Euler’s Elastica Strategy for Limited-Angle Computed Tomography Image Reconstruction

Limited-angle computed tomography (CT) image reconstruction is a noteworthy reconstruction problem in which image prior models are crucial. In this paper, we aimed to develop an appropriate regularization-based method to suppress the aliasing artifacts and improve the image quality for limited-angle CT image reconstruction. Given the observation that the isophotes of reconstructed images in specific missing directions, which are not tangent to the available projection rays, are usually broken and have large curvatures; a curvature-driven Euler's elastica regularization is introduced to rectify large curvatures and keep the isophotes smooth without erratic distortions. The developed approach is evaluated using numerical simulation and real CT data, and is compared with total variation, and total generalized variation-based reconstructions. Both qualitative and quantitative evaluation results demonstrate that a prospective performance can be obtained using the proposed Euler's elastica-based method for limited-angle tomography. This method has excellent aliasing artifact suppression and edge preservation.