基于M带对偶树小波变换图像融合及超分辨重建

Image Fusion and Super-resolution Restoration Based on M-band Dual-tree Wavelet Transform

提出基于M带对偶树小波变换图像融合及超分辨重建的新方法。首先介绍M带对偶树小波变换的基本原理,分析两个低频子带的关联性。待融合的两幅图像分别作为二维M带对偶树小波变换的两个低频子带,其他子带系数赋零,然后执行逆M带对偶树小波变换并调整像素范围得到高分辨的融合图像。算法结构简单,计算复杂性低。实验结果证实获得的图像信息融合充分,边缘和纹理明晰。

A new image fusion and super-resolution restoration method based on M-band dual-tree wavelet transform is proposed.The theory of the M-band dual-tree wavelet transform is introduced firstly.Then the relationship of the two low-frequency subbands is analyzed.The two different images are respectively considered as the two low frequency subbands of the 2D M-band dual-tree wavelet transform,and the coefficients of other subbands are set zero.Then the inverse M-band dual-tree wavelet transform is performed and the pixel range is adjusted to achieve the super-resolution fusion image.The proposed method has a simple structure and low complexity.Experiments testify that the obtained image contains sufficient information and has sharp edges and textures.