Research On Cosmic-ray Muon Tomography Reconstruction Algorithms And Imaging Quality Analysis

Muon tomography is a new non-destructive imaging technology for high atomic number materials with natural cosmic rays.It avoids the risk of radiation damage brought by artificial radiation sources,and can be utilized for detailed structure inspection of special nuclear materials in national defense.Natural muons have limited flux,incomplete incident angle rangeand detection noise,which degrades the imaging quality and makesthe application in detailsinspection difficult.With the aim at improving image quality of muon tomography,the dissertation studied both image reconstruction algorithms and imaging quality analysis.The research of the dissertation was of great importance for imaging noise reduction and system design optimization.First,the dissertation proposed an lp-MAP reconstruction algorithm based on edge-constrained prior functions.For the problem of large image noises,edge-sparsity prior function was added in the objective function as the penalty term,and lp norm prior with unified form was proposed.It had the advantages of each kind of priors,and low implementation cost to switch between different priors.For the problem that lp-MAP objective function had complex form and was hard to solve,surrogate of the objective function was constructed to simplify complicated optimization problems to paraboloid problems.Convex function separations were further used to transfer multivariable problems to single-variable problems,in order to achieve effective solution of lp-MAP problems.Besides,numerical experiments were performed to investigate the influence of strength parameter and norm parameter on the reconstruction results.Compared to existing methods,the proposed lp-MAP algorithm effectively suppressed noise,preserved edgesand improved spatial resolution of image.Second,the dissertation established the relationship between imaging parameters and quality with both theoretical and numerical analysis.Frequency domain analysis was first used to study the influence of system parameters on spatial resolution.The muon central slice theorem was proposed,which innovatively transferred the information of muons to frequency domain,the equivalent beam width model was established,and the analytical expression of spatial resolution was derived.Error analysis was then used to study the influence of system parameters on material resolution.The relationship between detection error and scattering angle measurement error was first established,and the statistical characteristics of muons were further used to derive the influence of detection error on reconstruction value biases.At last,numerical analysis was used to establish an empirical influence of imaging time on spatial and material resolution.Besides,performance indices and assessment methods were also proposed for muon tomography.This work filled the international research blank of muon tomography quality analysis,which provided references and basis for system optimization,quality prediction and imaging time evaluation.Last but not least,real data imaging experiments were carried out on TUMUTY imaging system,to verify the robustness of algorithms and analyze the influence of imaging time on image quality.It also provided references for real system optimization.