Silk Texture Defect Recognition System Using Computer Vision and Artificial Neural Networks

Competiveness of textile industries depends on the quality control of production. In order to minimize production cost, effort is directed towards less defectiveness and time spent on production operations. More accuracy in silk texture defect identification should be maintained so as eliminate any abnormality in the silk texture that hinders its acceptability by the consumer. In this paper, silk texture defect identification is achieved by implementing artificial neural network (ANN) technique. Methodology for feature selection that leads to high recognition rates and to simpler classification systems architectures is presented.