Real-time Fabric Defect Detection Technology

Because of the variety and complexity of fabric defects, the performance of the current fabric defect detection system is not good enough. The key point of real-time fabric defect detection system is how to improve the accuracy of fabric defect detection and classification.This paper presents a multi-channel Gabor filter scheme for fabric defect detection. A textile image is processed by using the Gabor-filters with multi-scale and multi-orientation and hereby form multi-images. Then, the images are reconstructed into one image for detecting defects. The experimental result shows that the algorithm is robust and computationally efficient. The overall detection rate is 95%.To classify the defects effectively, a novel learning method based on BP is employed. Experiments shows that the proposed approach efficiently improves the result of classifying and well fulfills the users requirement. In this paper, we also integrate SVM into the learning process of defect classification. Experimental results show that SVM approach achieves better performance than other techniques based on machine vision.