Hyperspectral data subspace dimension algorithm based on noise whitening
CHEN Jie1, 2, DU Lei1, LI Jing1, HAN Yachao1, GAO Zihong1
1. China Aero Geophysical Survey and Remote Sensing Center for Land and Resources, Beijing 100083, China; 2. Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing 100101, China
摘要高光谱影像数据的相邻波段间相关性较强,信号与噪声共存,根据最小二乘原理,使观测数据与噪声的投影误差之和最小化的HySime (hyperspectral signal identification by minimum error)算法,通过数据观测值减去噪声估计值后得到信号的估计值,进而可以计算信号相关矩阵的估计值。该算法在准确估计噪声的情况下是可行的,但实际上经光谱降维去相关后得到的各像元噪声估计值往往并不准确,因此,原始的HySime算法得到的结果可能并不理想。提出一种基于噪声白化的HySime改进算法,它不必进行逐像元的噪声去除,而是先对原始数据进行噪声白化处理,然后准确获取噪声的协方差矩阵估计值,再利用HySime算法进行信号相关矩阵计算,实现了提高算法精度的目的。通过模拟和实验数据的验证,改进的算法结果更准确稳定,与经典的NSP (noise subspace projection)算法在不同情况下所得结果有很好的一致性,通过引入噪声白化的过程,提高了算法对非白噪声的适应性。
Abstract：The correlation between adjacent bands of hyperspectral image data is relatively strong. However, signal coexists with noise. The HySime (hyperspectral signal identification by minimum error) algorithm which is based on the principle of least squares is designed to calculate the estimated noise value and the estimated signal correlation matrix value. The algorithm is effective with accurate noise value but ineffective with estimated noise value obtained from spectral dimension reduction and decorrelation process. This paper proposes an improved HySime algorithm based on noise whitening process. Instead of removing noise pixel by pixel, the algorithm carries out the noise whitening process on the original data first, obtains the noise covariance matrix estimated value accurately, and uses the HySime algorithm to calculate the signal correlation matrix value so as to improve the precision of the resultant value. Simulation and experiment have reached some conclusions: Firstly, the improved HySime algorithm is more accurate and stable than the original HySime algorithm; Secondly, the improved HySime algorithm results have better consistency under different conditions compared with the classic NSP (noise subspace the projection) algorithm; Finally, the improved HySime algorithm improves the adaptability of non-white data noise with the noise whitening process.
陈洁, 杜磊, 李京, 韩亚超, 高子弘. 基于噪声白化的高光谱数据子空间维数算法[J]. 国土资源遥感, 2017, 29(2): 60-66.
CHEN Jie, DU Lei, LI Jing, HAN Yachao, GAO Zihong. Hyperspectral data subspace dimension algorithm based on noise whitening. REMOTE SENSING FOR LAND & RESOURCES, 2017, 29(2): 60-66.
[1] Green A A,Berman M,Switzer P,et al.A transformation for ordering multispectral data in terms of image quality with implications for noise removal[J].IEEE Transactions on Geoscience and Remote Sensing,1988,26(1):65-74. [2] Lee J B,Woodyatt A S,Berman M.Enhancement of high spectral resolution remote-sensing data by a noise-adjusted principal components transform[J].IEEE Transactions on Geoscience and Remote Sensing,1990,28(3):295-304. [3] Chang C I,Du Q.Interference and noise-adjusted principal components analysis[J].IEEE Transactions on Geoscience and Remote Sensing,1999,37(5):2387-2396. [4] 张 兵,高连如.高光谱图像分类与目标探测[M].北京:科学出版社,2011. Zhang B,Gao L R.Hyperspectral Image Classification and Target Detection[M].Beijing:Science Press,2011. [5] Roger R E.Principal components transform with simple, automatic noise adjustment[J].International Journal of Remote Sensing,1996,17(14):2719-2727. [6] Roger R E,Arnold J F.Reliably estimating the noise in AVIRIS hyperspectral images[J].International Journal of Remote Sensing,1996,17(10):1951-1962. [7] 洪 波.高光谱遥感图像信噪比估算方法研究[D].北京:中国科学院大学,2013. Hong B.Study on Methods for SNR Estimation of Hyperspectral Remote Sensing Images[D].Beijing:University of Chinese Academy of Sciences,2013. [8] 高连如.高光谱遥感目标探测中的信息增强与特征提取研究[D].北京:中国科学院遥感应用研究所,2007. Gao L R.Research on Information Enhancement and Feature Extraction in Hyperspectral Remote Sensing Object Detection[D].Beijing:Institute of Remote Sensing and Digital Earth,Chinese Academy of Sciences,2007. [9] Chang C I,Du Q.Estimation of number of spectrally distinct signal sources in hyperspectral imagery[J].IEEE Transactions on Geoscience and Remote Sensing,2004,42(3):608-619. [10] Bioucas-Dias J M,Nascimento J M P.Hyperspectral subspace identification[J].IEEE Transactions on Geoscience and Remote Sensing,2008,46(8):2435-2445. [11] Eldar Y C,Oppenheim A V.MMSE whitening and subspace whitening[J].IEEE Transactions on Information Theory,2003,49(7):1846-1851. [12] Cawse K,Robin A,Sears M.The effect of noise whitening on methods for determining the intrinsic dimension of a hyperspectral image[C]//Proceedings of the 3rd Workshop on Hyperspectral Image and Signal Processing:Evolution in Remote Sensing.Lisbon,Portugal:IEEE Computer Society,2011. [13] Nascimento J M P,Dias J M B.Vertex component analysis:A fast algorithm to unmix hyperspectral data[J].IEEE Transactions on Geoscience and Remote Sensing,2005,43(4):898-910. [14] Chang C I,Du Q.Noise subspace projection approaches to determination of intrinsic dimensionality of hyperspectral imagery[C]//Proceedings of the SPIE 3871,Image and Signal Processing for Remote Sensing V.Florence,Italy:SPIE,1999:34-44. [15] Swayze G,Clark R N,Kruse F,et al.Ground-truthing AVIRIS mineral mapping at Cuprite,Nevada[C]//Summaries of the Third Annual JPL Airborne Geoscience Workshop.Denver,CO:JPL Publication,1992:47-49. [16] Swayze G A.The Hydrothermal and Structural History of the Cuprite Mining District,Southwestern Nevada:An Integrated Geological and Geophysical Approach[D].Boulder,Colorado:University of Colorado,1997. [17] Chang C I,Xiong W,Liu W M,et al.Linear spectral mixture analysis based approaches to estimation of virtual dimensionality in hyperspectral imagery[J].IEEE Transactions on Geoscience and Remote Sensing,2010,48(11):3960-3979.
2017, Vol. 29 
