Development of a real-time machine vision system for detecting defeats of cord fabrics

Automatic detection techniques based on machine vision can be used in fabric industry for quality control, which constantly pursues intelligent methods to replace human inspections of product. This work introduces the principle components of a real-time machine vision system for defeat detection of cord fabrics, which is usually a challenging task in practice. The work aims at solving some difficulties usually incurring in such kind of tasks. The design and implementation of the algorithm, software and hardware are introduced. Based on the Gabor wavelet techniques, the system can automatically detect regular texture defects. Our experiments show the proposed algorithm is favorably suited for detecting several types of cord fabric defects. The system testing has been carried in both on-line and off-line situations. The corresponding results show the system has good performance with high detection accuracy, quick response and strong robustness.