Research On The Key Technology Of Vision-based Loom On-line Defect Detection System

Fabric defect is the main factor affecting the quality of fabric in textile production,so fabric defect detection is one of the most important parts in quality control of textile production.In the field of textile production testing,manual testing is gradually replaced by machine vision which becomes the mainstream trend.At present,the on-line defect detection system is mainly applicable to the defect detection in the fabric finishing stage,and the research on the defect detection system in the on-line production stage is still in its primary stage.Therefore,it is of great value to research a loom on-line defect detection system in practical application,and there is more realistic significance to reduce the enterprise economical cost and enhance the productivity.In order to overcome the influence of body vibration during the image acquisition of loom operation stage,a design model of the camera damper for loom is proposed.According to the actual working condition of the loom,the model was verified by multiple simulations with different coefficients.The experiment proves that the matching coefficient of camera damper can be designed to weaken the influence of the loom vibration for different working speed.Aiming at the particularity of glass fiber electronic cloth,a defect detection method based on local normalization and adaptive weighting Otsu method is proposed.First,the local normalization technique can be used to compensate the illumination of the collected image,so that the gray value of the image is uniform,and the effect of threshold selection in the subsequent defect detection is eliminated.Then,the adaptive weighting Otsu method is used to select the optimal threshold to segment the fabric image and output the defect detection image.The experimental results show that the algorithm can solve the problem of illumination unevenness in industrial scene and have good detection accuracy for a variety of fabric defect types.In order to improve the accuracy of fabric defect detection,a defect detection algorithm based on Nonsubsampled Contourlet Transform and Naive Bayesian classifier is proposed in this paper.The experimental results show that the proposed algorithm is applicable not only to textured fabrics with uniform gray texture,but also to textured color fabrics,and the detection results are not affected by fabric texture direction.