摘要
文中在阐述小波变换原理的基础上,针对融合算子的构造,给出了一种新的基于小波系数相关性的融合方法。以分解后的高频子图像的区域均值、方差、协方差统计参量构造匹配度和加权因子,以此对高频子图像的对应区域进行融合计算;低频部分采取加权平均融合规则;最后,通过小波逆变换得到融合图像。用熵、交叉熵和交互信息量对融合结果进行了客观评价,并与其它几种算法进行了比较。实验结果表明,此方法融合效果要优于一般融合方法,计算也比较简单,有一定的工程实用价值。
Based on expatiate wavelet transform principle, aimed at the construction of fusion operator, a new scheme using the correlation of decomposition coefficients is presented. With mean, variance and covariance of decomposed high frequency sub-image, the matching degree and adding coefficient can be gained, and then the fusion calculation of high frequency sub-image can be completed. For low frequency sub-image, average coefficient is adopted. Finally the fusion image is obtained by taking inverse wavelet transform. In addition, with entropy, cross entropy and mutual information, the performance of fusion scheme is evaluated and compared with other schemes. The experimental results show this fusion scheme is effective than other fusion schemes, the calculation course is simple, and has application value to engineering.
出处
《激光与红外》
CAS
CSCD
北大核心
2006年第3期227-230,共4页
Laser & Infrared
基金
国家863高技术项目(2003AA823050)
关键词
图像融合
小波变换
分解系数
相关
融合性能
image fusion
wavelet transform
decomposition coefficients
correlation
fusion performance
