الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 119 |  |  |
| | | from | 119 | | |
Abstract
We introduced a fabric defect segmentation method based on a Simplified Pulse
Coupled Neural Network (SPCNN). Firstly, the quality of the fabric images is improved
using contrast image enhancement techniques, including Histogram Equalization (HE),
fuzzy technique and super resolution (SR). Secondly, the enhanced fabric images are
segmented using SPCNN to detect the defective regions. Finally, the SPCNN
performance is tested on FI-1662 grayscale dataset and compared to Entropy Otsu
automatic threshold, Gray Level Co-occurrence Matrix (GLCM) and Watershed methods.
The proposed SPCNN method achieves segmentation accuracy up to 99.67 %, in finding
the defective regions.