Grid Filters for Local Nonlinear Image Restoration

Abstract

A new approach to local nonlinear image restoration is described, based on approximating functions using a regular grid of points in a many-dimensional space. Symmetry reductions and compression of the sparse grid make it feasible to work with twelve-dimensional grids as large as 22¹². Unlike polynomials and neural networks whose filtering complexity per pixel is linear in the number of filter co-efficients, grid filters have O(1) complexity per pixel. Grid filters require only a single presentation of the training samples, are numerically stable, leave unusual image features unchanged, and are a superset of order statistic filters. Results are presented for additive noise, blurring, and superresolution.