mrthat / 2dpcal1-s Goto Github PK

Comparsion of five dimensionality reduction algorithms in face
classification and reconstruction. The five algorithms are: PCA, PCA-L1,
2DPCA, 2DPCA-L1, 2DPCAL1-S. Classifier is chosen to be Nearest Neighbor(NN).

--Reference--
Haixian Wang and Jing Wang, “2DPCA with L1-norm for
simultaneously robust and sparse modelling,” Neural Networks, vol.
46, no. 0, pp. 190–198, 2013.

--Usage-- 
1. Download Feret and AR face database from the web. 
2. Load them in Matlab. The data format should be height*width*nSub.
All images should be listed in original order. Resize images if necessary. 
3. Run demo_classification.m to do classification. 
4. Run demo_add_noise.m, then run demo_reconstruction.m to do reconstruction.

--Contact information-- 
Jing Wang
[email protected] 
[email protected]

2013-6-15 20:17:43