|
| 摘要: |
| 为了有效地解决由成像系统、记录设备和传输介质的不完善等因素所导致的图像退化问题, 本文提出了基于 Krylov 子空间的 LSQR 算法, 并将其与 Tikhonov 正则化技术相结合形成了一种新算法. 该方法不仅可以将高维度的问题转化为低维度问题, 还改善了图像恢复中的不适定性. 在数值模拟中, 将新算法与相关文献中所提的正则化 GMRES 方法进行了分析比较, 结果表明本文提出的方法是可行且高效的, 尤其在信噪比方面大大的提高了计算效率, 明显改善了图像恢复质量. |
| 关键词: 图像 恢复 Krylov 正则化 LSQR算法 |
| DOI： |
| 分类号:TP391.4 |
| 基金项目:国家自然科学基金(51679186); 陕西省自然科学基金(No.2019JM-284). |
|
| Hybrid regularization LSQR algorithm for image denoising |
|
Mintao,Hanxuan
|
| Abstract: |
| This paper proposes a new algorithm, which is formed by combining Tikhonov regularization technology with LSQR algorithm based on Krylov subspace, for effectively solving the image degradation caused by factors such as imaging system, recording equipment and transmission media imperfections. It can not only convert high-dimensional into low-dimensional problems, but also improve ill-posedness in image recovery. In numerical simulations, we compare and analyze the new algorithm with the regularization GMRES algorithm proposed in the related literature. The results show that the feasibility and effectiveness of the proposed algorithmt,especially in the aspect of SNR, which significantly improves the quality of image restoration. |
| Key words: image Recovery Krylov Regularization LSQR algorithm |
