Scintigraphic images restoration using jointly Fourier and Wavelet domains

Abstract

The paper focus on the problem of scintigraphic images restoration. These images are often disturbed by the bad detection equipment quality. The aim is to improve regions of interest perception, to help useful information extraction and so to allow good understanding of the pathological phenomenon. The restoration is an ill-posed problem. So, inverting the the distortion model in presence of additive noise is often numerically unstable. We propose, then, a new framework based on the Fourier and the Wavelet domain, in order to benefit from the advantages of each one. The Fourier regularized deconvolution exploits the Fourier representation efficiency of the colored noise. Whereas the wavelet packets denoising exploits the wavelet representation efficiency and the good localization of inherent noise in this domain. We noticed the performance of the proposed method in terms of edges preservation, contrast and uniformity in the images.