Visual quality is an important quality component of textile fabrics. This component considerably affects the value of the product and the consumer willingness to buy decision. Therefore, fabric inspection is very important for manufacturers before marketing the textile fabric goods. Until recently, fabric quality inspection is performed by human inspectors which have many drawbacks such as; the need for experienced workers increases cost; fatigue after working short periods of time; subjectivity, the same fault might be classified into different defect categories by different inspectors; small defects are difficult to be detected by human inspectors. Developing fabric defect detection approaches using image processing methods has been a popular research topic due to the importance of the automatic inspection systems in textiles.
In order to develop a robust automatic inspection system, fabric defects are divided into two classes: directional defects (DD) and regional defects (RD). Although general aim of the researchers and developers is to focus on an approach for detecting all the defect types at once, in this research only directional defects are considered. Partitioning the problem leads to effective development of method for each defect type. In this presentation it is shown that defect-free standard signals obtained from projection profiles and defected signals difference leads to detecting the location of directional defects due to the directional characteristics of the defects and fabric texture
Anahtar Kelimeler: Defect Detection, Image Processing, Projection Profiles, Directional Defects
|
