A application of vision system for the identification and defect detection on woven fabrics by using artificial neural network

In this research a computer vision system with artificial neural network is used for the identification and defect detection on several samples of the available textile´s woven fabrics. Several textural features that are extracted by the neighboring grey level dependence matrix, and the grey level run length matrix is used as input data for the network trained by the backpropagation method. Results of the experiment indicate that the system can identify three product-types: plain, twill, and sateen weaves, and it can detect several defects, including pick-broken, pick-inhomogeneity, reed-mark, and dirt, with more than 80% correct.