Tree based reconstruction from sparse samples of signals

Abstract

The study in sparse representation of signals and its applications has changed the field of signal processing into a quickly developing area. These studies have opened tremendous feasible research ideas. Compressed sensing is one among the greatest beneficial field of these studies. Signal sparsity property introduced the idea of reconstruction of signals from very few prototype atoms of an overcomplete dictionary. The novel approach in dictionary learning process is K-SVD method. We propose a method to improve the sparse representation of signal by incorporating A∗ search algorithm in K-SVD. The usage of tree data structure has provided a lot of advantages including the introduction of auxiliary function, tree pruning techniques etc. Experimental results show that the proposed system provides more efficient and improved results than the conventional ones.