approximation of piecewise smooth functions and images by edge.pdf

Approximation of piecewise smooth functions and images by edge-adapted (ENO-EA) nonlinear multiresolution techniques Francesc Arandiga, Albert Cohen, Rosa Donat, Nira Dyn and Basarab Matei Abstract This paper introduces and analyzes new approximation procedures for bi- variate functions. These procedures are based on an edge-adapted nonlinear reconstruction technique which is an intrinsically two-dimensional extension of the essentially non-oscillatory and subcell resolution techniques introduced in the one dimensional setting by Harten and Osher. Edge-adapted reconstruc- tions are tailored to piecewise smooth functions with geometrically smooth edge discontinuities, and are therefore attractive for applications such as image com- pression and shock computations. The local approximation order is investigated both in Lp and in the Hausdorff distance between graphs. In particular, it is shown that for general classes of piecewise smooth functions, edge-adapted re- constructions yield multiscale representations which are optimally sparse and adaptive approximations with optimal rate of convergence, similar to curvelets decompositions for the L2 error. 1. Introduction 1.1 Background and motivation Multiscale representations of images into wavelet bases have been successfully applied in applications such as compression and denoising. In these applications, one essentially takes advantage of the sparsity of the representation f = dλ λ , i.e. the concentration of the energy of the image f on a small number of coef- ficients d . The detailed analysis of such methods, e.g. in [21], [28] and [15], λ demonstrates that their performances are directly related to a mathematical measure of the sparsity which can be