摘要
对基于边缘的算法进行了改进,综合考虑了边缘和能量结构.首先利用非下采样contourlet(NSCT)变换,对图像进行多尺度和多方向的分解;然后将低频系数按边缘能量大小划分为边缘和平滑2部分,边缘部分采用边缘能量取最大的融合方法,平滑部分使用基于局部能量的规则进行融合,高频子带使用相关系数和局部方差相结合的重要性测度法进行融合;最后对融合系数进行NSCT反变换得到融合图像.实验结果表明,该算法融合效果较基于边缘算法有所改善,是一种有效兼顾细节和能量结构的方法.
Edge-based image fusion algorithm can not preserve energy structure which is important in indicating the reflection feature of the observed objects. This paper proposed an enhanced algorithm considering both edge and energy structures. Nonsubsampled contourlet transform is firstly deployed to analyze input image in both multi-scale and multi-direction sub-bands. Low frequency band is then divided into contour and smooth regions by edge energy. Edge energy is used as a fusion judgment for the contour region while local energy and correlation are selected as judgments for the smooth region. Local variance and correlation are chosen to merge the coefficients of high frequency sub-bands. Inverse NSCT transform is finally used to get the fused image. Experimental results show that the new algorithm is an effective method that introduces more detailed information while preserves power structure.
出处
《中国科学院研究生院学报》
CAS
CSCD
北大核心
2009年第5期657-662,共6页
Journal of the Graduate School of the Chinese Academy of Sciences
