GitHub - yuzhounh/G2DPCA_demo_2: Scripts for the paper: Generalized 2-D principal component analysis by Lp-norm for image analysis.

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 10 Commits
data		data
G2DPCA.m		G2DPCA.m
GPCA.m		GPCA.m
LICENSE		LICENSE
PCA2DL1S.m		PCA2DL1S.m
RSPCA.m		RSPCA.m
Readme		Readme
classify_G2DPCA.m		classify_G2DPCA.m
classify_GPCA.m		classify_GPCA.m
classify_PCA2DL1S.m		classify_PCA2DL1S.m
classify_RSPCA.m		classify_RSPCA.m
m1_add_noise.m		m1_add_noise.m
m1_prepare.m		m1_prepare.m
m2_task.m		m2_task.m
m3_classify_plot.m		m3_classify_plot.m
m3_reco_plot.m		m3_reco_plot.m
m3_reco_plot.png		m3_reco_plot.png
main.m		main.m
manuscript.pdf		manuscript.pdf
para_task.m		para_task.m
para_worker.m		para_worker.m
perct.m		perct.m
pnorm.m		pnorm.m
reco_G2DPCA.m		reco_G2DPCA.m
reco_GPCA.m		reco_GPCA.m
reco_PCA2DL1S.m		reco_PCA2DL1S.m
reco_RSPCA.m		reco_RSPCA.m

Repository files navigation

Generalized two dimensional principal component analysis by Lp-norm for image analysis. 
Copyright (C) 2018 Jing Wang

For 1D algorithms, the image data should be 2D matrix, nSub*(height*width). 
For 2D algorithms, the image data should be 3D matrix, height*width*nSub. 
The images are listed in the subject-by-subject manner. Please refer to the 
manuscript for more information about the experiments.

Variables:
FaceDB, Feret or ORL
x,   image data
rho, a tuning parameter in RSPCA or 2DPCAL1-S
s,   a tuning parameter in GPCA or G2DPCA
p,   a tuning parameter in GPCA or G2DPCA
nPV, number of projection vectors to be calculated
W,   projection vectors

Usage:
Run main.m to play this demo. It takes about 30 hours on a server with 40 CPUs.

Related codes:
2DPCAL1-S, https://github.com/yuzhounh/2DPCAL1-S
G2DPCA, https://github.com/yuzhounh/G2DPCA
G2DPCA_demo_1, https://github.com/yuzhounh/G2DPCA_demo_1
G2DPCA_demo_2, https://github.com/yuzhounh/G2DPCA_demo_2

Contact information:
Jing Wang
wangjing0@seu.edu.cn
yuzhounh@163.com

2018-6-13 22:58:14