A Novel Method for Knitted Fabric Defect Classification Using Image Processing and Weighted Voting Classifiers

This research proposes an application of the image processing device for the detection of knitted fabrics.The machine rolls up knitted fabrics and then the fabrics moves through a detection area, where the camera is connected to the image processing device.Next, an image captured after a defect exists is sent to be processed and analyzed by the NI myRIO device.An area pixel is computed based on Hue, Saturation, and Lightness (HSL) system and interpreted for defect selection.This procedure is performed on region of interest (ROI) to verify the defect.Three defects, examined in this research, consist of small holes, loose threads, and crook knitting needles.These knitted fabric defects are classified from the proposed weighted voting classifier, the results were found that the defects of three knitted fabrics can be accurately predicted.The proposed method can be available to the textile industry.