Textile Web Defect Inspection by Feature Analysis Method

Abstract

Nowadays in the textile industry are still used a human naked eyes to detect any kinds of defect on textile webs. The problems occurred when a human has their own limitations on different kind of perceptions in identifying a defect. In this paper, we intend to propose an object classification using a standard deviation value for classifying the defect on textile webs. First we describe the method of image segmentation that we applied in this study which is based on statistical technique. Further, we focus on the features analysis where we divide it into two phases; (1) learning phase and (2) analysis phase. Finally, we have been tested our propose algorithm into 5 (five) different types of textile webs with 500 images for each type of webs. We figure out that this method is suitable for distort and small defect as in textile webs. The details of results from our testing phases will be discussed at the end of this paper.