Nonlinear Decomposition of High Spectral Mixture Pixel Based on Approximate Hapke Model and Single Physical Volume Method

Author

Liu, Jia ; Yao, Guoqing

Author_Institution

Sch. of Inf. Eng., China Univ. of Geosci. (Beijing), Beijing, China

fYear

2012

fDate

11-13 Aug. 2012

Firstpage

393

Lastpage

396

Abstract

Traditional spectrum image analysis method stays in pixel-level image analysis, so its analysis results are not accurate enough. Hyper-spectrum image has more spectrum bands, thus making it possible to analyze image in sub-pixel level. The method of hyper-spectrum data analysis contains linear model and nonlinear model. This article combines Approximate Hapke Model and single physical volume method (H-SPVM) for mix-pixel analyzing. We use H-SPVM on analyzing data from labs, and error rate of end-member extraction fell in 1%, error rate of mix-pixel analysis fell in 3%.

Keywords

data analysis; geophysical image processing; remote sensing; H-SPVM; approximate Hapke model; end-member extraction error rate; high spectral mixture pixel; hyper spectrum remote sensing method; hyper-spectrum data analysis; hyper-spectrum image; linear model; mix-pixel analysis; nonlinear decomposition; nonlinear model; pixel-level image analysis; single physical volume method; spectrum bands; spectrum image analysis method; Analytical models; Data mining; Educational institutions; Minerals; Reflectivity; Remote sensing; Spectral analysis; approximate Hapke model; nonlinear; single physical volume; spectrum analysis;

fLanguage

English

Publisher

ieee

Conference_Titel

Computer Science & Service System (CSSS), 2012 International Conference on

Conference_Location

Nanjing

Print_ISBN

978-1-4673-0721-5

Type

conf

DOI

10.1109/CSSS.2012.105

Filename

6394343