Bspline based Super-Resolution Construction of Textured Images

Abstract

Super-Resolution image construction has gained increased importance recently. This is due to the demand for resolution enhancement for many imaging applications, as it is much efficient to capture images in a low resolution environment. The Bspline mathematical functions have long been utilized for signal representation. However they have been just recently been used for signal interpolation and zooming. This is due to the fact that they are flexible and provide the best cost/quality trade off relationship. In this paper we present a super-resolution image construction algorithm, where the high frequencies and edges of the high resolution constructed image are solely based on the Bspline signal representation. Mathematical explanation and derivation for the proposed Bspline prediction is analyzed. Several texture images from the Vistex database has been used to test the proposed technique. Extensive simulation results, that have been carried out with the proposed approach on different classes of images and demonstrated its usefulness, are proposed.