Sparse Approximation of Videos by Adaptive Linear Splines over Anisotropic Tetrahedralizations

摘要

We discuss theoretical aspects of sparse representations for videos by linear splines over anisotropic tetrahedralizations. In previous work, we have proposed a locally adaptive algorithm, adaptive thinning, for sparse approximation of images. Moreover, we have analyzed the asymptotic behaviour of N-term image approximations by linear splines over anisotropic Delaunay triangulations. In this paper, we generalize our previous results from image approximation to video approximation, i.e., from the approximation of bivariate to trivariate target functions.