Fabric defect detection using variance and intensity analysis

Abstract

The purpose of this paper is to design a fabric defect detection system using image processing techniques. Fabric defect inspection is one of the most important processes to discard the inferior quality fabrics during the manufacture of fabric and garment in textile industry. Currently, in Sri Lankan textile industries the inspection process is carried out manually by humans and therefore subjective and prone to human error. Also, there are many other drawbacks such as tiredness, boredom and inattentiveness which cause to reduce the efficiency of industrial process. Because of this, the process has been automated with new computer technologies as an effective alternative to increase the quality of fabric. In this paper, variance analysis and intensity analysis has been used as the preceding step to identify the defects in the fabric. Because the defect-free fabric has uniformity in the structure, the occurrence of a defect in the fabric alters the regular structure. Therefore, the fabric defects can be predetermined by analyzing the parameters such as variance and intensity. To improve the efficiency of the technique and to overcome the problem of detection errors, further thresholding, noise removal, morphological operations and connected components analysis were carried out. To verify the success of the technique, it is implemented on plain fabric samples with different colors containing five common defect types such as hole, missed yarn, pin hole, knitting fault, oil mark. Eventually, based on the methodologies employed in this paper, it provides a promising stage for the development of an automated fabric defect detection system. This fabric fault detection system was designed and implemented using high level programming language MATLAB.