Study On Fabric Defect Detection And Sutomati Grad-ing System

Fabric defect detection system is the method based on machine vision instead of arti-ficial detection. It can adapt people's improving requirements of higher labor productivity and textile technology development, which can overcome the artificial detection's short-comings of un-detection, mistakenly inspection and low speed. So, the exploratory re-search of this project can be of some value in technology and market.Double line-scan cameras match with encoder to acquire fabric image, including fab-ric defect image. Then quality of fabric image mosaic was discussed from feature point matching and image fusion of overlap region. The character of different image mosaic methods was analyzed based on experiment, and it was successfully realized the image mosaic which was acquired by two high resolution cameras. Generalized Gaussian Distri-bution Model was set up from acquired image after preprocessing. Fabric defect was sepa-rately recognized by two methods which ware image retrieval method based on image fea-ture database and BP neural network base on Contourlet transform. The experimental re-sults demonstrate that the recognition rate of two methods was 57.67% and 94.33% respec-tively.Using Access2007 database in order to accomplish automatic scoring on fabric defect and automated grading of fabric, based on correctly recognize the fabric defect. The method may provide some referential materials for production in industrial scale.