The Study Of Incomplete Projection Reconstruction Algorithm Based On Sparse Representation

Computed tomography (CT) imaging technique is to scan the object under differentangles with using a certain image reconstruction algorithm to reconstruct the image from theobtained projection data. While in practical application, due to the scanning equipment, theobject shapes or the limitation of radiation dose, incomplete projection data may occursometimes. The iterative reconstruction algorithm has a certain advantage for the incompleteprojection data with applying the priori knowledge and without being limited by the scanningpath and so on. In order to better realize the iterative reconstruction, this subject mainly studythe reconstruction of the incomplete projection data in CT image reconstruction based on thesparse representation theory. Major research content is as follows:(1) This paper focuses on the analysis and introduces of the problem of TV-minimizeconstraints based on the CS theory which can effectively solve the incomplete projection dataimage reconstruction problem. This article improved the existing TV algorithm. Thisalgorithm is verified to perform feasibly for the Head model in the situation of limited-angleprojection data. The simulation experiment shows that this algorithm has higher quality andhigher signal-to-noise ratio. Finally, the real image was reconstructed from the real projectiondata and satisfactory results were obtained.(2) Then detailed introduce of the sparse representation theory and dictionary learning(DL) method of study is given. This article combines traditional ART algorithm with thedictionary learning method, and then the ART-DL algorithm is given. The proposed algorithmwas applied to the sparse-angle projection data in CT image reconstruction. In addition, thispaper sets up the size of little image block and the sliding distance parameters in the processof the dictionary construction, the results show that these parameters have effects on the CTreconstruction images. Then the simulation projection data experiment was performed and compared with traditional algorithm. Finally the satisfactory results were obtained from realprojection data. The results of the simulation and the real projection data experiment in thesituation of sparse-angle projection data verify the feasibility and effectiveness of thecombination of algebra iteration method and dictionary learning algorithm in the CT imagereconstruction...