QQ客服 官方微博 反馈留言

	改进拉普拉斯能量和的尖锐频率局部化Contourlet 域多聚焦图像融合方法; Sum-modified-Laplacian-based Multifocus Image Fusion Method in Sharp Frequency Localized Contourlet Transform Domain
	Xiaobo Qu ; 屈小波 ; Jingwen Yan ; 闫敬文 ; Guide Yang ; 杨贵德
	2009-05-18
关键词	图像融合 多聚焦图像 Contourlet 变换 伪吉布斯现象 小波变换 Image fusion multifocus images contourlet transform pseudo-Gibbs phenomena wavelet transform
英文摘要	为了克服Contourlet 融合在远离支撑区间上出现的混叠成分，抑制融合图像在奇异处产生伪吉布斯现象，提出改进拉普拉斯能量和的尖锐频率局部化Contourlet ( Sharp Frequency Localized Contourlet Transform-SFLCT)域多聚焦图像融合方法。首先，采用SFLCT 而不是原始的Contourlet 对多聚焦图像进行分解。接着，将多聚焦图像空域融合方法中评价图像清晰度的指标引入到SFLCT 变换域，采用拉普拉斯能量来选择变换域系数。然后，逆SFLCT 重构得到融合结果。最后，采用循环平移(Cycle Spinning)来提高SFLCT 的平移不变性，有效抑制融合图像在奇异处产生伪吉布斯现象。实验结果表明：对于多聚焦图像，所提方法比循环平移小波变换互信息提高5.87%, QAB/F 提高2.70%,比循环平移Contourlet 方法互信息提高1.77%,QAB/F 提高1.29%,视觉效果优于典型的空域分块拉普拉斯能量方法和平移不变小波变换方法 ============ Abstract: In order to suppress pseudo-Gibbs phenomena around singularities of fused image and reduce significant amount of aliasing components which are located far away from the desired support when the original contourlet is employed in image fusion, Sum-modified-Laplacian-based multifocus image fusion method in sharp frequency localized contourlet transform (SFLCT) domain is proposed. First, SFLCT, instead of the original contourlet, is utilized as the multiscale transform to decompose the source multifocus images into subbands. Second, typical measurements for multifocus image fusion in spatial domain are introduced into contourlet domain and Sum-modified-Laplacian (SML), evidenced in this paper with the best capability to distinguish SFLCT coefficients is from the clear parts or blurry parts of images, is employed in SFCLT subbands as measurement to select SFLCT transform coefficients. Third, inverse SFLCT is used to reconstruct fused image. Finally, cycle spinning is applied to compensate for the lack of translation invariance property and suppress pseudo-Gibbs phenomena of fused images. Using the proposed fusion method, experimental results demonstrate that mutual information is improved by 5.87% and transferred edge information QAB/F is improved by 2.70% compared with cycle spinning wavelet method, while mutual information is improved by 1.77% and QAB/F is improved by 1.29% compared with cycle spinning contourlet method. Meanwhile the proposed fusion method outperforms block-based spatial SML method and shift-invariant wavelet method in term of visual appearance.; 国家自然科学基金(No.60472081),航空基础科学基金(No.05F07001)
语种	中文
出版者	光学精密工程
内容类型	期刊论文
源URL	[http://dspace.xmu.edu.cn/handle/2288/8087]
专题	信息技术－已发表论文
推荐引用方式 GB/T 7714	Xiaobo Qu,屈小波,Jingwen Yan,等. 改进拉普拉斯能量和的尖锐频率局部化Contourlet 域多聚焦图像融合方法, Sum-modified-Laplacian-based Multifocus Image Fusion Method in Sharp Frequency Localized Contourlet Transform Domain[J],2009.
APA	Xiaobo Qu,屈小波,Jingwen Yan,闫敬文,Guide Yang,&杨贵德.(2009).改进拉普拉斯能量和的尖锐频率局部化Contourlet 域多聚焦图像融合方法..
MLA	Xiaobo Qu,et al."改进拉普拉斯能量和的尖锐频率局部化Contourlet 域多聚焦图像融合方法".(2009).

个性服务

查看访问统计

相关权益政策

暂无数据

收藏/分享

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。