Abstract
A fast speckle suppression algorithm for synthetic aperture radar(SAR) images was proposed. Firstly, using split augmented Lagrange shrinkage method, the total variation regularization denoising model was transformed into a two-parameter model. The alternating iteration method was introduced to decompose the two-parameter model into two single-parameter models with non-quadratic fidelity term and non-differentiable regularity term. Secondly, Newton iteration method was used to solve the model with non-quadratic fidelity term, and dual method was used to solve the model with non-differentiable regularity term. The fast iterative shrinkage-thresholding algorithm(FISTA) was used to optimize the speed of dual method. Experiments on "GF 3" image and TerraSAR-X image of a certain area in Wuhan show that the proposed algorithm can suppress SAR image speckle better, which improves the efficiency by about two times compared with the total variation regularization method.
| Translated title of the contribution | A fast SAR image speckle suppression algorithm |
|---|---|
| Original language | Chinese |
| Pages (from-to) | 662-667 |
| Number of pages | 6 |
| Journal | Zhongguo Kuangye Daxue Xuebao/Journal of China University of Mining and Technology |
| Volume | 48 |
| Issue number | 3 |
| State | Published - 1 Jun 2019 |
Funding
国家自然科学基金项目(41501504);辽宁省自然科学基金项目(20180551209);辽宁省教育厅一般项目(LJYL011)
Indexed by
- CSCD
- EI
Fingerprint
Dive into the research topics of 'A fast SAR image speckle suppression algorithm'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver