Feasibility and optimization study of a two-dimensional density reconstruction method for large-object muography

Cosmic muon imaging (muography) has gained popularity in recent years for density reconstruction in various applications. However, evaluations of image quality in reconstructed images have been lacking. This work addresses this gap by applying muon transmission imaging to a teaching building on a university campus. Multiple impacts on reconstruction image quality, including filtering methods, total counts, building thickness thresholds, and reconstruction formulas, were integrally studied. The area under the curve (AUC) of the receiver operating characteristic analysis and mean squared error (MSE) were calculated to quantitatively evaluate and optimize these parameters. Based on the simulation using the Geant4 toolkit, the recommended parameters for the building model were median filtering, a total count of 5 x 104 in the building region, a thickness threshold of 3 m, and either the constant energy loss method or cubic polynomial form fitting method. Implementing these parameters in the experimental data analysis, the reconstructed image well revealed the inner structure of the teaching building from the view angle of the detector. The optimization of the twodimensional (2D) reconstruction method facilitates the detection of hidden cavities or regions with abnormal density in large-scale objects.