一种改进多分辨率图像融合算法

提出一种基于局部熵的多分辨图像融合算法。利用小波变换得到待融合图像的多分辨结构，同时得到图像的多分辨局部熵序列。以局部熵为判据，在图像多分辨结构相应各级上进行融合，得到融合图像的多分辨结构，利用小波逆变换重构融合图像。实验结果表明，该图像融合方法在保留TM多光谱图像光谱分辨率的同时，通过融合SPOT全色图像提高了空间分辨率，丰富了图像细节信息。

Improved algorithm for image fusion based on multi-resolution analysis

This paper introduces a local entropy-based multi-resolution image fusion technique.The algorithm consists of three steps: First,decomposing the input multi-spectral image by wavelet transform to obtain the multi-resolution local entropy sequences of each input image.Second,according to local entropy-based criterion,the multi-resolution analysis of the fused image can be obtained on the corresponding levels of the multi-resolution analysis of the input images.Finally,the output image can be obtained through inverse wavelet transform.Experiment results show that image details are enriched while spectral solution of TM image is reserved by fusing TM image with SPOT image with the image fusion method.