Discrete Tomography in Discrete Deconvolution: Deconvolution of Binary Images Using Ryser's Algorithm

Abstract

A new deconvolution algorithm for binary images based on the theory of discrete tomography is proposed. The proposed algorithm is inherently binary as opposed to traditional filtering techniques such as Wiener filtering which require thresholding to produce binary images. Time and space complexity of the proposed algorithm are polynomial in the image size whereas the two-dimensional Viterbi method has an exponential complexity. Application of the proposed method in equalization of two-dimensional inter-symbol interference channels such as page-oriented optical memories is demonstrated. Through numerical simulations, it is shown that the method can outperform the traditional methods such as Wiener filtering especially for low singal-to-noise scenarios.