非抽样形态小波的构造及在图像融合中的应用  

Construction of undecimated morphological wavelets and application in image fusion

作　　者：邓苗[1] 张基宏[1,2] 任仙怡[2] 梁永生[1,2] 

机构地区：[1]深圳大学信息工程学院,深圳518060 [2]深圳信息职业技术学院,深圳市可视媒体处理与传输重点实验室,深圳518172

出　　处：《深圳大学学报（理工版）》2013年第3期254-260,共7页Journal of Shenzhen University(Science and Engineering)

基　　金：国家自然科学基金资助项目(61271420);广东省自然科学基金资助项目(S2012020011034)~~

摘　　要：提出一种适合图像融合的非抽样形态小波的构造方法.采用新的非线性滤波器组描述形态小波,去掉其中下采样部分实现非抽样,并将其用于图像融合,用以克服形态小波方法的块状效应.构造适用于所有类型图像融合的S-变换形态小波,通过增加通道和引入增强效果,提高图像融合效果.A method of constructing undecimated morphological wavelets （UMW） applicable to image fusion is proposed in this paper. Firstly, the traditional morphological waveletes （MW） are represented with a novel non-linear filter banks. Secondly, the UMW is constructed by removing the downsampling to overcome the block effect of the MW based image fusion methods. Thirdly, a non-separable S-transform MW applicable to all kinds of image fusion is presented. By expanding UMW to multiplechannels and adding an enhancement factor, the quality of image fusion is improved.

关 键 词：计算机图象处理 图像融合 形态小波 非抽样形态小波 多尺度变换 图像增强 不可分离滤波器 非线性滤波器 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]