Comparison of few-view CT image reconstruction algorithms by constrained total-variation minimization based on different sampling bases

The aim of this paper is to investigate how the choice of the sampling basis affects the reconstruction results of constrained total-variation (TV) minimization algorithms used for few-view computed tomography (CT). The reconstruction results between the algorithm based on frequency samples (AFS) and the algorithm based on projection samples (APS) are compared using numerical simulated data. Under same conditions, AFS exhibits better results compared with APS. These experimental results confirm that provided same conditions, the more incoherence between the basis for sampling Psi and the basis of the sparse signal Phi, the better reconstruction results of the constrained TV minimization algorithm can be achieved and better algorithms can be developed with the basis pairs (Phi, Psi).