Gait Pose Estimation Based on Manifold Learning

A manifold learning based approach for gait pose estimation is proposed in this paper. It consists of two manifold learning based dimension reductions and three mapping functions based on General Regression Neural Network (GRNN). A model of various people walking gait is built so as to find the correspondence between a new gait pose image and the model. The reduced low-dimensional data can be used to realize the mapping between 2D gait pose model and 3D body configuration. When inputting a 2D gait pose image, it can provide the corresponding pose image in the model which can be used to carry out the mapping by the trained GRNN. Simulated experiments manifested the effectiveness of the approach.

[1]  J. Tenenbaum,et al.  A global geometric framework for nonlinear dimensionality reduction. , 2000, Science.

[2]  H. Abdi,et al.  Principal component analysis , 2010 .

[3]  Ahmed M. Elgammal,et al.  Tracking People on a Torus , 2009, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[4]  Christian Wallraven,et al.  View Invariant Body Pose Estimation Based on Biased Manifold Learning , 2010, 2010 20th International Conference on Pattern Recognition.

[5]  J. Leeuw.  Applications of Convex Analysis to Multidimensional Scaling , 2000 .

[6]  Ralph Gross,et al.  The CMU Motion of Body (MoBo) Database , 2001 .

[7]  Christian Wallraven,et al.  Supervised manifold learning based on biased distance for view invariant body pose estimation , 2012, 2012 IEEE International Conference on Systems, Man, and Cybernetics (SMC).

[8]  Donald F. Specht,et al.  A general regression neural network , 1991, IEEE Trans. Neural Networks.

[9]  Ahmed M. Elgammal,et al.  Modeling View and Posture Manifolds for Tracking , 2007, 2007 IEEE 11th International Conference on Computer Vision.