一种无监督的层间互补性增强的环形伪影移除方法
By employing an unsupervised interlayer complementarity enhancement method, utilizing polynomial compensation and TV algorithm to optimize detector response, and combining Gaussian step convolution and online dual-domain transformation, the problems of cumbersome ring artifact removal and noise amplification in traditional methods are solved, achieving fast, universal, and efficient artifact removal results, suitable for automated image reconstruction of CT scanners.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SOUTHEAST UNIV
- Filing Date
- 2024-08-27
- Publication Date
- 2026-07-17
AI Technical Summary
Existing techniques for removing ring artifacts caused by inconsistent detector responses in computed tomography (CT) systems are cumbersome, difficult to adjust hyperparameters, and lack universality. Furthermore, traditional methods increase the manufacturing difficulty and cost of CT systems and may amplify noise in reconstructed images.
An unsupervised interlayer complementarity enhancement method is adopted, which optimizes the detector response through polynomial compensation coefficients. Combined with the Total Variation (TV) algorithm and Gaussian strided convolution, the gradient of stripe artifacts is extracted using the horizontal differential operator. Iterative optimization is performed to remove ring artifacts, and the artifact removal in the image domain is achieved through an online dual-domain transformation algorithm.
It achieves efficient removal of annular artifacts without hyperparameter tuning, is versatile, can run quickly on low-computing-power GPU devices, significantly improves image quality, and is suitable for automated image reconstruction of CT scanners.
Smart Images
Figure CN119151838B_ABST