Correction of ring artifacts with Swin-Conv-U-Net for x-ray computed tomography

Abstract

X-ray tomography is widely used in diverse research domains owing to its capacity for observing high-resolution three-dimensional structures nondestructively. However, ring artifacts appear in the tomographic reconstruction because of the inconsistent response of detector pixels, seriously affecting the image quality and causing nonuniform bias. To solve this problem, a method for correction of ring artifacts based on Swin-Conv-U-Net is proposed for x-ray tomography. When applied to simulation and experimental data, the proposed method achieves high accuracy and strong robustness and shows advantages over several classical methods in quantitative and qualitative evaluation.