Nonnegative matrix factorization with spatial prior and reference spectra application to remote hyperspectral image understanding

This paper dealswith nonnegativematrix factorization (NMF) dedicated to unmixing of hyperspectral images (HSI). We propose several improvements to better relate the output endmember spectra to the physical properties of the input data: firstly, we introduce a regularization term which enforces the closeness of the output endmembers to automatically selected reference spectra. Secondly, we account for these reference spectra and their locations in the initialization matrices. We exemplify our methods on self-acquired HSIs. The first scene is compound of leaves at the macroscopic level. In a controlled environment, we extract the spectra of three pigments. The second scene is acquired froman airplane: we distinguish between vegetation, water, and soil.

[1]  Patrik O. Hoyer,et al.  Non-negative Matrix Factorization with Sparseness Constraints , 2004, J. Mach. Learn. Res..

[2]  John P. Kerekes,et al.  SHARE 2012: subpixel detection and unmixing experiments , 2013, Defense, Security, and Sensing.

[3]  Mireille Guillaume,et al.  HALS-based NMF with flexible constraints for hyperspectral unmixing , 2012, EURASIP J. Adv. Signal Process..

[4]  Anne-Katrin Mahlein,et al.  Recent advances in sensing plant diseases for precision crop protection , 2012, European Journal of Plant Pathology.

[5]  Hyunsoo Kim,et al.  Nonnegative Matrix Factorization Based on Alternating Nonnegativity Constrained Least Squares and Active Set Method , 2008, SIAM J. Matrix Anal. Appl..

[6]  Antonio J. Plaza,et al.  A Signal Processing Perspective on Hyperspectral Unmixing: Insights from Remote Sensing , 2014, IEEE Signal Processing Magazine.

[7]  Andrzej Cichocki,et al.  Non-Negative Matrix Factorization , 2020 .

[8]  Anne-Katrin Mahlein,et al.  Hyperspectral imaging for small-scale analysis of symptoms caused by different sugar beet diseases , 2012, Plant Methods.

[9]  Nicolas Gillis,et al.  Using underapproximations for sparse nonnegative matrix factorization , 2009, Pattern Recognit..

[10]  Paul D. Gader,et al.  A Signal Processing Perspective on Hyperspectral Unmixing , 2014 .

[11]  Sándor Lenk,et al.  Multicolor fluorescence imaging for early detection of the hypersensitive reaction to tobacco mosaic virus. , 2007, Journal of plant physiology.

[12]  Julien Marot,et al.  Lower-Rank Tensor Approximation and Multiway Filtering , 2008, SIAM J. Matrix Anal. Appl..

[13]  Andrzej Cichocki,et al.  Nonnegative Matrix and Tensor Factorization T , 2007 .

[14]  Nicolas Dobigeon,et al.  Robust nonnegative matrix factorization for nonlinear unmixing of hyperspectral images , 2013, 2013 5th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS).

[15]  José M. Bioucas-Dias,et al.  Vertex component analysis: a fast algorithm to unmix hyperspectral data , 2005, IEEE Transactions on Geoscience and Remote Sensing.

[16]  Qian Du,et al.  Investigation on constrained matrix factorization for hyperspectral image analysis , 2005, Proceedings. 2005 IEEE International Geoscience and Remote Sensing Symposium, 2005. IGARSS '05..

[17]  Julien Marot,et al.  Leaf marker spectra identification by hyperspectral image acquisition and vertex component analysis , 2013, European Workshop on Visual Information Processing (EUVIP).