Tracking 3D hand in hierarchical latent variable space

Since the dimensionality of hands state space is too high, we employ a hierarchical Gaussian process latent variable model (GPLVM) to simultaneously learn the hierarchical latent space of hands motion and the nonlinear mapping from the hierarchical latent space to the state space of human hands. Nonlinear mappings from the hierarchical latent space to the space of hand images are constructed using radial basis function interpolation method. With these mappings, particles can be projected into hand images and measured in the images space directly. Then particle filters with fewer particles are used to track hands in the learned hierarchical low-dimensional space. Experimental results show that our proposed method can track articulated hand robustly and efficiently.