Study Of Image Reconstruction TV Algorithms Based On Chambolle-Pock

Computed Tomography is a kind of technology which can reconstruction the inner intersecting surface information through the projection data that derive from the outside,and it has been widely used in medical treatment and industry field.CT now is regarded as the most advanced tool of nondestructive examination.Nowadays,the primal issue of this field is that try to decrease the radiation dose to patients as much as possible,based on guaranteeing the precision of reconstruction image.But under the limitation of the traditional Nyquist sampling theorem,it is too hard to realize.The proposition of Compressive Sensing provides a new way to solve the issue above,which can reconstruct the objective signals with high precision,according to the sampling frequency much lower than Nyquist,so that the incomplete projection CT image reconstruction can be realized.TV?Total Variation?minimization algorithm is a classical image reconstruction algorithm based on CS.ASD-POCS algorithm is a kind of effective way to solve TV minimization reconstruction modeling,it is designed through optimization theory but not inferred,so that there are too many parameters for debugging,which could be hard.For that reason,we studied TV algorithm based on Chambolle-Pock?CP?frame.In this paper,we realized three CP algorithm:l₂²-TV algorithm?l₁-TV algorithm and constrained l₂-TV algorithm.We analyze how the balanced factor influenced the reconstruction precision in l₂²-TV algorithm and l₁-TV algorithm,discuss the their sensibility to noise,and made compare with traditional FBP algorithm on reconstruction precision.Finally,we make comparison about convergence rate between general algorithm and precondition algorithm.The experiments showed that three algorithms based on CP frame can realize sparse reconstruction with high precision,and l₁-TV algorithm has good performance when dealing with noisy projection data.