Research On CT Image Reconstruction Technology Based On Compressive Sensing

Computer tomography CT(Computed Tomography)application in the medical field,image reconstruction algorithm are compared with the traditional iterative method and the analytical method of two.But they are not suitable for the reconstruction of incomplete projection data calculation and can not meet the increasingly high requirements of practical application.In order to make up for this defect,the theory of compressed sensing is introduced.In the lower frequency of Binet Quist sampling frequency,the purpose of shortening the sampling time is realied the random projection to reduce the sampling time,using the compressibility of the signal.Combining the compressed sensing theory with CT,it can not only greatly reduce the amount of reconstruction calculation,but also can control the cost input of signal processing.This paper introduces the traditional CT image reconstruction algorithm,mathematical theory foundation and CT imaging physical principle and studies several commonly used reconstruction algorithms,and makes systematic analysis for pixel quality and operation response time of reconstructed image.The relevant content of the theory of compression perception is described in detail,especially its function in the field of image processing.The problem of low efficiency and slow speed reflected by CT technology in reconstruction of incomplete projection data can be remedied and solved by compressed sensing theory.The original projection is sparse represented based on wavelet transform basis.Under the premise of minimizing the original L1 norm,the algorithm of algebraic iteration and image gradient total variation algorithm are combined to achieve the optimal solution of image gradient.And on this basis,with more sparse norm Lp(0<p<1)instead of norm L1,CT reconstruction,CT the image reconstruction problem so successfully is transformed into solving Lp minimization problem.The simulation experiment verifies the results and efficiency of the reconstructed image,and compares it with the traditional algorithm.In the premise of determining the accuracy of the simulation process,the result is analyzed.Compared with the traditional reconstruction method,in the Lp range,when the 0.4<p<0.6 is running,the minimum reconstruction budget is achieved,and the image quality obtained is better and the information contained is more.In this case,the reconstructed image has lower noise than the traditional reconstruction image,and is more suitable for the practical application requirements.