基于DTCWT和稀疏表示的红外偏振与光强图像融合  

Infrared Polarization and Intensity Image Fusion Based on Dual-Tree Complex Wavelet Transform and Sparse Representation

作　　者：朱攀[1] 刘泽阳[1] 黄战华[1] 

机构地区：[1]天津大学精密仪器与光电子工程学院,光电信息技术教育部重点实验室,天津300072

出　　处：《光子学报》2017年第12期207-215,共9页Acta Photonica Sinica

基　　金：国家自然科学基金(Nos.61475113,61275009)资助~~

摘　　要：针对红外偏振与光强图像彼此包含共同信息和特有信息的特点,提出了一种基于双树复小波变换和稀疏表示的图像融合方法.首先,利用双树复小波变换获取源图像的高频和低频成分,并用绝对值最大值法获得融合的高频成分;然后,用低频成分组成联合矩阵,并使用K-奇异值分解法训练该矩阵的冗余字典,根据该字典求出各个低频成分的稀疏系数,通过稀疏系数中非零值的位置信息判断共有信息和特有信息,并分别使用相应的规则进行融合;最后,将融合的高低频系数经过双树复小波反变换得到融合图像.实验结果表明,本文提出的融合算法不仅能较好地凸显源图像的共有信息,而且能很好地保留它们的特有信息,同时,融合图像具有较高的对比度和细节信息.Considering that infrared polarization and intensity image contain common information and their own unique information,a method of image fusion based on Dual-Tree Complex Wavelet Transform and sparse representation was proposed.Firstly,the high and low frequency components of source images wereobtained by using Dual-Tree Complex Wavelet Transform,and the high frequency components werecombined by absolute maximum method.Secondly,ajoint matrix was constructed by low frequency components,and a redundant dictionarywas acquired by using K-singular value decomposition to train the matrix.Based on the dictionary,the sparse coefficient of low frequency component was calculated,and the common information and unique information werejudged by the location of non-zero value of the sparse coefficient,and two kinds of information was merged by proper fusion rules.Finally,the fusion image was obtained by performing inverse Dual-Tree Complex Wavelet Transform on the fused high and low frequency components.The experimental results show that the proposed fusion method can highlight the common information of source images and keep their own unique information,and the fusion image own higher contrast and clearer details.

关 键 词：红外偏振图像 图像融合 稀疏表示 双树复小波变换 K-奇异值分解 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]