基于小波包的遥感图像融合新算法

提出了一种基于正交小波包的局域方差遥感图像融合新算法．该算法利用正交小波包变换，把图像分解成不同尺度的低频和高频部分，采用小波包局域窗口和子区域窗口统计，把小波包系数分类成边缘和非边缘系数．在融合处理中，把低频图像的小波包系数平均值作为融合后的低频系数，高频细节系数根据不同区域特征选择方法以及对应图像小波包系数的多窗口区域方差，来确定融合后高频小波包系数．实验结果表明，这种方法能够在保留图像微小细节方面获得满意的结果，有效且优于其他图像融合算法．

New Remote Sensing Image Fusion Based on Wavelet Packet Algorithm

In this paper we presented a new wavelet packet transform method based on local deviation. Wavelet packet transform decomposed an image into low frequency hand and high frequency hand in different level; local window and local sub-window statistic in the wavelet domain were used to further classify the wavelet coefficients in- to the edge and non-edge coefficients. In the fusion processing, the fused approximate coefficients were obtained by average method, the fused detailed coefficients were obtained by several different area-based feature selection method and the multi-window local deviation of the corresponding image wavelet packet coefficients. Experimental resulks show that the proposed algorithm can retain subtle image detail and achieve an excellent effect and the new image fusion algorithm provides a performance better than that of other methods.