摘要
针对传统融合算法处理聚焦区域能力弱、边缘效果差以及目标轮廓提取存在缺陷等问题,提出了基于稀疏分解和背景差分融合的方法.稀疏分解经过鲁棒主成分分析方式提取多聚焦图像轮廓,从而为源图像的准确分离奠定基础;背景差分融合依照源图像与增强图像的差异图提取轮廓信息以准确定位聚焦区域,从而防止引入人工干扰.结果表明,与传统方法相比,本文提出的方法在很大程度上提升了融合效果,能够很好地加强其对噪声的鲁棒性,同时表现出很好的视觉效果.
Aiming at the problems that in the traditional fusion algorithm,the capacity to deal with the focus area is weak,the edge effect is poor,and the defects exist in the target contour extraction,a method based on the sparse decomposition and background difference fusion was proposed. The contours of multi-focus image were extracted with the sparse decomposition method in the robust principal component analysis mode,which could provide the foundation for the exact separation of source image. According to the difference diagram between both source and enhanced images,the contour information was extracted in the background difference fusion method to accurately locate the focus area,and thus the introduction of manual interference could be prevented. The results showthat compared with the traditional method,the proposed method can greatly increase the fusion effect,enhance its robustness to noise,and exhibits good visual effect at the same time.
作者
王金茹
WANG Jin-ru(School of Management and Journalism,Shenyang Sport University,Shenyang 110102,China)
出处
《沈阳工业大学学报》
EI
CAS
北大核心
2018年第4期436-440,共5页
Journal of Shenyang University of Technology
基金
辽宁省自然科学基金资助项目(2015020031)
2015年度辽宁省社科联与高校社科联合作课题(lslgslhl-156)
关键词
稀疏分解
背景差分融合
鲁棒主成分分析
图像融合
多尺度变换
剪切波变换
图像处理
融合算法
sparse decomposition
background differential fusion
robust principal component analysis
image fusion
multi-scale transform
shear wave transform
image processing
fusion algorithm