ICA
所属分类:matlab编程
开发工具:matlab
文件大小:541KB
下载次数:18
上传日期:2013-03-15 09:59:43
上 传 者:
T.T.S
说明: Independent Component Analysis Play a Role
in Unmixing Hyperspectral Data
(This paper studies the impact of hyperspectral source statistical dependence on ICA and IFA performances)
文件列表:
ICA (0, 2012-10-11)
ICA\ICA Does_independent_component_analysis_play_a_role_in_unmixing_hyperspectral_data.pdf (579827, 2012-07-05)
ICA\Untitled.m (2638, 2012-07-06)
ICA\demo_ica.m (10631, 2004-05-28)
ICA\dirichlet_rnd.m (1333, 2002-11-05)
ICA\fastica.m (18146, 2001-01-15)
ICA\fpica.m (26742, 2001-01-15)
ICA\pcamat.m (11747, 2001-01-15)
ICA\remmean.m (444, 2001-01-15)
ICA\signatures.mat (40352, 2004-05-12)
ICA\whitenv.m (2017, 2001-01-15)
This set of files contains the Matlab code
test FastICA algorithm describe in the paper:
José M. P. Nascimento and José M. B. Dias
"Does Independent Component Analysis Play a Role in Unmixing Hyperspectral Data?"
submited to IEEE Trans. Geosci. Remote Sensing, vol. .., no. .., pp. .-., 2004
A short version can be found in
José M. P. Nascimento and José M. B. Dias
"Does Independent Component Analysis Play a Role in Unmixing Hyperspectral Data?"
in Pattern Recognition and Image Analysis, ser. Lecture Notes in Computer Science,
F. j. Perales, A.J.C. Campilho, N. P. Blanca and A. Sanfeliu, eds.,
vol. 2652, Springer-Verlag, 2003, pp. 616-625.
Files:
readme.txt - this file
demo_ica.m - Matlab script file to start demo
dirichlet_rnd.m - Matlab function file for generation of
abundance fractions with
dirichlet distributions
signatures.mat - Matlab data file with a set of mineral signatures
extracted from USGS spectral library.
This file contains three variables:
wavlen (224 x 1) - wavelength
A (224 x 21) - 21 mineral signatures
names (21 x 29) - names of the minerals
cup_ref.mat - Matlab data file with the Cuprite, Nevada
reflectance subimage (250 x 190 pixels)
from data set acquired on the AVIRIS flight 19 June 1997
the subimage starts at line 1620 and column 420 and it
ends on line 1869 and column 610, noisy channels
{1, 2, 104...113, 148...167, 221...224} were removed.
This file contains three variables:
wavlen (224 x 1) - wavelength
BANDS (1 x 188) - selected bands
Lines - number of lines of the subimage
Columns - number of columns of the subimage
L - number of channels selected
x (188 x 47750) - subimage (channels x number of pixels)
fastica.m - fastica file (all rigths reserved)
fpica.m - fastica file
whitenv.m - fastica file
remmean.m - fastica file
pcamat.m - fastica file
(FAstICA files are a subset of the fastICA package that can be downloaded
from http://www.cis.hut.fi/projects/ica/fastica/ )
Getting started:
To run different demos, set the "demo_number" on the second line of the demo_ica.m file:
demo_number=1
demo_number=2
demo_number=3
demo_number=4
demo_number=5
this will set different simulation:
first demo the abundances are independent with unnmiform distributions.
There are 3 endmembers (p=3) mixed in the image;
second demo the abundances are Dirichlet distributed with 3 endmembers;
third demo the abundances on strips with 3 endmembers (p=3) mixed in the image;
fourth demo the abundances on strips with 3 endmembers (p=3) mixed in the image;
SNR is set to 30dB and there are ilumination fluctuations and signature variability
fifth demo real data from Cuprite (AVIRIS).
Run the demo_ica.m script Matlab file
近期下载者:
相关文件:
收藏者: