A CT Local Reconstruction Algorithm Based on Prior Image and Compressed Sensing

In order to improve the local reconstruction quality of CT images, a CT local reconstruction algorithm based on prior images and compressed sensing（CS） was proposed on the basis of CS theory. First, the filtered backprojection（FBP） algorithm was used to reconstruct the projection data in local region of interest（ROI）, and the reconstructed CT image was adopted as the prior image of iteration. Then the total variation minimization（TVM） of CT image was chosen as the criterion for the projection onto convex sets-TVM reconstruction of local ROI. Shepp-Logan model and one solid rocket motor were chosen to carry out relative experiment. Results show that the proposed algorithm has better local reconstruction quality of CT images and anti-noise performance.