基于多小波分析的图像融合算法

Image fusion algorithm using multiwavelet analysis

从电磁波频谱带宽的角度,提出了图像融合的一种物理解释。介绍了多小波变换的一般理论,描述了多小波图像融合的方法。用GHM多小波、CL多小波及Db4单小波对可见光与红外图像进行了融合。利用平均熵作为判定准则,比较了GHM多小波、CL多小波及Db4单小波的融合效果,实验结果表明CL多小波算法效果最好。

A physics-based explanation of image fusion is proposed in this paper. Multiwavelet transform offers simultaneous orthogonality, symmetry, compact support, and vanishing moment, which are not possible with scalar wavelet transform. In the case of multiwavelet, there are more than one different coefficient detail blocks in the same direction and in the same resolution. Objectives of image fusion schemes extract all useful information from the source images. In this paper, optical image and IR image are fused using multiwavelets and scalar wavelet. Image fusion performances of Db4 scalar wavelet, GHM multiwavelet, and CL multiwavelet are compared quantitatively. Experimental results demonstrate that the fused image entropy using CL multiwavelet is the maximum in all fused images.