摘要
在基于协同表示(CR)的高光谱图像分类问题中,通常直接选用训练样本构建各类字典,但各类字典内训练样本基元间的相关性往往会降低协同表示分类性能。为此,该文提出采用主成分分析(PCA)对各类训练样本进行去相关处理以构建字典,降低了训练样本间的相关性对分类结果的影响,可有效提高协同表示分类的有效性。高光谱真实数据分类实验结果表明,该算法可有效提高传统协同表示分类的正确率。
In traditional collaborative representation(CR) based hyperspectral image classification, the training samples are directly used to construct a dictionary for representation. However, the correlation among the training samples within a class tends to degrade the performance of collaborative representation based classification. In the paper, the principal component analysis(PCA) is used to de-correlate the training samples within a class. As a result, the influence of correlation among training samples on subsequent collaborative representation-based classification can be alleviated. Experimental results on two benchmark datasets show that the proposed algorithm can effectively improve the performance of traditional collaborative representation-based classification.
引文
[1]韩震,陈西庆,恽才兴.海洋高光谱遥感研究进展[J].海洋科学,2003,27(1):22-25.HAN Zhen,CHEN Xi-qing,YUN Cai-xing.Research progress of marine hyperspectral remote sensing[J].Marine Sciences,2003,27(1):22-25.
[2]辛勤,粘永健,万建伟,等.基于FastICA的高光谱图像压缩技术[J].电子科技大学学报,2010,39(5):711-715.XIN Qin,NIAN Yong-jian,WAN Jian-wei,et al.Compression technique for hyperspectral imagery based on FastICA[J].Journal of University of Electronic Science and Technology of China,2010,39(5):711-715.
[3]MEI S,HE M,WANG Z,et al.Spatial purity based endmember extraction for spectral mixture analysis[J].IEEETransaction on Geoscience and Remote Sensing,2010,48(9):3434-3445.
[4]张良培,张立福.高光谱遥感[M].武汉:武汉大学出版社,2005.ZHANG Liang-pei,ZHANG Li-fu.Hyperspectral remote sensing[M].Wuhan:WuHan University Publishing Press,2005.
[5]WANG M,ZHANG L,CHEN S,et al.The analysis about factors influencing the supervised classification accuracy for vegetation hyperspectral remote sensing imagery[C]//The4th International Congress on Image and Signal Processing(CISP).[S.l.]:IEEE,2011,3:1685-1689.
[6]ZHANG L,YANG M,FENG X.Sparse representation or collaborative representation:Which helps face recognition?[C]//International Conference on Computer Vision.[S.l.]:IEEE Computer Society,2011:471-478.
[7]LI W,TRAMEL E W,PRASAD S,et al.Nearest regularized subspace for hyperspectral classification[J].IEEETransactions on Geoscience and Remote Sensing,2013,52(1):477-489.
[8]李占利,崔磊磊,刘金瑄.基于协同表示的步态识别[J].计算机应用研究,2016,33(9):2878-2880.LI Zhan-li,CUI Lei-lei,LIU Jin-xuan.Gait recognition based on collaborative representation[J].Journal of Computer Applications,2016,33(9):2878-2880.
[9]TARABALKA Y,BENEDIKTSSON J A,CHANUSSOT J.Spectral-spatial classification of hyperspectral imagery based on partitional clustering techniques[J].IEEETransactions on Geoscience&Remote Sensing,2009,47(8):2973-2987.
[10]PLAZA A,BENEDIKTSSON J A,BOARDMAN J W,et al.Recent advances in techniques for hyperspectral image processing[J].Remote Sensing of Environment,2009,113(1):S110-S122.
[11]赵蔷.主成分分析方法综述[J].软件工程,2016,19(6):1-3.ZHAO Qiang.A review of principal component analysis[J].Software Engineering,2016,19(6):1-3.
[12]姜斌,潘景昌,郭强,等.PCA和相融性度量在聚类算法中的应用[J].电子科技大学学报,2007,36(6):1292-1295.JIANG Bin,PAN Jing-chang,GUO Qiang,et al.Application of PCA and coherence measure in clustering algorithm[J].Journal of University of Electronic Science and Technology of China,2007,36(6):1292-1295.