CT Image Reconstruction Algorithm With Incomplete Projection For Polymer Bonded Explosive Material Testing

Computed tomography(CT)is widely used for defect detection of polymer bonded explosive(PBX)with the advantages of non-destructive testing and three-dimensional internal structure imaging.After CT system acquires complete projections by scanning PBX,filtered backprojection(FBP)algorithm and simultaneous algebraic reconstruction technique(SART)algorithm can reconstruct high-quality CT images.When testing PBX,due to various factors,it is impossible to collect complete projection data.This paper studies the reconstruction of high-quality CT images with incomplete projection data of PBX,the main work is to study suitable image reconstruction algorithms The main contents are as follows:Firstly,according to the shape,characteristics and size of the PBX internal defects,select the micro-focus CT system for inspection.There may be initial defects such as small cracks,bubbles and residual pores inside the PBX,and the scale of these defects is in the order of hundreds of microns to millimeters.The micro-focus CT system has a resolution of 1-10?m in this experiment,which can accurately detect and analyze the initial defects of the PBX.Secondly,to improve noise and serious artifacts in the reconstructed image by the traditional image reconstruction algorithm with incomplete projection data for PBX.the compressed sensing theory(CS),total variation(TV)model and SART-TV algorithm are studied.The SART-TV algorithm incorporates prior knowledge that the CT image is sparse or nearly sparse in the gradient domain.The algorithm model is solved using the fastest gradient descent method.The experimental results show that the SART-TV algorithm is better at suppressing noise and reducing artifacts than the SART algorithm.Thirdly,to further improve the image quality reconstructed from the incomplete projection data of PBX and reduce the influence of blocky artifacts in the reconstructed image of the SART-TV algorithm,non-local mean(NLM)algorithm,non-local variation(NLTV)model and SART-NLTV algorithm are studied.The SART-NLTV algorithm uses non-local ideas to improve the TV model,and the Split-Bregman method is used to solve the algorithm model.The experimental results show that the SART-NLTV algorithm can reconstruct images better than the SART-TV algorithm,has stronger denoising ability,and has certain solutions to block artifacts.Finally,to reduce the influence of shadow artifacts in the reconstructed image with limited-angle projections for PBX,and further improve the image quality of sparse angle projection data reconstruction,the prior image constrained compressed sensing(PICCS)algorithm is studied.The PICCS algorithm introduces the prior image information,the fastest gradient descent method is used to solve the algorithm model.The reconstruction result of the algorithm will be affected by the quality of the prior image.Given a high-quality prior image,the PICCS algorithm can reconstruct a better image than SART-TV and SART-NLTV.It also has a better suppression effect on shadow artifacts.