Effect of Wavelet Basis and Decomposition Levels on Performance of Fusion Images from Remotely Sensed Data
GONG Jian-zhou1,2,LIU Yan-sui2,XIA Bei-cheng3,CHEN Jian-fei1 (1.School of Geographical Sciences,Guangzhou University,Guangzhou 510006;2.Institute of Geographic Sciences and Natural Resources Research,CAS,Beijing 100101;3.School of Environmental Science and Engineering,Sun Yat-Sen University,Guangzhou 510275,China)
The integration of wavelet transform with IHS has apparently become popular to incorporate multi-sources remotely sensed data to create a new image,in that better fusion results might be produced by the method.But related factors are poorly considered,which would directly affect the fusion results,such as wavelet basis,decomposition levels.In Matlab,SPOT images,including panchromatic band and three multispectral bands,were used as analytical data here.Fusion images were created by highlighting those factors:different cluster of wavelets,wavelet basis with different serial number,and decomposition levels.Three indices for image performance,including entropy(joint entropy),average gradient and deviation,were calculated.All fusion images had similar performance with only one wavelet decomposition level.But fusion images had different performance with increasing levels and different wavelet basis.Three wavelet basis,including coif5,sym5,dmey,did not show different performance with changing levels.Two wavelet basis,db1 and bior3.1,had unique response characteristic within different ranges of level,such as indices value were stable while levels between 1 and 4,and then monotonically decreasing.Taking rbio3.1 as wavelet basis,the distortion image was created when the wavelet decomposition levels was equal to 4.With the increasing levels,fusion images were created and characterized by dramatic distortion in image spectral.Besides,different cluster of wavelets in Matlab has different fusion image in performance.The paper could better support the practical application of remote sensed data in defining method to improve image performance by fusion multi-sources data and make the operation simpler and faster.
【Fund】： 国家自然科学基金重点项目(40635029);; 中国博士后科学基金项目(20080440511);; 中国博士后科学基金特别基金项目(200902132);; 广州市属高校科技计划项目(08C027)
【CateGory Index】： TP751
【CateGory Index】： TP751