Objective measurement and grading of seam pucker using texture analysis

摘要

Seam pucker is one of the most frequently occurring defects in garment industry. It is due to wide range of factors including fabric properties, sewing thread properties and sewing machine parameters. These factors separately or interacted affect the pucker grade. Many studies have been made on factors affecting Seam Pucker, but in the area of subjective grading using AATCC reference images. However, in the area of using image processing in objective measurement of seam pcker very little have been published. In the present work, seam pucker grade is tested on open seams resembling normal condition of industrial inspection. Seam assemblies are prepared according to a three dimension experimental design changing six factors affecting pucker grade. Image acquisition module obtains the digitized image of the seam assembly and stores it as a Bmp format in a computer file. Texture features are analyzed statistically. Three main computer programs are used on Matlab Version 5.3 for identifying fabric type and color, extracting first order statistics features from digitized image, and extracting the second order statistical texture feature. Fast Fourier Transform (FFT) and Discrete Cosine Transforms (DCT) Transforms were also tested for Pcker grading. AATCC Seam Pucker reference images are stored in the computer and delt with in the same way in order to set numbers for grading through comparing and quantifying grade. Fabric type, sewing machine teeth height and presser foot height were the factors affecting pucker grade. First order statistical feature analysis showed better results than second order one. FFT transform gave better results than DCT transform.