Determining the Intrinsic Dimension of a Hyperspectral Image Using Random Matrix Theory

Determining the intrinsic dimension of a hyperspectral image is an important step in the spectral unmixing process and under- or overestimation of this number may lead to incorrect unmixing in unsupervised methods. In this paper, we discuss a new method for determining the intrinsic dimension using recent advances in random matrix theory. This method is entirely unsupervised, free from any user-determined parameters and allows spectrally correlated noise in the data. Robustness tests are run on synthetic data, to determine how the results were affected by noise levels, noise variability, noise approximation, and spectral characteristics of the end-members. Success rates are determined for many different synthetic images, and the method is tested on two pairs of real images, namely a Cuprite scene taken from Airborne Visible InfraRed Imaging Spectrometer (AVIRIS) and SpecTIR sensors, and a Lunar Lakes scene taken from AVIRIS and Hyperion, with good results.

[1]  Amandine Robin,et al.  The effect of spectrally correlated noise on noise estimation methods for hyperspectral images , 2012, 2012 4th Workshop on Hyperspectral Image and Signal Processing (WHISPERS).

[2]  José M. Bioucas-Dias,et al.  Hyperspectral Subspace Identification , 2008, IEEE Transactions on Geoscience and Remote Sensing.

[3]  Chein-I Chang,et al.  Hyperspectral Data Exploitation , 2007 .

[4]  Noureddine El Karoui.  Spectrum estimation for large dimensional covariance matrices using random matrix theory , 2006, math/0609418.

[5]  Amandine Robin,et al.  Using Random Matrix Theory to determine the number of endmembers in a hyperspectral image , 2010, 2010 2nd Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing.

[6]  Amandine Robin,et al.  The effect of noise whitening on methods for determining the intrinsic dimension of a hyperspectral image , 2011, 2011 3rd Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS).

[7]  J. W. Silverstein,et al.  Eigenvalues of large sample covariance matrices of spiked population models , 2004, math/0408165.

[8]  Chein-I Chang,et al.  Exploration of methods for estimation of number of endmembers in hyperspectral imagery , 2006, SPIE Optics East.

[9]  Peter Bajorski.  Does virtual dimensionality work in hyperspectral images? , 2009, Defense + Commercial Sensing.

[10]  David W. Messinger,et al.  A generalized linear mixing model for hyperspectral imagery , 2008, SPIE Defense + Commercial Sensing.

[11]  José M. Bioucas-Dias,et al.  Estimation of signal subspace on hyperspectral data , 2005, SPIE Remote Sensing.

[12]  Marco Diani,et al.  A New Algorithm for Robust Estimation of the Signal Subspace in Hyperspectral Images in the Presence of Rare Signal Components , 2009, IEEE Transactions on Geoscience and Remote Sensing.

[13]  I. M. Johnstone,et al.  Detection performance of Roy's largest root test when the noise covariance matrix is arbitrary , 2011, 2011 IEEE Statistical Signal Processing Workshop (SSP).

[14]  Fred A. Kruse,et al.  Comparison of AVIRIS and Hyperion for Hyperspectral Mineral Mapping , 2002 .

[15]  David W. Messinger,et al.  Geometric estimation of the inherent dimensionality of single and multi-material clusters in hyperspectral imagery , 2009, Journal of Applied Remote Sensing.

[16]  Ho-Young Jung,et al.  Subband-based blind signal separation for noisy speech recognition , 1999 .

[17]  David Malah,et al.  Rank Estimation and Redundancy Reduction of High-Dimensional Noisy Signals With Preservation of Rare Vectors , 2007, IEEE Transactions on Signal Processing.

[18]  I. Johnstone.  On the distribution of the largest eigenvalue in principal components analysis , 2001 .

[19]  Wei Xiong,et al.  Maximum Orthogonal Subspace Projection Approach to Estimating the Number of Spectral Signal Sources in Hyperspectral Imagery , 2011, IEEE Journal of Selected Topics in Signal Processing.

[20]  S. Hook,et al.  Laboratory Reflectance Spectra of 160 Minerals, 0.4 to 2.5 Micrometers , 1992 .

[21]  José M. P. Nascimento,et al.  Signal subspace identification in hyperspectral imagery , 2012 .

[22]  B. Nadler,et al.  Determining the number of components in a factor model from limited noisy data , 2008 .

[23]  Chein-I Chang,et al.  Estimation of number of spectrally distinct signal sources in hyperspectral imagery , 2004, IEEE Transactions on Geoscience and Remote Sensing.

[24]  Boaz Nadler,et al.  Non-Parametric Detection of the Number of Signals: Hypothesis Testing and Random Matrix Theory , 2009, IEEE Transactions on Signal Processing.

[25]  Zongming Ma,et al.  Accuracy of the Tracy–Widom limits for the extreme eigenvalues in white Wishart matrices , 2012, 1203.0839.

[26]  Chein-I Chang,et al.  A New Growing Method for Simplex-Based Endmember Extraction Algorithm , 2006, IEEE Transactions on Geoscience and Remote Sensing.

[27]  F. Kruse.  Identification and mapping of minerals in drill core using hyperspectral image analysis of infrared reflectance spectra , 1996 .

[28]  R. Nalepka,et al.  Estimating the proportions of objects within a single resolution element of a multispectral scanner. , 1971 .

[29]  Azriel Rosenfeld,et al.  A Fast Parallel Algorithm for Blind Estimation of Noise Variance , 1990, IEEE Trans. Pattern Anal. Mach. Intell..

[30]  Marco Diani,et al.  Hyperspectral Signal Subspace Identification in the Presence of Rare Signal Components , 2010, IEEE Transactions on Geoscience and Remote Sensing.

[31]  S. J. Sutley,et al.  Ground-truthing AVIRIS mineral mapping at Cuprite, Nevada , 1992 .