快中子层析成像与锥束几何三维重建算法研究OA

Neutron tomography uses the characteristics of neutron-transmitted objects to reconstruct images based on neutron transmission intensity information and analyze internal structures.This paper focused on the reconstruction of neutron tomography images with three-dimensional cone-beam geometry.Fast neutron transmission data were obtained through particle transport program simulation.Image reconstruction was performed using the filtered back-projection algorithm(FDK),simultaneous algebraic reconstruction(SART),ordered subset simultaneous algebraic reconstruction(OSSART),simultaneous iterative reconstruction(SIRT),and least squares QR decomposition(LSQR)algorithms.The image reconstruction effects of various algorithms with different numbers of projection angles were compared and analyzed using indicators such as the structural similarity index(SSIM).The results showed that under different numbers of projection images,the reconstruction efficiency of the FDK algorithm was significantly better than that of the iterative algorithm;in sparse projection scenes,the comprehensive performance of the SIRT algorithm was relatively good.The study laid a foundation for the subsequent optimization of fast neutron three-dimensional reconstruction algorithms.