Pilling evaluation of patterned fabrics based on a gradient field method

Abstract

An objective method for the detection of pills on different kinds of patterned fabric based on image analysis techniques has been developed. The proposed method for pilling evaluation includes the 3D reconstruction of a fabric surface with pills and the detection of pills of fabrics with various patterns. Shadows created by pills with lateral illumination of the samples, while images of the samples are being captured, are the basic element for the 3D reconstruction of the fabric surface using a gradient field method. By using this method, the effect of fabric pattern on the results of 3D reconstruction is suppressed. The reconstructed fabric surface enables effective segmentation of pills from a fabric texture using simple image processing tools. Important characteristics of pills, such as number, area, perimeter, density and contrast, are extracted for the objective pilling evaluation. The results show that the proposed method is applicable for the evaluation of pilling not only for unicolor fabrics but even for fabrics with various kinds of patterns.