Description: S-isomap is a manifold learning algorithm, which is a supervised variant of isomap.
Reference: X. Geng, D.-C. Zhan, and Z.-H. Zhou. Supervised nonlinear dimensionality reduction for visualization and classification. IEEE Transactions on Systems, Man, and Cybernetics - Part B: Cybernetics, 2005, vol.35, no.6, pp.1098-1107.
We propose a novel approach for head tracking, which combines particle filters with isomap. The particle filter works on the low-dimensional embedding of training images. It indexes into the isomap with its state variables to find the closest template for each particle. The most weighted particle approximates the location of head. We develop a synthetic video sequence to test our technique. The results we get show that the tracker tracks the head which changes position, poses and lighting conditions.
mani: MANIfold learning demonstration GUI by Todd Wittman, Department of Mathematics, University of Minnesota E-mail wittman@math.umn.edu with comments & questions.
MANI Website: httP://www.math.umn.edu/~wittman/mani/index.html
Last Modified by GUIDE v2.5 10-Apr-2005 13:28:36
Methods obtained from various authors.
(1) MDS -- Michael Lee
(2) isomap -- J. Tenenbaum, de Silva, & Langford
(3) LLE -- Sam Roweis & Lawrence Saul
(4) Hessian LLE -- D. Donoho & C. Grimes
(5) Laplacian -- M. Belkin & P. Niyogi
(6) Diffusion Map -- R. Coifman & S. Lafon
(7) LTSA -- Zhenyue Zhang & Hongyuan Zha
