English

On Cartesian line sampling with anisotropic total variation regularization

Information Theory 2016-02-09 v1 math.IT Numerical Analysis

Abstract

This paper considers the use of the anisotropic total variation seminorm to recover a two dimensional vector xCN×Nx\in \mathbb{C}^{N\times N} from its partial Fourier coefficients, sampled along Cartesian lines. We prove that if (xk,jxk1,j)k,j(x_{k,j} - x_{k-1,j})_{k,j} has at most s1s_1 nonzero coefficients in each column and (xk,jxk,j1)k,j(x_{k,j} - x_{k,j-1})_{k,j} has at most s2s_2 nonzero coefficients in each row, then, up to multiplication by log\log factors, one can exactly recover xx by sampling along s1s_1 horizontal lines of its Fourier coefficients and along s2s_2 vertical lines of its Fourier coefficients. Finally, unlike standard compressed sensing estimates, the log\log factors involved are dependent on the separation distance between the nonzero entries in each row/column of the gradient of xx and not on N2N^2, the ambient dimension of xx.

Cite

@article{arxiv.1602.02415,
  title  = {On Cartesian line sampling with anisotropic total variation regularization},
  author = {Clarice Poon},
  journal= {arXiv preprint arXiv:1602.02415},
  year   = {2016}
}
R2 v1 2026-06-22T12:45:03.230Z