A Novel Spatial-Spectral Sparse Representation for Hyperspectral Image Classification Based on Neighborhood Segmentation
WANG Cai-ling1, 2, WANG Hong-wei3, HU Bing-liang1, WEN Jia4, XU Jun5, LI Xiang-juan2
1. Key Lab of Spectral Imaging, Xi’an Institute of Optics and Precision Mechanics of Chinese Academy of Sciences, Xi’an 710119, China 2. School of Computer Science, Xi’an Shiyou University, Xi’an 710065, China 3. Engineering University of CAPF, Xi’an 710086, China 4. Institute of Software of Chinese Academy of Sciences, Beijing 100080, China 5. School of Information Engineering, East China Jiaotong Univeristy, Nanchang 330013, China
Abstract：Traditional hyperspectral image classification algorithms focus on spectral information application, however, with the increase of spatial resolution of hyperspectral remote sensing images, hyperspectral imaging presents clustering properties on spatial domain for the same category. It is critical for hyperspectral image classification algorithms to use spatial information in order to improve the classification accuracy. However, the marginal differences of different categories display more obviously. If it is introduced directly into the spatial-spectral sparse representation for image classification without the selection of neighborhood pixels, the classification error and the computation time will increase. This paper presents a spatial-spectral joint sparse representation classification algorithm based on neighborhood segmentation. The algorithm calculates the similarity with spectral angel in order to choose proper neighborhood pixel into spatial-spectral joint sparse representation model. With simultaneous subspace pursuit and simultaneous orthogonal matching pursuit to solve the model, the classification is determined by computing the minimum reconstruction error between testing samples and training pixels. Two typical hyperspectral images from AVIRIS and ROSIS are chosen for simulation experiment and results display that the classification accuracy of two images both improves as neighborhood segmentation threshold increasing. It concludes that neighborhood segmentation is necessary for joint sparse representation classification.
[1] Gillespie A R. Remote Sensing of Environment, 1992, 42(2): 137. [2] Theiler J, Gloce K. Proc SPIE Algorithms and Technologies for Multispectral, Hyperspectral, and Ultraspectral Imagery Ⅻ, 2006, 6233: 62330H. [3] Chen Yi, Nasrabadi Nasser M, Tran Trac D. IEEE Geoscience and Remote Sensing Letters, 2011, 8(4): 676. [4] ZHAO Chun-hui, LI Xiao-hui, ZHU Hai-feng(赵春晖，李晓慧，朱海峰). Journal of Harbin Engineering University(哈尔滨工程大学学报), 2013, 34(9): 1171. [5] Qazi Sami ul Haq，Shi Lixin, Tao Linmi，et al. Key Engineering Materials: Advanced Materials in Microwaves and Optics, 2012, 500: 675. [6] Chen Yi, Nasrabadi Nasser M, Tran Trac D. IEEE Trans. Geoscience and Remote Sensing, 2011, 49(10): 3973. [7] Zhao Chunhui, Li Xiaohui, Ren Jinchang, et al. International Journal of Remote Sensing , 2013, 34(24): 8669. [8] SONG Lin, CHENG Yong-mei, ZHAO Yong-qiang(宋 琳，程永梅，赵永强). Acta Optical Sinica(光学学报), 2012, 32(3): 0330003. [9] HU Zheng-ping, JIA Qian-wen, XU Cheng-qian(胡正平，贾千文，许成谦). Signal Processing(信号处理), 2010, 26(4): 533. [10] GAO Rui, ZHAO Rui-zhen, HU Shao-hai(高 瑞，赵瑞珍，胡绍海). Acta Optica Sinia(光学学报), 2010, 30(6): 1639. [11] SHI Guang-ming, LIU Dan-hua, GAO Da-hua，et al(石光明，刘丹华，高大化，等). Acta Electronica Sinica(电子学报), 2009, 37(5): 1070. [12] Tropp J A, Gilbert A C, Strauss M J. Signal Processing, 2006, 86(3): 572. [13] Cotter S F, Rao B D, Engan K, et al. IEEE Trans. on Signal Process, 2005, 53(7): 2477. [14] Chang Chein-I. Hyperspectral Data Exploitation: Theory and Applications. John Wiley & Sons, 2007. 379. [15] FANG Hong, YANG Hai-rong(方 红，杨海蓉). Acta Automatica Sinica(自动化学报), 2011, 37(12): 1413. [16] Ji S H, Xue Y, Carin L. IEEE Trans. On Signal Processing, 2008, 56(6): 2346. [17] Rakotomamonjy A. Signal Processing, 2011, 91(7): 1505.
