Detection of Weft Knitting Fabric Defects Based on Windowed Texture Information And Threshold Segmentation by CNN

Methods for detecting weft knitting fabric defects are studied in this article. A new method to analyze the texture information on the fabric image with multi-window for enhancing the defects feature is introduced. The feature information of defect is segmented by Cellular Neural Network and three terms of variables are defined to represent the feature. Using interlock fabric with the defects of hole, course mark, dropped stitch and fly as experiment materials, the experiment proved the acquired feature information involved adequate information of defects with less effect of noise and the result of classification by Artificial Neural Network was well performed.