Research On Regularization Model Of Image Reconstruction

Computed tomography(CT)is extensively applied to biomedical imaging and industrial non-destructive examination,agriculture and other fields because of its advantages of noninvasive and high resolution.Image reconstruction-an important stage of CT imaging,which can reconstruct the internal structure of the scanned object based on the obtained projection data.In the case that the obtained projection data is complete,compared with the analytic reconstruction algorithm,the reconstruction effect is better.But in practice,due to the limitation of scanning environment and the structure of the scanned object,complete projection data cannot be obtained.In practical situations,regularization technique is often used to introduce prior information to improve the quality of reconstructed images.The regularization reconstruction model of images is a hot topic and important content in current social research.This paper studies the iterative reconstruction algorithm with SART as the data fidelity and regular as the constraint and introduces the specific algorithm flow.At the beginning of this article,I briefly explained what the CT imaging system is mainly composed of,and then introduces several algorithms frequently used in the field of CT reconstruction,which based on Total Variation Minimization(TVM)reconstruction algorithm was introduced.The solving process of reconstruction algorithm based on TVM is extremely slow,and TV regularization is easy to cause too smooth.These widely used algorithms have more or less shortcomings,so it is extremely important to find appropriate regularization constraints.On this basis,this paper will Truncated Total Variation(Truncated TV,TTV)as the regular term,combined with SART algorithm,put forward CT reconstruction algorithm based on Truncated TV(SART-TTV algorithm).Finally,the simulation data are used for experiments.Through the comparison experiment of reconstructed images,it can be clearly seen that the image quality is better after the image is reconstructed by the algorithm proposed in this paper.Finally,Guided Image Filtering(GIF)is introduced,which can make better use of Guided Image and Image structure,transfer the important features contained in Guided Image to the input Image,and let the filtered Image retain part of the input Image information.On this basis,we propose a CT reconstruction algorithm based on guided image filtering and truncated total variation(ART-TTV-GIF algorithm).The input image is reconstructed by SART algorithm,and the guided image is the image after the base truncation TV is minimized.Then,the guided image filtering is used to solve the sparse Angle CT reconstruction problem.Experiments are carried out using simulated data with and without noise,and the experimental results clearly show that the algorithm proposed by us can reconstruct images with higher quality by comparison.Moreover,the proposed algorithm is superior to SART-TTV algorithm in some aspects.