A wavelet technique applied to estimate image shifts and enhance image resolution during processing of a sequence of shifting digital frames

Abstract

The problem of high-resolution image reconstruction from noisy frames that are received by a photodetector array and shifted with respect to each other is considered. A wavelet technique is proposed for solution of this problem. The technique uses a 2D wavelet decomposition as a parametric model of the brightness function of an image in the plane of the photodetector array. A recurrent-iteration algorithm is developed for combined estimation of variable frame shifts and of the coefficients of the wavelet docomposition corresponding to the doubly improved resolution. It is shown that, under various conditions, the algorithm provides for estimation of shifts with an error no greater than 0.06 of the array pitch and for estimation of an observed-object’s brightness function with a resolution twice as high as the input-frame resolution.