高光谱图像异常检测算法研究
|
摘要
高光谱图像是新型的遥感数据,其良好的光谱诊断能力使得它非常适合对照自然背景发现人工目标。其中,异常检测算法能够在没有先验光谱信息的情况下检测到与周围环境存在光谱差异的目标,具有较强的实用性,成为了目标检测领域的一个研究热点。本文在深入分析高光谱图像数据特点的基础上,针对高光谱图像异常检测中面临的高数据维、非线性信息提取、同物异谱、混合像元等问题,做了以下几方面的研究:
首先,在研究高光谱图像数据降维技术的基础上,提出了一种基于二代曲波变化和脉冲耦合神经网络(PCNN)的自适应波段融合降维算法。该方法以高光谱图像自适应子空间分解后的各子空间为处理单元,通过对子空间内各波段图像二代曲波变换后的粗尺度系数的熵加权融合和细尺度系数的PCNN选取,实现子空间内图像的融合。将融合后的图像用于异常检测,不仅极大地降低了高光谱图像的数据量,而且能够有效地提取图像的细节信息。
其次,通过对核函数方法理论的研究,提出了一种基于核函数的特征空间加权RX异常检测算法,以利用核函数性质有效提取图像波段间隐含的非线性信息。该算法在图像的高维特征空间进行目标的异常检测,并依据背景协方差矩阵中各光谱向量到质心的距离对协方差矩阵进行自适应加权,削弱了协方差矩阵中异常数据的比重,从而使背景分布更加符合实际。另外,在利用核函数性质进行特征空间的内积运算转化时线性组合所构造的光谱核函数和径向基核函数,减弱了同物异谱现象引起的能量差异对检测精度的影响。
最后,在分析线性混合模型理论的基础上,提出了一种基于背景误差数据的高光谱图像非线性异常检测算法。它通过分块快速端元提取方法得到背景端元后,利用光谱解混技术将背景数据从高光谱图像中分离出来,然后将包含丰富目标信息的背景误差数据映射到图像的高维特征空间进行异常检测。该算法抑制了光谱混合现象带来的背景干扰,而且通过有效地利用高光谱图像的非线性信息进一步提升了异常检测的性能。
Hyperspectral imagery is a new type of remote sensing data. Its high spectral resolution makes it suitable for the detection of human-made targets surrounded by natural environment background. An anomaly detector can enable one to detect targets whose signatures are spectrally distinct from their surroundings with no prior knowledge, so it is practicable in real sences, and then it becomes a focus in the field of target detection. Based on the analysis of characteristics of hyperspectral imagery, the dissertation does some researches as following in order to solve the difficulties in anomaly detection, such as high dimensionality, nonlinear feature extraction, spectral variety, mixed pixels.
Firstly, dimension reduction methods are studied and a new adaptive band fusion algorithm based on the second generation curvelet transform and pulse-coupled neural networks (PCNN) is proposed to solve the problems caused by the high dimensions of hyperspectral imagery. After the whole data space is divided into several subspaces by adaptive subapace decomposition method, every subspace is considered an independent processing unit and curvelet transform is performed in every unit. Then the coarse scale coefficients from curvelet transform are weighted fused based on the entropy of every band image and the fine scale coefficients are selected intelligently by PCNN. Finally the fused coefficients are reconstructed to obtain the fusion image of every subspace by inverse curvelet transform. Therefore, the way that uses the fused images to detect targets can not only reduce the data greatly but also extract the detail information of hyperspectral imagery effectively.
Secondly, kernel methods are used to extract the nonlinear information of hyperspectral imagery. According to the study of the theory of kernel methods, a new kernel weighted RX algorithm is proposed for anomaly detection. The algorithm is implemented in the feature space of original hyperspectral data, and for the purpose of decreasing the portion of anomaly pixels in the background covariance matrix, each pixel in the covariance matrix is given weight by its distance to the data center. In addition, when the dot products in the high dimensional feature space are converted into the kernel computation in the low dimensional input space, the new spectral kernel function and the radial basis kernel function are mixed to conquer the difficulty brought by the spectral variety.
Finally, a nonlinear anomaly detection algorithm based on the background error data is proposed on the basis of the analysis of linear mixed model. After the background endmembers are gained by the proposed fast endmember extraction method, spectral unmixing technique is applied to all mixed spectral pixels for the purpose to separate target information from complicated background information. Then the error data that include abundant target information were transformed into a high dimensional feature space to finish the target detection. This way the serious background interferences brought by the spectral mixing was overcame, and the exploitation of nonlinear information in hyperspectral imagery greatly improved the performance of the proposed algorithm.
首先,在研究高光谱图像数据降维技术的基础上,提出了一种基于二代曲波变化和脉冲耦合神经网络(PCNN)的自适应波段融合降维算法。该方法以高光谱图像自适应子空间分解后的各子空间为处理单元,通过对子空间内各波段图像二代曲波变换后的粗尺度系数的熵加权融合和细尺度系数的PCNN选取,实现子空间内图像的融合。将融合后的图像用于异常检测,不仅极大地降低了高光谱图像的数据量,而且能够有效地提取图像的细节信息。
其次,通过对核函数方法理论的研究,提出了一种基于核函数的特征空间加权RX异常检测算法,以利用核函数性质有效提取图像波段间隐含的非线性信息。该算法在图像的高维特征空间进行目标的异常检测,并依据背景协方差矩阵中各光谱向量到质心的距离对协方差矩阵进行自适应加权,削弱了协方差矩阵中异常数据的比重,从而使背景分布更加符合实际。另外,在利用核函数性质进行特征空间的内积运算转化时线性组合所构造的光谱核函数和径向基核函数,减弱了同物异谱现象引起的能量差异对检测精度的影响。
最后,在分析线性混合模型理论的基础上,提出了一种基于背景误差数据的高光谱图像非线性异常检测算法。它通过分块快速端元提取方法得到背景端元后,利用光谱解混技术将背景数据从高光谱图像中分离出来,然后将包含丰富目标信息的背景误差数据映射到图像的高维特征空间进行异常检测。该算法抑制了光谱混合现象带来的背景干扰,而且通过有效地利用高光谱图像的非线性信息进一步提升了异常检测的性能。
Hyperspectral imagery is a new type of remote sensing data. Its high spectral resolution makes it suitable for the detection of human-made targets surrounded by natural environment background. An anomaly detector can enable one to detect targets whose signatures are spectrally distinct from their surroundings with no prior knowledge, so it is practicable in real sences, and then it becomes a focus in the field of target detection. Based on the analysis of characteristics of hyperspectral imagery, the dissertation does some researches as following in order to solve the difficulties in anomaly detection, such as high dimensionality, nonlinear feature extraction, spectral variety, mixed pixels.
Firstly, dimension reduction methods are studied and a new adaptive band fusion algorithm based on the second generation curvelet transform and pulse-coupled neural networks (PCNN) is proposed to solve the problems caused by the high dimensions of hyperspectral imagery. After the whole data space is divided into several subspaces by adaptive subapace decomposition method, every subspace is considered an independent processing unit and curvelet transform is performed in every unit. Then the coarse scale coefficients from curvelet transform are weighted fused based on the entropy of every band image and the fine scale coefficients are selected intelligently by PCNN. Finally the fused coefficients are reconstructed to obtain the fusion image of every subspace by inverse curvelet transform. Therefore, the way that uses the fused images to detect targets can not only reduce the data greatly but also extract the detail information of hyperspectral imagery effectively.
Secondly, kernel methods are used to extract the nonlinear information of hyperspectral imagery. According to the study of the theory of kernel methods, a new kernel weighted RX algorithm is proposed for anomaly detection. The algorithm is implemented in the feature space of original hyperspectral data, and for the purpose of decreasing the portion of anomaly pixels in the background covariance matrix, each pixel in the covariance matrix is given weight by its distance to the data center. In addition, when the dot products in the high dimensional feature space are converted into the kernel computation in the low dimensional input space, the new spectral kernel function and the radial basis kernel function are mixed to conquer the difficulty brought by the spectral variety.
Finally, a nonlinear anomaly detection algorithm based on the background error data is proposed on the basis of the analysis of linear mixed model. After the background endmembers are gained by the proposed fast endmember extraction method, spectral unmixing technique is applied to all mixed spectral pixels for the purpose to separate target information from complicated background information. Then the error data that include abundant target information were transformed into a high dimensional feature space to finish the target detection. This way the serious background interferences brought by the spectral mixing was overcame, and the exploitation of nonlinear information in hyperspectral imagery greatly improved the performance of the proposed algorithm.
引文
[1]张良培,张立福.高光谱遥感.武汉:武汉大学出版社.2005:58-126页
[2]陈述彭,童庆禧,郭东华.遥感信息机理研究.北京:科学出版社.1988:1-133页
[3]浦瑞良,宫鹏.高光谱遥感及其应用.高等教育出版社.2000:81-97页
[4]I.S.Reed,X.Yu.Adaptive Multiple-band CFAR Detection of An Optical Pattern with Unknown Spectral Distribution.IEEE Trans.Acoust,Speech Singal Process.1990,38(10):1760-1770 P
[5]Chein-I Chang,Shao-Shan Chiang.Anomaly Detection and Classification for Hyperspectral Imagery.IEEE Trans.Geoscience and Remote Sensing.2002,40(6):1314-1325 P
[6]Chein-I Chang,Hsuan Ren,Shao-Shan Chiang.Real-Time Processing Algorithms for Target Detection and Classification in Hyperspectral Imagery.IEEE Trans.Geoscience and Remote Sensing.2001,39(4):760-768 P
[7]Hsuan Ren,Chien-Wen Chen,Hsien-Ting Chen.Weighted Anomaly Detection for Hyperspectral Remotely Sensed Images.Proceedings of SPIE-The International Society for Optical Engineering.2005,5995:599507 P
[8]Hsuan Ren,Wang Jing.Background Whitened Target Detection Algorithm for hyperspectral imagery.Proceedings of SPIE-The International Society for Optical Engineering.2005,5806:511-518 P
[9]Dimitris Manolakis,Christina Siracusa,David mardcn,Gry Show Hyperspectral Adaptive Matched Filter Detectors:Practical Performance Comparison.SPIE 2001,4831:18-33 P
[10]Geoffrey G.Hazel.Multivariate Gaussian MRF for Multispectral Scene Segmentation and Anomaly Detection.IEEE Trans.Geoscience and Remote Sensing.2000,38(3):1199-1211 P
[11] A Schaum. Spectral Subspace Matched Filtering. SPIE 2001, 4831:1-17 P
[12]Dimitris Manolakis, Gary Show, Nirma Keshava. Comparative Analysis of Hyperspectral Adaptive Matched Filter Detectors. SPIE 2000, 4049: 2-17 P
[13] Kraut S, Scharf L L, McWhorter L T. Adaptive Subspace Detectors. IEEE Transactions on Signal Processing. 2001, 49(1): 1-16 P
[14] Manolakis D, Siracusa C, Shaw G. Adaptive Matched Subspace Detectors for Hyperspectral Imaging Applications. IEEE International Conference on Acoustics, Speech, and Signal Processing. 2001, 5(5):3153-3156P
[15]Bajorski Peter. Analytical Comparison of the Matched Filter and Orthogonal Subspace Projection Detectors in Structured Models for Hyperspectral Images. Proceedings of SPIE-The International Society for Optical Engineering. 2006, 62331: 623301 P.
[16] Chowdhury A, Alarn M.S. Target Detection In Hyperspectral Imagery Using Noise-adjusted Principal Component Analysis and Orthogonal Subspace Projection. Proceedings of SPIE-The International Society for Optical Engineering. 2006, 6365: 636510 P
[17]Heesung Kwon, Nasser M.Nasrabadi. Kernel RX-Algorithm: A Nonlinear Anomaly Detector for Hyperspectral Imagery. IEEE Trans. Geoscience and Remote Sensing. 2005, 43(2): 388-397 P
[18]Heesung Kwon, Nasser M.Nasrabadi. Kernel Adaptive Subspace Detector for Hyperspectral Imagery. IEEE Geoscience and Remote Sensing Letters. 2006, 3(2): 271-275 P
[19]Heesung Kwon, Nasser M.Nasrabadi. Kernel Matched Subspace Detectors for Hyperspectral Target Detection. IEEE Trans Pattern Analysis and Machine Intelligence. 2006, 28(2): 178-194 P
[20] Heesung Kwon, Nasser M.Nasrabadi. Kernel Orthogonal Subspace Projection for Hyperspectral Signal Classification. IEEE Geoscienc and Remote Sensing Letters. 2005, 43(12): 2952-2962 P
[21]Banerjee Arnit,Burlina Philippe,Diehl Chris.A Support Vector Method for Anomaly Detection in Hyperspectral Imagery.IEEE Transactions on Geoscience and Remote Sensing.2006,44(8):2282-2291 P
[22]Smits G F,Jordaan E M.Improved SVM Regression Using Mixtures of Kernels.Proc of the 2002 Int Joint Conf on Neural Networks,Honolulu,USA:IEEE,2002,3:2785-2790 P
[23]路威,余旭初,刘娟,杨国鹏.基于分布异常的高光谱遥感影像小目标检测算法.测绘学报.2006,35(1):40-45页
[24]田宏伟,赵慧洁.基于高阶统计的高光谱目标检测的研究.光学技术.2005,31 Suppl:223-225页
[25]李智勇.高光谱图像异常检测方法研究.国防科学技术大学博士论文,2004:47-69页
[26]李智勇,郁文贤,匡纲要.基于高维几何特性的高光谱异常检测算法研究.遥感技术与应用.2003,18(6):379-383页
[27]Gu Yanfeng,Zhang Ye,Zhang Junping.A Kernel Based Nonlinear Subspace Projection Method For Reduction of Hyperspectral Image Dimensionality.ICIP2002,Proceeding,USA.IEEE Signal Processing Society.2002,2:357-360 P
[28]Gu Yanfeng,Zhang Ye,Quan Taifan.A Kernel-based Nonlinear Subspace Projection Method For Dimensionality Reduction of Hyperspectral Image Data.Chinese Journal of Electronics.2003,12(2):203-207 P
[29]Gu Yanfeng,Zhang Ye.Hyperspectral Small Target Detection by Combining Kernel PCA with Linear Mixture Model.Chinese Journal of Electronics.2005,14(1):130-134 P
[30]Zhang Ye,Zhang Junping,Jin Ming.Adaptive Subspace Decomposition and Classification for Hyperspectral Images.Chinese Journal of Electronics.2000,9(1):82-88 P
[31]Le Moigne.J,Cole-Rhodes.A,Eastman.R..Multiple Sensor Image Registration,Image Fusion and Dimension Reduction of Earth Science Imagery.Information Fusion,Proceeding of the Fifth International Conference.2002,2(8):999-1006 P
[32]L A Zadeh.Fuzzy Sets.Information and Control.1965,8:338-353 P
[33]L A Zadeh.Fuzzy Logic.IEEE Computer.1998,4:20-26 P
[34]J A Benediktsson,Philip H.Swain.Consensus Theoretic Classification Methods.IEEE Transcations on System Man and Cybernetics.1992,22(4):688-704 P
[35]E J Candes,L Demanet,D L Donoho.Fast Discrete Curvelete Transform.Applied and Computational Mathematics.California Institute of Technology.2005.1-43 P
[36]Eckhorn R,Reitboeck H J.Fature Linking via Synchronization among Distributed Asemblies:Simulation of Results from Cat Cortex.Neural Computer,1990,2(3):293-307 P
[37]刘春红,赵春晖.基于提升算法的超谱遥感图像融合分类研究.哈尔滨工程大学学报,2004,25(6):794-798页
[38]Courant R.,Hilbert D.Methods of Mathematical Physics.New York:Interscience Publishers.1953:21-126 P
[39]John Shawe-Taylor,Nello Cristianini.Kernel Methods for Pattern Analysis.Cambridge:Cambridge University.2004:1-198 P
[40]谷延锋,刘颖,贾友华,张晔.基于光谱解译的高光谱图像异常检测算法.红外与毫米波学报.2006,25(6):473-477页
[41]D.Heinz,C-I Chang.Fully Constrained Least Squares Linear Mixture Analysis for Material Quantification in Hyperspectral Imagery.IEEE Trans.on Geosci.Remote Sensing.2001,39(3):529-545 P
[2]陈述彭,童庆禧,郭东华.遥感信息机理研究.北京:科学出版社.1988:1-133页
[3]浦瑞良,宫鹏.高光谱遥感及其应用.高等教育出版社.2000:81-97页
[4]I.S.Reed,X.Yu.Adaptive Multiple-band CFAR Detection of An Optical Pattern with Unknown Spectral Distribution.IEEE Trans.Acoust,Speech Singal Process.1990,38(10):1760-1770 P
[5]Chein-I Chang,Shao-Shan Chiang.Anomaly Detection and Classification for Hyperspectral Imagery.IEEE Trans.Geoscience and Remote Sensing.2002,40(6):1314-1325 P
[6]Chein-I Chang,Hsuan Ren,Shao-Shan Chiang.Real-Time Processing Algorithms for Target Detection and Classification in Hyperspectral Imagery.IEEE Trans.Geoscience and Remote Sensing.2001,39(4):760-768 P
[7]Hsuan Ren,Chien-Wen Chen,Hsien-Ting Chen.Weighted Anomaly Detection for Hyperspectral Remotely Sensed Images.Proceedings of SPIE-The International Society for Optical Engineering.2005,5995:599507 P
[8]Hsuan Ren,Wang Jing.Background Whitened Target Detection Algorithm for hyperspectral imagery.Proceedings of SPIE-The International Society for Optical Engineering.2005,5806:511-518 P
[9]Dimitris Manolakis,Christina Siracusa,David mardcn,Gry Show Hyperspectral Adaptive Matched Filter Detectors:Practical Performance Comparison.SPIE 2001,4831:18-33 P
[10]Geoffrey G.Hazel.Multivariate Gaussian MRF for Multispectral Scene Segmentation and Anomaly Detection.IEEE Trans.Geoscience and Remote Sensing.2000,38(3):1199-1211 P
[11] A Schaum. Spectral Subspace Matched Filtering. SPIE 2001, 4831:1-17 P
[12]Dimitris Manolakis, Gary Show, Nirma Keshava. Comparative Analysis of Hyperspectral Adaptive Matched Filter Detectors. SPIE 2000, 4049: 2-17 P
[13] Kraut S, Scharf L L, McWhorter L T. Adaptive Subspace Detectors. IEEE Transactions on Signal Processing. 2001, 49(1): 1-16 P
[14] Manolakis D, Siracusa C, Shaw G. Adaptive Matched Subspace Detectors for Hyperspectral Imaging Applications. IEEE International Conference on Acoustics, Speech, and Signal Processing. 2001, 5(5):3153-3156P
[15]Bajorski Peter. Analytical Comparison of the Matched Filter and Orthogonal Subspace Projection Detectors in Structured Models for Hyperspectral Images. Proceedings of SPIE-The International Society for Optical Engineering. 2006, 62331: 623301 P.
[16] Chowdhury A, Alarn M.S. Target Detection In Hyperspectral Imagery Using Noise-adjusted Principal Component Analysis and Orthogonal Subspace Projection. Proceedings of SPIE-The International Society for Optical Engineering. 2006, 6365: 636510 P
[17]Heesung Kwon, Nasser M.Nasrabadi. Kernel RX-Algorithm: A Nonlinear Anomaly Detector for Hyperspectral Imagery. IEEE Trans. Geoscience and Remote Sensing. 2005, 43(2): 388-397 P
[18]Heesung Kwon, Nasser M.Nasrabadi. Kernel Adaptive Subspace Detector for Hyperspectral Imagery. IEEE Geoscience and Remote Sensing Letters. 2006, 3(2): 271-275 P
[19]Heesung Kwon, Nasser M.Nasrabadi. Kernel Matched Subspace Detectors for Hyperspectral Target Detection. IEEE Trans Pattern Analysis and Machine Intelligence. 2006, 28(2): 178-194 P
[20] Heesung Kwon, Nasser M.Nasrabadi. Kernel Orthogonal Subspace Projection for Hyperspectral Signal Classification. IEEE Geoscienc and Remote Sensing Letters. 2005, 43(12): 2952-2962 P
[21]Banerjee Arnit,Burlina Philippe,Diehl Chris.A Support Vector Method for Anomaly Detection in Hyperspectral Imagery.IEEE Transactions on Geoscience and Remote Sensing.2006,44(8):2282-2291 P
[22]Smits G F,Jordaan E M.Improved SVM Regression Using Mixtures of Kernels.Proc of the 2002 Int Joint Conf on Neural Networks,Honolulu,USA:IEEE,2002,3:2785-2790 P
[23]路威,余旭初,刘娟,杨国鹏.基于分布异常的高光谱遥感影像小目标检测算法.测绘学报.2006,35(1):40-45页
[24]田宏伟,赵慧洁.基于高阶统计的高光谱目标检测的研究.光学技术.2005,31 Suppl:223-225页
[25]李智勇.高光谱图像异常检测方法研究.国防科学技术大学博士论文,2004:47-69页
[26]李智勇,郁文贤,匡纲要.基于高维几何特性的高光谱异常检测算法研究.遥感技术与应用.2003,18(6):379-383页
[27]Gu Yanfeng,Zhang Ye,Zhang Junping.A Kernel Based Nonlinear Subspace Projection Method For Reduction of Hyperspectral Image Dimensionality.ICIP2002,Proceeding,USA.IEEE Signal Processing Society.2002,2:357-360 P
[28]Gu Yanfeng,Zhang Ye,Quan Taifan.A Kernel-based Nonlinear Subspace Projection Method For Dimensionality Reduction of Hyperspectral Image Data.Chinese Journal of Electronics.2003,12(2):203-207 P
[29]Gu Yanfeng,Zhang Ye.Hyperspectral Small Target Detection by Combining Kernel PCA with Linear Mixture Model.Chinese Journal of Electronics.2005,14(1):130-134 P
[30]Zhang Ye,Zhang Junping,Jin Ming.Adaptive Subspace Decomposition and Classification for Hyperspectral Images.Chinese Journal of Electronics.2000,9(1):82-88 P
[31]Le Moigne.J,Cole-Rhodes.A,Eastman.R..Multiple Sensor Image Registration,Image Fusion and Dimension Reduction of Earth Science Imagery.Information Fusion,Proceeding of the Fifth International Conference.2002,2(8):999-1006 P
[32]L A Zadeh.Fuzzy Sets.Information and Control.1965,8:338-353 P
[33]L A Zadeh.Fuzzy Logic.IEEE Computer.1998,4:20-26 P
[34]J A Benediktsson,Philip H.Swain.Consensus Theoretic Classification Methods.IEEE Transcations on System Man and Cybernetics.1992,22(4):688-704 P
[35]E J Candes,L Demanet,D L Donoho.Fast Discrete Curvelete Transform.Applied and Computational Mathematics.California Institute of Technology.2005.1-43 P
[36]Eckhorn R,Reitboeck H J.Fature Linking via Synchronization among Distributed Asemblies:Simulation of Results from Cat Cortex.Neural Computer,1990,2(3):293-307 P
[37]刘春红,赵春晖.基于提升算法的超谱遥感图像融合分类研究.哈尔滨工程大学学报,2004,25(6):794-798页
[38]Courant R.,Hilbert D.Methods of Mathematical Physics.New York:Interscience Publishers.1953:21-126 P
[39]John Shawe-Taylor,Nello Cristianini.Kernel Methods for Pattern Analysis.Cambridge:Cambridge University.2004:1-198 P
[40]谷延锋,刘颖,贾友华,张晔.基于光谱解译的高光谱图像异常检测算法.红外与毫米波学报.2006,25(6):473-477页
[41]D.Heinz,C-I Chang.Fully Constrained Least Squares Linear Mixture Analysis for Material Quantification in Hyperspectral Imagery.IEEE Trans.on Geosci.Remote Sensing.2001,39(3):529-545 P