| DEMO_KPCADENOIS | |
Idea of image denoising based on Kernel PCA.
Description:
The kernel PCA model is trained for to model input 2D vectors.
The free model parameters (kernel argument, dimension) are
tuned by the script train_kpca_denois. The denosing of corrupted
vectors is based on projecting onto the kernel PCA model and
take the resulting image as the reconstructed vector [Mika99b]. This
idea is demonstrated on a toy 2D data.
See also
GREEDYKPCA, KPCAREC, KPCA.
About: Statistical Pattern Recognition Toolbox
(C) 1999-2003, Written by Vojtech Franc and Vaclav Hlavac
Czech Technical University Prague
Faculty of Electrical Engineering
Center for Machine Perception
Modifications:
06-jun2004, VF