Abstract
New probabilistic restoration of noisy blurred images is presented. The optimum filter minimizing mean-squared errors of restored images is derived for images with known statistical characteristics. The Wiener filter technique utilizing power spectra can be applied to yield the optimum filter. Unlike a Kalman filter, the filter is noncausal in the sense that estimation at a point depends not only on past estimates, but also on future estimates yet to be found. The problem can be solved by an iterative method proposed in this study. It is shown that the proposed filter is effective for uniformly distributed noise as well as impulsive noise.
| Original language | English |
|---|---|
| Pages (from-to) | 40-47 |
| Number of pages | 8 |
| Journal | Systems, computers, controls |
| Volume | 15 |
| Issue number | 1 |
| Publication status | Published - 1984 Jan |
| Externally published | Yes |
ASJC Scopus subject areas
- Engineering(all)