The CT Image Reconstruction Algorithm Based On Improved K-SVD Dictionary Learning

In the field of CT image reconstruction,many researches on reconstruction algorithms have made great achievements.The theory of Compressed Sensing is a new way to study the problem of incomplete angle reconstruction.It can recover the original signal with only a small number of sample points.And the K-SVD dictionary learning as one of the typical algorithms of image sparseness,It can improve the matching degree of sparse signal and original image by sparse decomposition and update dictionary.Firstly,the CT image reconstruction,K-SVD dictionary learning algorithm and several image reconstruction algorithms are expounded in this paper,and it also gives simulation experiment results and objective evaluation analysis.Secondly,considering that the adaptive dictionary construction by K-SVD algorithm is also related to the selection of initialization dictionary,so this paper introduces a new initialization dictionary matrix to make the K-SVD dictionary learning algorithm have a better expression ability,and make the image corresprond to the image reconstruction algorithm.Finally,considering the structural similarity of the image blocks,we make the similar blocks are clustered by NLM algorithm,and trian each class of image blocks respectively by K-SVD algorithm.We combined the improved K-SVD algorithm with the SART algorithm to reconstruct more ideal photograph.