Fabric defect detection using a hybrid particle swarm optimization-gravitational search algorithm and a Gabor filter.

Area of Science:

• Textile Engineering
• Computer Vision
• Artificial Intelligence

Background:

• Growing demand for diverse textile fabrics necessitates advanced inspection techniques.
• Manual fabric defect detection is challenging due to complex patterns and varied defects.
• Automated defect detection is crucial for improving textile production efficiency.

Purpose of the Study:

• To develop a novel, efficient, and cost-effective method for fabric defect detection.
• To optimize ellipse Gabor filter (EGF) parameters using a hybrid metaheuristic algorithm.
• To enhance the accuracy and speed of fabric surface defect identification.

Main Methods:

• A hybrid Particle Swarm Optimization-Gravitational Search Algorithm (PSO-GSA) was employed for parameter optimization.
• Ellipse Gabor Filter (EGF) parameters were optimized for non-defective fabric textures.
• Defective images were processed using the optimized filter and thresholding for defect localization.

Main Results:

• The hybrid PSO-GSA demonstrated good convergence and solution characteristics for filter optimization.
• The proposed method achieved accurate detection of fabric defects on various surfaces.
• The use of a single optimized filter significantly increased inspection speed and reduced costs.

Conclusions:

• The proposed PSO-GSA and EGF hybrid method offers a robust solution for automated fabric defect inspection.
• This technique improves upon traditional visual inspection methods in terms of speed and accuracy.
• The approach is cost-effective, making it suitable for industrial textile production.