基于多尺度变换的运动图像融合与图像融合工具开发
|
摘要
图像融合是以图像数据为对象的信息融合,运用图像处理和计算机技术,将初始图像数据加以分析,最大限度地提取各个初始图像数据的有利信息,并将这些有利信息汇集到一幅图像中去。本论文对基于多尺度变换的运动图像融合进行研究。论文完成的主要工作如下：
(1)提出了一种基于NSCT变换的图像融合算法,利用NSCT变换的多尺度、多方向和平移不变性以及其对图像中轮廓和细节信息的精确表示,设计了低频子带系数采用加权差分盒维数融合规则,各带通方向子带系数采用局部空间频率取大的融合规则,考虑了低频子带整体对融合效果的影响,充分保留了低频子带中的图像纹理细节信息,同时又突出了带通方向子带局部轮廓特征。通过大量对比实验表明了算法具有良好的融合效果。
(2)提出了一种基于UDCT变换的图像融合算法。UDCT变换结合了基于快速傅里叶变换(FFT)算法的曲波变换和基于滤波器组的轮廓波变换的思想。在设计融合规则时,低频子带的融合采用加权差分盒维数,高频子带的融合采用区域模值取大,在图像稀疏表示和系数融合上都具有很大优势。通过对比多种融合算法,表明了本文所提算法具有很好的融合效果。
(3)设计开发了基于多尺度变换的图像融合工具。完成了工具界面设计、多尺度变换方法模块、高低频融合规则模块、融合效果评价模块四个部分的具体实现,并且基于标准的测试方法对工具的正确性和功能性进行了测试,系统具有良好的可扩展性。
本文研究了多尺度变换方法和图像融合算法,实验表明本文所提的两种算法对标准多聚焦图像的融合、空间运动目标图像的融合和多源传感器图像的融合均具有很好的性能。为了方便进行融合实验和融合效果评价,本文还设计开发了基于多尺度变换的图像融合工具。
Image fusion is a kind of information fusion which is operating on image. By using image processing and computer technology, image fusion can extract image feature information from original images maximally, and then generate a high-quality fused image. The main contributions can be summarized as the follows:
1. Proposed an image fusion algorithm based on NSCT. In the algorithm, NSCT transform is first used to decompose source images at each scale and direction to get low-pass sub-band coefficients and band-pass directional sub-band coefficients. Then, the fusion rule of weighted box-counting dimension is adopted in low-pass sub-band, as well as the fusion rule of local space frequency in band-pass directional sub-band. Finally, the NSCT inverse transform is employed to get the fused image. Through check experiment, our algorithm is proved to be simple and effective.
2. Proposed a multi-focus image fusion algorithm based on UDCT (Uniform Discrete of Curvelet Transform). The UDCT is a new mathematical and computational tool for multi-resolution data representation which is also able to provide the higher approximation precision of geometric shape and better sparsity representation ability. Firstly, source images are decomposed at each scale and direction by UDCT to get The UDCT sub-band coefficients. Then, the fusion rule of weighted box-counting dimension in low-pass bands and window-based maximum modulus in directional high-pass bands is adopted. Finally, the fusion image is obtained by using inverse UDCT. The contrast test shows that the proposed algorithm is simple and effective.
3. Designed and developed a multi-scale transform image fusion tool including the tool interface design, multi-scale transformation method module, high-low-frequency fusion rule module and fusion effect evaluation module.
This thesis studies the multi-scale transform and image fusion algorithms. After a lot of experiments, the two proposed algorithms are proved that they both have good performance on the fusion of standard multi-focus image, space motion target image and multi-sensor image. At the same time, in order to facilitate the integration of experimental and fusion performance evaluation, this paper designed and developed the tool of image fusion based on multi-scale transform.
(1)提出了一种基于NSCT变换的图像融合算法,利用NSCT变换的多尺度、多方向和平移不变性以及其对图像中轮廓和细节信息的精确表示,设计了低频子带系数采用加权差分盒维数融合规则,各带通方向子带系数采用局部空间频率取大的融合规则,考虑了低频子带整体对融合效果的影响,充分保留了低频子带中的图像纹理细节信息,同时又突出了带通方向子带局部轮廓特征。通过大量对比实验表明了算法具有良好的融合效果。
(2)提出了一种基于UDCT变换的图像融合算法。UDCT变换结合了基于快速傅里叶变换(FFT)算法的曲波变换和基于滤波器组的轮廓波变换的思想。在设计融合规则时,低频子带的融合采用加权差分盒维数,高频子带的融合采用区域模值取大,在图像稀疏表示和系数融合上都具有很大优势。通过对比多种融合算法,表明了本文所提算法具有很好的融合效果。
(3)设计开发了基于多尺度变换的图像融合工具。完成了工具界面设计、多尺度变换方法模块、高低频融合规则模块、融合效果评价模块四个部分的具体实现,并且基于标准的测试方法对工具的正确性和功能性进行了测试,系统具有良好的可扩展性。
本文研究了多尺度变换方法和图像融合算法,实验表明本文所提的两种算法对标准多聚焦图像的融合、空间运动目标图像的融合和多源传感器图像的融合均具有很好的性能。为了方便进行融合实验和融合效果评价,本文还设计开发了基于多尺度变换的图像融合工具。
Image fusion is a kind of information fusion which is operating on image. By using image processing and computer technology, image fusion can extract image feature information from original images maximally, and then generate a high-quality fused image. The main contributions can be summarized as the follows:
1. Proposed an image fusion algorithm based on NSCT. In the algorithm, NSCT transform is first used to decompose source images at each scale and direction to get low-pass sub-band coefficients and band-pass directional sub-band coefficients. Then, the fusion rule of weighted box-counting dimension is adopted in low-pass sub-band, as well as the fusion rule of local space frequency in band-pass directional sub-band. Finally, the NSCT inverse transform is employed to get the fused image. Through check experiment, our algorithm is proved to be simple and effective.
2. Proposed a multi-focus image fusion algorithm based on UDCT (Uniform Discrete of Curvelet Transform). The UDCT is a new mathematical and computational tool for multi-resolution data representation which is also able to provide the higher approximation precision of geometric shape and better sparsity representation ability. Firstly, source images are decomposed at each scale and direction by UDCT to get The UDCT sub-band coefficients. Then, the fusion rule of weighted box-counting dimension in low-pass bands and window-based maximum modulus in directional high-pass bands is adopted. Finally, the fusion image is obtained by using inverse UDCT. The contrast test shows that the proposed algorithm is simple and effective.
3. Designed and developed a multi-scale transform image fusion tool including the tool interface design, multi-scale transformation method module, high-low-frequency fusion rule module and fusion effect evaluation module.
This thesis studies the multi-scale transform and image fusion algorithms. After a lot of experiments, the two proposed algorithms are proved that they both have good performance on the fusion of standard multi-focus image, space motion target image and multi-sensor image. At the same time, in order to facilitate the integration of experimental and fusion performance evaluation, this paper designed and developed the tool of image fusion based on multi-scale transform.
引文
[1]杨露菁 余华.多源信息融合理论与应用.[M].北京：北京邮电大学出版社.2006
[2]潘泉,于昕,程咏梅等.信息融合理论的基本方法与进展[J].自动化学报,2003,29(4)：599-615.
[3]陈浩.基于多尺度变换的多源图像融合技术研究[D].中国科学院研究生院(长春光学精密机械与物理研究所),2010.
[4]孙野.医学图像融合算法的研究[D].吉林大学,2011.
[5]陈博.遥感图像融合及应用研究[D].中国科学技术大学,2009.
[6]Davy Sannen, Hendrik Van Brussel. A multilevel information fusion approach for visual quality inspection[J].Information Fusion,2012,13(1):48-59.
[7]Bin Yang, Shutao Li. Pixel-level image fusion with simultaneous orthogonal matching pursuit[J].Information Fusion,2012,13(1):10-19.
[8]叶传奇.基于多尺度分解的多传感器图像融合算法研究[D].西安电子科技大学,2009.
[9]沈洁.基于多分辨率分析的图像融合技术研究[D].扬州大学,2009.
[10]Dwijesh Dutta Majumder, Dipankar Ray. Approaches of Multimodal Medical Images Registration and Fusion:Efficacy on Diagnostic and Therapeutic Planning[J].IETE Journal of Research,2011,57(6):498-514.
[11]王洪锋,周磊,单甘霖等.国外军事信息融合理论与应用的研究进展[J].电光与控制,2007,14(4)：13-18.
[12]Abdullah S. Alghamdi. Evaluating Defense Architecture Frameworks for C4I System Using Analytic Hierarchy Process [J].Journal of Computer Science,2012,5(12):32-41.
[13]Yan-Li Liu,Jin Wang,Xi Chen. A Robust and Fast Non-Local Means Algorithm for Image Denoising[J].2008,23(2):270-279.
[14]Zhiqiang Li, Tao Fang, Hong Huo et al. Color conspicuity map based on wavelet low-pass pyramid for popping out contours of salient objects[J].Optical Engineering,2010,49(5):050502-1-050502-3.
[15]Zhong Guo-sheng, Ao Li-ping, Zhao Kui. Influence of explosion parameters on wavelet packet frequency band energy distribution of blast vibration[J]. 2012,19(9):2674-2680.
[16]Duncan D.-Y. Po, Minh N. Do. Directional multiscale modeling of images using the contourlet transform [C].2006 IEEE Workshop on Statistical Signal Processing.2006,15(6):1610-1620.
[17]Da Cunha, A.L., Zhou, J.,Do, M.N. et al. The Nonsubsampled Contourlet Transform:Theory, Design, and Applications[J].IEEE Transactions on Image Processing,2006,15(10):3089-3101.
[18]Truong T. Nguyen, Herve Chauris. Uniform Discrete Curvelet Transform[J].IEEE Transactions on Signal Processing,2010,58(7):3618-3634.
[19]覃征,鲍复民,李爱国等.多传感器图像融合及其应用综述[J].微电子学与计算机,2004,21(2)：1-5.
[20]黄敏敏.基于小波分析的图像去噪、图像融合的应用研究[D].电子科技大学,2009.
[21]张强.基于多尺度几何分析的多传感器图像融合研究[D].西安电子科技大学，2008.
[22]Zhang, W.-F., Yan, H. Exon prediction using empirical mode decomposition and Fourier transform of structural profiles of DNA sequences[J].Pattern Recognition, 2012,45(3):947-955.
[23]胡钢,刘哲,高瑞,徐小平.基于小波变换的自适应图像融合算法[J].西安理工大学学报,2007(3)：286-290.
[24]曾杰,龚声蓉,刘纯平等.一种新的基于小波变换的多聚焦图像融合算法[J].计算机工程与应用,2007,43(24)：47-50,90.
[25]李建林,俞建成,孙胜利等.像素级的图像融合方法[J].红外,2007,28(11)：9-13,47.
[26]Yi Zhou, Mohammed Omar. Pixel-level fusion for infrared and visible acquisitions [J].International Journal of Optomechatronics,2009,3(1):41-53.
[27]李玲玲.像素级图像融合方法与应用.[M].2006年6月第1版.兰州：甘肃人民出版社,2006.
[28]Yong Xu, David Zhang. Represent and fuse bimodal biometric images at the feature level:complex-matrix-based fusion scheme[J].Optical Engineering,2010, 49(3):1-6.
[29]Kyoungro Yoon, Jonghyung Lee, Min-Uk Kim et al. Music recommendation system using emotion triggering low-level features[J].IEEE Transactions on Consumer Electronics,2012,58(2):612-618.
[30]邓海波.多分辨率图像融合及其实现[D].兰州大学,2010.
[31]Ming Qian, Aguilar, M., Zachery, K.N. et al. Decision-Level Fusion of EEG and Pupil Features for Single-Trial Visual Detection Analysis[J].IEEE Transactions on Biomedical Engineering,2009,56(7):1929-1937.
[32]许占伟.基于小波变换的CT/MRI图像融合技术[D].中国科学院研究生院(长春光学精密机械与物理研究所),2010.
[33]张良,邵琳.图像融合在高光谱遥感数据处理中的应用[J].计算机与数字工程,2010(2)：118-120
[34]李光鑫,王珂,张立保等.加权多分辨率图像融合的快速算法[J].中国图象图形学报,2005,10(12)：1529-1536.
[35]喻怀义,朱谷昌,杨自安等.ASTER多光谱波段与SPOT全色波段融合方法研究[J].矿产与地质,2008,22(1)：69-73.
[36]吕超峰.图像融合算法研究及DSP实现[D].西北工业大学,2007.
[37]Kim, YG, Song, YJ, Chang, UD et al. Face recognition using a fusion method based on bidirectional 2DPCA[J].Applied Mathematics and Computation,2008, 205(2):601-607.
[38]Se-Hwan Yun, Jin Heon Kim, Suki Kim et al. Image enhancement using a fusion framework of histogram equalization and laplacian pyramid[J].IEEE Transactions on Consumer Electronics,2010,56(4):2763-2771.
[39]Bo Gu, Wujing Li, Jiangtao Wong et al. Gradient field multi-exposure images fusion for high dynamic range image visualization[J].Journal of visual communication; image representation,2012,23(4):604-610.
[40]李建林,俞建成,孙胜利等.基于梯度金字塔图像融合的研究[J].科学技术与工程,2007,7(22)：5818-5822.
[41]Yankui sun, Yong Chen, Hao Feng et al. Two-dimensional stationary dyadic wavelet transform, decimated dyadic discrete wavelet transform and the face recognition application [J].International Journal of Wavelets, Multiresolution and Information Processing,2011,9(3):397-416.
[42]崇元,徐晓刚.基于BEMD与NMF的多源遥感图像融合[J].计算机工程,2012,38(23)：224-226.
[43]王海晖,彭嘉雄,吴巍等.评价多传感器图像融合效果方法的比较[J].红外与激光工程,2004,33(2)：189-1 93.
[44]田宝玉.工程信息论.[M].北京：北京邮电大学出版社,2004.
[45]Xydeas CS.,Petrovic V..Objective image fusion performance measure[J].Electronics Letters,2000,36(4):308-309.
[46]G. Qu, D. Zhang, P. Yan, Information measure for performance of image fusion, Electronics Letters 2001,38 (7):313-315.
[47]Li, J, Du, Q, Sun, CX et al. An improved box-counting method for image fractal dimension estimation [J].Pattern Recognition,2009,42(11):2460-2469.
[48]赵海英,杨光俊,徐正光等.图像分形维数计算方法的比较[J].计算机系统应用,2011,20(3)：238-241,246.
[49]Parul shah, S. N. Merchant, U. B. Desai et al. Fusion of surveillance images in infrared and visible band using curvelet, wavelet and wavelet packet transform [J].International Journal of Wavelets, Multiresolution and Information Processing,2010,8(2):271-292.
[50]敬照亮MATLAB教程与应用.[M].清华大学出版社,2011.
[2]潘泉,于昕,程咏梅等.信息融合理论的基本方法与进展[J].自动化学报,2003,29(4)：599-615.
[3]陈浩.基于多尺度变换的多源图像融合技术研究[D].中国科学院研究生院(长春光学精密机械与物理研究所),2010.
[4]孙野.医学图像融合算法的研究[D].吉林大学,2011.
[5]陈博.遥感图像融合及应用研究[D].中国科学技术大学,2009.
[6]Davy Sannen, Hendrik Van Brussel. A multilevel information fusion approach for visual quality inspection[J].Information Fusion,2012,13(1):48-59.
[7]Bin Yang, Shutao Li. Pixel-level image fusion with simultaneous orthogonal matching pursuit[J].Information Fusion,2012,13(1):10-19.
[8]叶传奇.基于多尺度分解的多传感器图像融合算法研究[D].西安电子科技大学,2009.
[9]沈洁.基于多分辨率分析的图像融合技术研究[D].扬州大学,2009.
[10]Dwijesh Dutta Majumder, Dipankar Ray. Approaches of Multimodal Medical Images Registration and Fusion:Efficacy on Diagnostic and Therapeutic Planning[J].IETE Journal of Research,2011,57(6):498-514.
[11]王洪锋,周磊,单甘霖等.国外军事信息融合理论与应用的研究进展[J].电光与控制,2007,14(4)：13-18.
[12]Abdullah S. Alghamdi. Evaluating Defense Architecture Frameworks for C4I System Using Analytic Hierarchy Process [J].Journal of Computer Science,2012,5(12):32-41.
[13]Yan-Li Liu,Jin Wang,Xi Chen. A Robust and Fast Non-Local Means Algorithm for Image Denoising[J].2008,23(2):270-279.
[14]Zhiqiang Li, Tao Fang, Hong Huo et al. Color conspicuity map based on wavelet low-pass pyramid for popping out contours of salient objects[J].Optical Engineering,2010,49(5):050502-1-050502-3.
[15]Zhong Guo-sheng, Ao Li-ping, Zhao Kui. Influence of explosion parameters on wavelet packet frequency band energy distribution of blast vibration[J]. 2012,19(9):2674-2680.
[16]Duncan D.-Y. Po, Minh N. Do. Directional multiscale modeling of images using the contourlet transform [C].2006 IEEE Workshop on Statistical Signal Processing.2006,15(6):1610-1620.
[17]Da Cunha, A.L., Zhou, J.,Do, M.N. et al. The Nonsubsampled Contourlet Transform:Theory, Design, and Applications[J].IEEE Transactions on Image Processing,2006,15(10):3089-3101.
[18]Truong T. Nguyen, Herve Chauris. Uniform Discrete Curvelet Transform[J].IEEE Transactions on Signal Processing,2010,58(7):3618-3634.
[19]覃征,鲍复民,李爱国等.多传感器图像融合及其应用综述[J].微电子学与计算机,2004,21(2)：1-5.
[20]黄敏敏.基于小波分析的图像去噪、图像融合的应用研究[D].电子科技大学,2009.
[21]张强.基于多尺度几何分析的多传感器图像融合研究[D].西安电子科技大学，2008.
[22]Zhang, W.-F., Yan, H. Exon prediction using empirical mode decomposition and Fourier transform of structural profiles of DNA sequences[J].Pattern Recognition, 2012,45(3):947-955.
[23]胡钢,刘哲,高瑞,徐小平.基于小波变换的自适应图像融合算法[J].西安理工大学学报,2007(3)：286-290.
[24]曾杰,龚声蓉,刘纯平等.一种新的基于小波变换的多聚焦图像融合算法[J].计算机工程与应用,2007,43(24)：47-50,90.
[25]李建林,俞建成,孙胜利等.像素级的图像融合方法[J].红外,2007,28(11)：9-13,47.
[26]Yi Zhou, Mohammed Omar. Pixel-level fusion for infrared and visible acquisitions [J].International Journal of Optomechatronics,2009,3(1):41-53.
[27]李玲玲.像素级图像融合方法与应用.[M].2006年6月第1版.兰州：甘肃人民出版社,2006.
[28]Yong Xu, David Zhang. Represent and fuse bimodal biometric images at the feature level:complex-matrix-based fusion scheme[J].Optical Engineering,2010, 49(3):1-6.
[29]Kyoungro Yoon, Jonghyung Lee, Min-Uk Kim et al. Music recommendation system using emotion triggering low-level features[J].IEEE Transactions on Consumer Electronics,2012,58(2):612-618.
[30]邓海波.多分辨率图像融合及其实现[D].兰州大学,2010.
[31]Ming Qian, Aguilar, M., Zachery, K.N. et al. Decision-Level Fusion of EEG and Pupil Features for Single-Trial Visual Detection Analysis[J].IEEE Transactions on Biomedical Engineering,2009,56(7):1929-1937.
[32]许占伟.基于小波变换的CT/MRI图像融合技术[D].中国科学院研究生院(长春光学精密机械与物理研究所),2010.
[33]张良,邵琳.图像融合在高光谱遥感数据处理中的应用[J].计算机与数字工程,2010(2)：118-120
[34]李光鑫,王珂,张立保等.加权多分辨率图像融合的快速算法[J].中国图象图形学报,2005,10(12)：1529-1536.
[35]喻怀义,朱谷昌,杨自安等.ASTER多光谱波段与SPOT全色波段融合方法研究[J].矿产与地质,2008,22(1)：69-73.
[36]吕超峰.图像融合算法研究及DSP实现[D].西北工业大学,2007.
[37]Kim, YG, Song, YJ, Chang, UD et al. Face recognition using a fusion method based on bidirectional 2DPCA[J].Applied Mathematics and Computation,2008, 205(2):601-607.
[38]Se-Hwan Yun, Jin Heon Kim, Suki Kim et al. Image enhancement using a fusion framework of histogram equalization and laplacian pyramid[J].IEEE Transactions on Consumer Electronics,2010,56(4):2763-2771.
[39]Bo Gu, Wujing Li, Jiangtao Wong et al. Gradient field multi-exposure images fusion for high dynamic range image visualization[J].Journal of visual communication; image representation,2012,23(4):604-610.
[40]李建林,俞建成,孙胜利等.基于梯度金字塔图像融合的研究[J].科学技术与工程,2007,7(22)：5818-5822.
[41]Yankui sun, Yong Chen, Hao Feng et al. Two-dimensional stationary dyadic wavelet transform, decimated dyadic discrete wavelet transform and the face recognition application [J].International Journal of Wavelets, Multiresolution and Information Processing,2011,9(3):397-416.
[42]崇元,徐晓刚.基于BEMD与NMF的多源遥感图像融合[J].计算机工程,2012,38(23)：224-226.
[43]王海晖,彭嘉雄,吴巍等.评价多传感器图像融合效果方法的比较[J].红外与激光工程,2004,33(2)：189-1 93.
[44]田宝玉.工程信息论.[M].北京：北京邮电大学出版社,2004.
[45]Xydeas CS.,Petrovic V..Objective image fusion performance measure[J].Electronics Letters,2000,36(4):308-309.
[46]G. Qu, D. Zhang, P. Yan, Information measure for performance of image fusion, Electronics Letters 2001,38 (7):313-315.
[47]Li, J, Du, Q, Sun, CX et al. An improved box-counting method for image fractal dimension estimation [J].Pattern Recognition,2009,42(11):2460-2469.
[48]赵海英,杨光俊,徐正光等.图像分形维数计算方法的比较[J].计算机系统应用,2011,20(3)：238-241,246.
[49]Parul shah, S. N. Merchant, U. B. Desai et al. Fusion of surveillance images in infrared and visible band using curvelet, wavelet and wavelet packet transform [J].International Journal of Wavelets, Multiresolution and Information Processing,2010,8(2):271-292.
[50]敬照亮MATLAB教程与应用.[M].清华大学出版社,2011.