A second-generation wavelet framework for super-resolution with noise filtering

The use of second-generation wavelets to attain super-resolution with noise filtering is described for a captured sequence of low-resolution frames without any assumptions on grid (sampling lattice) structure. The approach is based on 2D methods. The procedure allows the incorporation of the more general projective camera motion model into the framework, instead of only displacement and rotational models. Several simulations that compare the implementations of the algorithm presented here with other related approaches help illustrate the suitability of SGWs (coupled with hard or soft thresholding) in the task of image sequence superresolution with simultaneous noise filtering. (C) 2004 Wiley Periodicals, Inc.