摘要
文章提出了一种基于区域分割的图像融合算法。首先对各个源图像使用基于金字塔链接的多分辨率分割,将源图像在各个尺度上划分为一系列区域,在划分过程中为了克服根标识时的不可预知性,先给定根的数目,然后在这些区域上采用小波变换的方法进行融合。实验表明,该方法可较好的消除边界的模糊性。
<Abstrcat>An image fusion method based on region-segmentation is introduced. The original images are partitioned into a serial of regions in every scales by multiresolution segmentation based on pyramid linking. In order to overcome the unpredictable root-labeling during the segmentation, the root number is given first,then the images are fused by using wavelet transform in all regions. The experimentation results show that the proposed method can eliminate the blurring of the boundary.
出处
《合肥工业大学学报(自然科学版)》
CAS
CSCD
北大核心
2005年第6期577-580,共4页
Journal of Hefei University of Technology:Natural Science
基金
国家自然科学基金资助项目(10171026)
安徽省自然科学基金资助项目(03046102)
关键词
链式金字塔
图像融合
区域划分
小波变换
linking pyramid
image fusion
region-segmentation
wavelet transform