U-Net Based Image Enhancement for Short-time Muon Scattering Tomography

TL;DR

The paper tackles the challenge of poor image quality in short-time MST caused by low muon flux by introducing a U-Net-based image enhancer trained on simulated PoCA images. It adds a novel Stamp-ing augmentation to bridge sim-to-real gaps, enabling effective domain adaptation for experimental MST data. Quantitatively, the method achieves substantial improvements in image quality metrics, with $SSIM$ and $LPIPS$ demonstrating strong gains on real data. This approach paves the way for faster, more practical MST deployments with reduced detector quality requirements and acquisition times.

Abstract

Muon Scattering Tomography (MST) is a promising non-invasive inspection technique, yet the practical application of short-time MST is hindered by poor image quality due to limited muon flux. To address this limitation, we propose a U-Net-based framework trained on Point of Closest Approach (PoCA) images reconstructed with simulation MST data to enhance image quality. When applied to experimental MST data, the framework significantly improves image quality, increasing the Structural Similarity Index Measure (SSIM) from 0.7232 to 0.9699 and decreasing the Learned Perceptual Image Patch Similarity (LPIPS) from 0.3604 to 0.0270. These results demonstrate that our method can effectively enhance low-statistics MST images, thereby paving the way for the practical deployment of short-time MST.

Figures (9)

• Figure 1: The structure of the U-Net in our studies, the encoder, bottleneck layer, and decoder are displayed in different blue depths
• Figure 2: Schematic of the Geant4 simulation setup.
• Figure 3: Representative PoCA images (Image-1 to 4) reconstructed from experimental data, the definition of Image-1 to 11 can be found in the Section. \ref{['sssec:det']}
• Figure 4: Representative PoCA images reconstructed from simulated data, the resolution and the event level of the simulated data corresponding to the image are displayed in the upper left.
• Figure 5: The change of a simulated data reconstructed PoCA image before and after "Stamping" augmentation.