Sampling of two-dimensional signals below Nyquist density with application to computer aided tomography

The problem of sampling a two-dimensional signal at below the conventional Nyquist sampling density is considered. Although optimal sampling at below Nyquist density is possible through a process of periodic sample deletion, the resulting irregular sampling pattern has limited utility in real-world applications. In this paper, a novel algorithm is proposed to perform uniform sampling of the data at below Nyquist density, yet allows for perfect reconstruction of the signal. By exploring the special signal structure of sinograms in computer aided tomography, the technique is used for signal reconstruction from sinograms sampled at below Nyquist rate. Computer simulation results are presented to demonstrate the validity of the algorithm.>