Denim fabric defect detection by using correlation method

Fabrics are produced by at least two yarn systems which are perpendicular to each other in a certain order. Warp fault defect detection can not be done by using human eye easily and it is harder for denim fabrics especially. Therefore this hard task requires care and it is stated that can be done by trained quality controllers with %70 correct detection rate. The correct detection of the fabric defects is very important. To detect the fabric defects by using image processing techniques the identity report must be calculated correctly and searched compatibility all over the fabric. The aim of this study is the detection of different fabric and color defects that may be seen in denim fabrics by using image segmentation techniques. To detect defects such as pilling and stain (oil, color) during denim fabrics production, the length and width of identity report matrix are computed from denim fabrics images by using correlation method. Before this process images are preprocessed and satisfactory results are achieved in fabric defect detection by applied methods.