Sparse Image Reconstruction via Fast ICI Based Adaptive Thresholding

In this paper we propose the algorithm for sparse signal reconstruction by introducing the fast intersection of confidence intervals (FICI) to the two-step iterative shrinkage thresholding (TwIST) algorithm. The performance of sparse reconstruction algorithms which are based on the iterative shrinkage is often highly dependent on selection of the proper shrinkage (threshold) parameter, and this is why such state-of-the-art algorithms often implement some technique to vary the iterative step; the simplest one is to start with the relative high parameter value, and decrease it in each iteration. In order to attack this problem, we employ the FICI method in order to adaptively calculate the threshold value in each iteration of the TwIST algorithm. The performance of the proposed algorithm has been tested on three grey-scale images, and the results show that the proposed algorithm runs competitively with the state-of-the-art algorithms.