Abstract
A Non-negativity and Support Constrains Recursive Inverse Filtering (NAS-RIF) algorithm based on support estimation and noise removal is proposed. To deal with the noise sensitivity of NAS-RIF wavelet-based denoising is introduced. Self-estimation on image support based on image segmentation is used to change the square support limitation. Scale factor is introduced to construct a new cost function for blind image restoration and the golden section method is used to expedite the convergence in iterative computation. At last simulation is done to various SNR images and backgrounds and the results show that the improved algorithm still works well under the condition of low SNR reserving the detailed information in the image as well as suppressing the noise.
| Original language | English |
|---|---|
| Pages (from-to) | 18-22 |
| Number of pages | 5 |
| Journal | Yi Qi Yi Biao Xue Bao/Chinese Journal of Scientific Instrument |
| Volume | 30 |
| Issue number | SUPPL. |
| Publication status | Published - Jun 2009 |
Keywords
- Golden section
- NAS-RIF blind image restoration
- Support estimation