基于多小波的高分辨率与多光谱图像融合

摘要

为了将高分辨率与多光谱图像进行有效地融合，本文提出了一种将 HIS 变换，多小波变换及 PCNN神经网络相结合的图像融合算法。该算法首先采用HIS变换将多光谱图像分为 H、I、S 3个分量，然后使用多小波分别分解I分量和高分辨率图像得到各自的高频和低频分量，接着低频采用矢量融合，高频采用PCNN融合，最后采用多小波逆变换以及HIS逆变换得到融合图像。实验结果显示本文方法在融合时间、视觉效果及客观评价指标上综合效果比较好。%In order to fuse high-resolution and multi-spectral image better,a algorithm is presented in this paper of combining IHS transformation, multiwavelet transform with PCNN.First transfer the Multi-spectral image with HIS into 3 components of I, H S.Then High-resolution image and I component after the first step about IHS transformation are decomposed by Multiwavelet Transformto get to the sub-band coefficients at different scales are obtained. Thirdly fuses the images with combining of vector intergration and PCNN.At last the fusion images are reconstructed by HIS inverse and Multiwavelet inverse transform. It can get better results in both time and fusion in test.