Multi-segment limited-angle CT reconstruction via a BM3D filter

Authors:

Abstract
Limited-angle reconstruction is an ill-posed problem in CT reconstruction due to incomplete data in frequency domain. Multi-segments limited-angle projection is a special case in image reconstruction, which combines the properties of limited-angle and few-view projections. In this work, we use compressive sensing method to solve multi-segments limited-angle problem. A typical compressive sensing method uses total variation (TV) as a sparse representation of original signals and reconstructs the image by total variation minimization during an iterative process. We apply a BM3D filter in CT reconstruction and give a different sparse representation for compressive sensing. Numerical simulation is performed to compare the performance between ART-TV algorithm and ART-BM3D algorithm. Simulation results demonstrate the feasibility of our ART-BM3D algorithm.