Reconstruction of object contours using directional wavelet transform /
The purpose of this paper is to introduce an edge detector using directional wavelet transform. Directional wavelet transform decomposes an image into four-dimensional space which augments the image by the scale and directional information. We show the directional information significantly improves...
Shranjeno v:
| Main Authors: | , |
|---|---|
| Format: | Book Chapter |
| Jezik: | English |
| Teme: | |
| Sorodne knjige/članki: | Vsebovano v:
WSEAS transactions on computers |
| Oznake: |
Označite
Brez oznak, prvi označite!
|
| Izvleček: | The purpose of this paper is to introduce an edge detector using directional wavelet transform. Directional wavelet transform decomposes an image into four-dimensional space which augments the image by the scale and directional information. We show the directional information significantly improves edge detection in noisy images in comparison with the classical multiscale edge detector, and enables a reliable reconstruction of object contours. Detection is based on the modulus maximum principle with adaptive threshold whose value is calculated via maximum entropy measure. The method was tested on 100 synthetic noisy images with signal-to-noise ratios (SNRs) ranging from 50 dB down to 3 dB. Its reliability, robustness and acuracy were assessed by the mean absolute distance (MAD) measure. The obtained MAD results averaged across all SNRs are O.70 +/- O.O6 pixels (mean +/- std). The method' s superiority clearly predominates over the classical multiscale edge detection for the SNRs below 12 dB. |
|---|---|
| Fizični opis: | str. 1305-1312. |
| ISSN: | 1109-2750 |