Improving reconstructed images using hybridization between local search and harmony search meta-heuristics

Abstract

Image reconstruction from projections in tomography is an ill-posed, inverse problem. This problem is always a challenge because there are no standard methods which give satisfactory results. In this paper we propose hybridization between Local Search (LS) and Harmony Search (HS) metaheuristics to improve quality of reconstructed images. The proposed method is implemented, tested on some images and compared to LS and Filtered backprojection (FBP) methods. The preliminary results are promising and prove the efficiency of our method.