Λεπτομέρειες βιβλιογραφικής εγγραφής
| Τίτλος: |
Piecewise data-driven tight frame and ℓ 0 -balanced image denoising method. |
| Συγγραφείς: |
Du, Wenjian, Li, Jia, Zhou, Qiwen |
| Πηγή: |
Inverse Problems & Imaging; Jun2026, Vol. 22, p1-27, 27p |
| Θεματικοί όροι: |
Image denoising, Sparse approximations, Signal-to-noise ratio, Iterative methods (Mathematics), Image segmentation |
| Περίληψη: |
Data-driven tight frames are widely used for sparse representation in imaging tasks. However, conventional methods that train the tight frame globally across the entire image often suffer from interference between regions due to feature variations, which degrades representation quality and data processing performance. In this paper, we propose a piecewise data-driven tight frame(DDTF) based approach to mitigate the interference between regions. Based on simple segmentation, we generate a set of tight frames in each region respectively.Furthermore, existing methods typically train the tight frame on degraded images, resulting in frames that sparsely represent the degraded image rather than the realunderlying image. To overcome this limitation, we propose an $ \ell_0 $-balanced DDTF image denoising model, in which sparsity and regularity of the redundant frame coefficients are balanced. We employ an iterative algorithm that alternates between refining the tight frame and adjusting the sparse coefficients, utilizing information from both the degraded image and the reconstructed image. We prove that the whole sequence generated by our proposed algorithm converges to a stationary point of the $ \ell_0 $-balanced DDTF model.We apply piecewise DDTF and the above iterative scheme to image denoising and demonstrate its effectiveness through numerical experiments. The results indicate that our approach enhances the quality of the reconstructed images, and for images with piecewise textures, it outperforms several representative nonlocal methods in terms of PSNR and the preservation of texture details. [ABSTRACT FROM AUTHOR] |
|
Copyright of Inverse Problems & Imaging is the property of American Institute of Mathematical Sciences and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) |
| Βάση Δεδομένων: |
Complementary Index |