet transform. The results show that texture features extracted by GLCMpossess more variable, which lead to unsatisfied classification precision by those featuresalone, on the other hand the texture features gain by wavelet transform show stabilizationcharacteristic, which possibly due to the multi-scale characteristic of wavelet transform andthe same one texture imageries own. The classification precision has been improved togetherwith spectrums information.This article includes four parts. Study background is introduced in the first chapter,including the signification and progress of texture information, common analysis measures aswell. In chapter 2, gray level co-occurrence matrix is explained in detail, and an experience isimplemented. In consequent chapter analogous experience is toke after introduction ofwavelet transform. The final part of this article is conclusions and prospects.