Abstract
Pre-filtering is necessary before the edge detection from images corrupted by additive noise. However, even if nonlinear filtering degrade the image edges and details. We have already proposed the method of edge detection from images that corrupted by mixed noise. This edge detection consists of fuzzy rules and realizes edge detection and noise attenuation at same time. However, the proposed method which is processed with one-size mask, has the limits in the quantity of noise. In this paper, a novel method of edge detection with multi-size masks in order to detect of edges and details from images that corrupted by high probability impulse noise is proposed. The proposed method combines the large mask process for high probability impulse noise reduction and the small mask process for detection of image details by means of fuzzy rules.
| Translated title of the contribution | An edge detection with multi-scale processing from images corrupted by high probability impulse noise |
|---|---|
| Original language | Japanese |
| Pages (from-to) | 33 - 40 |
| Journal | Technical report of IEICE. DSP |
| Volume | 98 |
| Issue number | 80 |
| State | Published - 29 May 1998 |