基于边缘和多特征选择的多聚焦图像融合
|
摘要
为了有效度量多聚焦图像中的聚焦区域,提出一种基于边缘和多特征选择的多聚焦图像融合算法。利用NSCT对源图像进行分解,对得到的低频子带系数采用基于边缘的融合策略;对于高频子带系数,通过拉普拉斯能量和、区域能量和方差提取邻域窗口的显著性特征,提出一种新颖的基于多特征选择的高频融合策略,该融合策略可以综合考虑区域特征的贡献,自适应的选取显著性特征和融合方法;最后进行NSCT逆变换得到融合图像。实验结果表明,与常用的融合方法相比,在主观效果上,本文方法能够更有效保留源图像中的边缘和细节信息;在客观指标上,本文方法的融合图像在互信息、边缘保持度、熵、结构相似度以及标准差等客观评价具有明显的优势。
In order to measure effectively the area of focus in a multi-focus image,a novel multi-focus image fusion algorithm based on edge and multi-feature selection is proposed in this paper.Two registered original images are decomposed using non-subsampled Contarlet transform(NSCT) separately,and the selection principle of the low frequency subband coefficients is based on edges of images.For the high frequency subband coefficients,the prominent features are extracted through Laplace energy sum,region energy and variance,and a novel high frequency fusion strategy based on multi-feature selection is proposed.This fusion strategy can comprehensively consider the contribution of regional features and select prominent features and fusion methods adaptively.Finally,the fused image is obtained by performing inverse NSCT on the combined coefficients.Compared with common fusion rules,experimental results show that the fusion rule proposed in this paper preserves edges and the details of original images well.Objectively,fusion image of the new fusion rule is more superior in the objective evaluation index,such as mutual information,edge preserving degree,entropy,structural similarity and the standard deviation.
In order to measure effectively the area of focus in a multi-focus image,a novel multi-focus image fusion algorithm based on edge and multi-feature selection is proposed in this paper.Two registered original images are decomposed using non-subsampled Contarlet transform(NSCT) separately,and the selection principle of the low frequency subband coefficients is based on edges of images.For the high frequency subband coefficients,the prominent features are extracted through Laplace energy sum,region energy and variance,and a novel high frequency fusion strategy based on multi-feature selection is proposed.This fusion strategy can comprehensively consider the contribution of regional features and select prominent features and fusion methods adaptively.Finally,the fused image is obtained by performing inverse NSCT on the combined coefficients.Compared with common fusion rules,experimental results show that the fusion rule proposed in this paper preserves edges and the details of original images well.Objectively,fusion image of the new fusion rule is more superior in the objective evaluation index,such as mutual information,edge preserving degree,entropy,structural similarity and the standard deviation.
引文
[1] Zhu P,Ma X,Huang Z.Fusion of infrared-visible images using improved multi-scale top-hat transform and suitable fusion rules[J].Infrared Physics &Technology,2017,81:282-295.
[2] Geng P, Huang M, Liu S, et al.Multifocus image fusion method of Ripplet transform based on cycle spin-ing[J].Multimedia Tools and Applications,2016,75(17):10583-10593.
[3] Ancuti C O,Ancuti C,De V C,et al.Single-scale fusion:an effective approach to merging images[J].IEEE Transactions on Image Processing,2017,26(1):65-78.
[4] ZHU Pan,LIU Ze-yang,HUANG Zhan-hua.Infrared polarization and intensity image fusion based on dual-tree complex wavelet transform and sparse representation[J].Acta Photonica Sinica,2017,46(12):1210002. 朱攀,刘泽阳,黄战华.基于DTCWT和稀疏表示的红外偏振与光强图像融合[J].光子学报,2017,46(12):1210002.
[5] Donoho D L,Duncan M R.Digital curvelet transform:strategy,implementation and experiments[C].Proc.of SPIE.San Jose:SPIE,2000,12-30.
[6] WU Peng,LI Wen-lin,QI De-yu,et al.An image fusion algorithm based on threshold-improved contourlet transform[J].Journal of South China University of Technology(Natural Science Edition) ,2017,45(1):35-41. 吴鹏,李雯霖,齐德昱,等.基于阈值改进Contourlet变换的图像融合算法[J].华南理工大学学报(自然科学版),2017,45(1):35-41.
[7] Ikuta C,Zhang S,Uwate Y,et al.A novel fusion algorithm for visible and infrared image using non-subsampled contourlet transform and pulse-coupled neural network[A].The 9th International Conference on Computer VisionTheory and Applications[C].Lisbon,Portugal:ACM,2014,160-164.
[8] YANG Yang,GUO Bao,Ni Wei.Multifocus image fusion algorithm based on region statistics in contour-let domain[J].Journal of Xi′an Jiaotong University,2007,41(4):448-452. 杨NE742,郭宝,倪伟.基于区域特性的Contourlet域多聚焦图像融合算法[J].西安交通大学学报,2007,41(4):448-452.
[9] WU Yi-quan,YIN Jun,ZHU Li.Medical image fusion based on shearlet transform and total variation model[J].Journal of Data Acquisition & Processing,2013,28(5):566-571. 吴一全,殷骏,朱丽.基于Shearlet变换和TV模型的医学图像融合[J].数据采集与处理,2013,28(5):566-571.
[10] WANG Feng,CHENG Yong-mei.Image fusion method ba-sed on multi-scale non-local mean filter and shear direction filter[J].Control and Decision,2017,32(12):2183-2189. 王峰,程咏梅.基于MNLMF和SF方向滤波的图像融合算法[J].控制与决策,2017,32(12):2183-2189.
[11] da Cunha A L,Zhou J P,Do M N.The nonsubsampled contourlet transform:theory,design,and applications[J].IEEE Transactions on Image Processing,2006,15(10):3089-3101.
[12] CHAO Rui,ZHANG Ke,LI Yan-jun.An image fusion algorithm using wavelet transform[J].Acta Electronica Sinic,2004,32(5):750-753.晁锐,张科,李言俊.一种基于小波变换的图像融合算法[J].电子学报,2004,32(5):750- 753.
[13] Li Mei-li,Li Yan-jun,Wang Hong-mei.Multi-focus image fusion algorithm based on Contourlet transform[J].Computer Engineering and Applications,2009,45(10):20-22. 李美丽,李言俊,王红梅.基于Contourlet变换的多聚焦图像融合新算法[J].计算机工程与应用,2009,45(10):20-22.
[14] HUO Guan-ying,LI Qing-wu,SHI Dan.Multi-sensor image fusion algorithm considering neighborhood consi-stency in the nonsubsampled contourlet transform domain[J].Journal of Xi Dian University,2010,37(4):770-776. 霍冠英,李庆武,石丹.一种邻域一致性的NSCT域多传感器图像融合算法[J].西安电子科技大学学报,2010,37(4):770- 776.
[15] TONG Tao,YANG Guang,TAN Hai-feng,et al.Multi-sensor image fusion algorithm based on NSCT[J].Geogra-phy and Geo-Information Science,2013,29(2):22-25. 童涛,杨桄,谭海峰,等.基于NSCT变换的多传感器图像融合算法[J].地理与地理信息科学,2013,29(2):22-25.
[2] Geng P, Huang M, Liu S, et al.Multifocus image fusion method of Ripplet transform based on cycle spin-ing[J].Multimedia Tools and Applications,2016,75(17):10583-10593.
[3] Ancuti C O,Ancuti C,De V C,et al.Single-scale fusion:an effective approach to merging images[J].IEEE Transactions on Image Processing,2017,26(1):65-78.
[4] ZHU Pan,LIU Ze-yang,HUANG Zhan-hua.Infrared polarization and intensity image fusion based on dual-tree complex wavelet transform and sparse representation[J].Acta Photonica Sinica,2017,46(12):1210002. 朱攀,刘泽阳,黄战华.基于DTCWT和稀疏表示的红外偏振与光强图像融合[J].光子学报,2017,46(12):1210002.
[5] Donoho D L,Duncan M R.Digital curvelet transform:strategy,implementation and experiments[C].Proc.of SPIE.San Jose:SPIE,2000,12-30.
[6] WU Peng,LI Wen-lin,QI De-yu,et al.An image fusion algorithm based on threshold-improved contourlet transform[J].Journal of South China University of Technology(Natural Science Edition) ,2017,45(1):35-41. 吴鹏,李雯霖,齐德昱,等.基于阈值改进Contourlet变换的图像融合算法[J].华南理工大学学报(自然科学版),2017,45(1):35-41.
[7] Ikuta C,Zhang S,Uwate Y,et al.A novel fusion algorithm for visible and infrared image using non-subsampled contourlet transform and pulse-coupled neural network[A].The 9th International Conference on Computer VisionTheory and Applications[C].Lisbon,Portugal:ACM,2014,160-164.
[8] YANG Yang,GUO Bao,Ni Wei.Multifocus image fusion algorithm based on region statistics in contour-let domain[J].Journal of Xi′an Jiaotong University,2007,41(4):448-452. 杨NE742,郭宝,倪伟.基于区域特性的Contourlet域多聚焦图像融合算法[J].西安交通大学学报,2007,41(4):448-452.
[9] WU Yi-quan,YIN Jun,ZHU Li.Medical image fusion based on shearlet transform and total variation model[J].Journal of Data Acquisition & Processing,2013,28(5):566-571. 吴一全,殷骏,朱丽.基于Shearlet变换和TV模型的医学图像融合[J].数据采集与处理,2013,28(5):566-571.
[10] WANG Feng,CHENG Yong-mei.Image fusion method ba-sed on multi-scale non-local mean filter and shear direction filter[J].Control and Decision,2017,32(12):2183-2189. 王峰,程咏梅.基于MNLMF和SF方向滤波的图像融合算法[J].控制与决策,2017,32(12):2183-2189.
[11] da Cunha A L,Zhou J P,Do M N.The nonsubsampled contourlet transform:theory,design,and applications[J].IEEE Transactions on Image Processing,2006,15(10):3089-3101.
[12] CHAO Rui,ZHANG Ke,LI Yan-jun.An image fusion algorithm using wavelet transform[J].Acta Electronica Sinic,2004,32(5):750-753.晁锐,张科,李言俊.一种基于小波变换的图像融合算法[J].电子学报,2004,32(5):750- 753.
[13] Li Mei-li,Li Yan-jun,Wang Hong-mei.Multi-focus image fusion algorithm based on Contourlet transform[J].Computer Engineering and Applications,2009,45(10):20-22. 李美丽,李言俊,王红梅.基于Contourlet变换的多聚焦图像融合新算法[J].计算机工程与应用,2009,45(10):20-22.
[14] HUO Guan-ying,LI Qing-wu,SHI Dan.Multi-sensor image fusion algorithm considering neighborhood consi-stency in the nonsubsampled contourlet transform domain[J].Journal of Xi Dian University,2010,37(4):770-776. 霍冠英,李庆武,石丹.一种邻域一致性的NSCT域多传感器图像融合算法[J].西安电子科技大学学报,2010,37(4):770- 776.
[15] TONG Tao,YANG Guang,TAN Hai-feng,et al.Multi-sensor image fusion algorithm based on NSCT[J].Geogra-phy and Geo-Information Science,2013,29(2):22-25. 童涛,杨桄,谭海峰,等.基于NSCT变换的多传感器图像融合算法[J].地理与地理信息科学,2013,29(2):22-25.