基于K-L变换的小波图像融合方法

A Wavelet Transform Image Fusion Method Based on Karhunen-Loeve Transform

在多源信息融合中,小波多分辨率分析是一种最常用的方法。这里提出在小波多分辨率分析下,利用K-L变换的融合方法。首先利用小波变换对序列图像进行多分辨率分解,对相应的小波系数矩阵进行K-L变换,计算出小波系数权重,按照所得的权重融合小波系数,最后将小波融合系数逆变换实现图像的融合处理。实验结果证实这种方法有效的利用了图像的相关性,主观视觉效果分析和客观统计参数评价分析都表明,新方法的性能优于直接对小波系数进行平均的融合方法。

Wavelet multi- resolution analysis is an important method in image fusion. An image fusion method of wavelet multi-resolution based on Karhunen - Loeve transform is presented. Image sequences are decomposed using wavelet transform, fusion weight value for the coefficients of wavelet are resolved using K - L transform, and the coef- ficients of wavelet are fused by the fusion weight value, then images are fused by inverse transform of fusion wavelet coefficients. The experimental results show the method has made good use of the relativity of image sequences, and it＇ s better than direct fusing the coefficients of wavelet.