Hyperspectral data compression using lasso algorithm for spectral decorrelation

Simplice A. Alissou; Ye Zhang

doi:10.1117/12.2053265

22 May 2014 Hyperspectral data compression using lasso algorithm for spectral decorrelation

Simplice A. Alissou, Ye Zhang

Author Affiliations +

Proceedings Volume 9124, Satellite Data Compression, Communications, and Processing X; 91240A (2014) https://doi.org/10.1117/12.2053265
Event: SPIE Sensing Technology + Applications, 2014, Baltimore, MD, United States

Abstract

Among discrete orthogonal transforms, Karhunen-Loeve transform (KLT) achieves the most optimal spectral decorrelation for hyperspectral data compression with minimum mean square error. A common approach for those spectral decorrelation transform techniques such as KLT is to select m coefficient using some threshold value and then treating the rest of the coefficients as zero, this will result in loss of information. In order to preserve more information on small target data, this paper focused on a new technique called joint KLT-Lasso. The Lasso was applied to KLT coefficient. Sparse loadings were obtained using the Lasso constraint on KLT regression coefficients and more coefficients were shrunk to exact zero. The goal of our new method is to introduce a limit on the sum of the absolute values of the KLT coefficients and in which some coefficients consequently become zero without using any threshold value. A simulation on different hyperspectral data showed encouraging results.

Citation Download Citation

Simplice A. Alissou and Ye Zhang "Hyperspectral data compression using lasso algorithm for spectral decorrelation", Proc. SPIE 9124, Satellite Data Compression, Communications, and Processing X, 91240A (22 May 2014); https://doi.org/10.1117/12.2053265

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

;

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE