NTC Project: S01-PH07 (formerly I01-P07) Mohamed Abou-iiana, leader; Christopher Pastore, Claire Beevers (PhilaU), Yasser Gowayed (Auburn) Recent advances in computer technology have brought process automation to many areas of the textile industry. However, knitted fabric structure parameters, such as courses per unit length, wales per unit length, cover factor, fabric thickness and weight per unit area, are still measured manually. This requires large fabric samples and many testing tools. Moreover, manually counting the courses and wales per unit length and unraveling the fabric to determine stitch length are subject to human error and are time consuming. We developed an on-line system that uses image analysis during the knitting process to predict knitted fabric defects much faster than human experts and with greater accuracy. The key to characterizing a knitted structure lies within its basic element, the single knitted loop. Previously, researchers have shown that the length of a yarn knitted into a single loop determines such overall fabric qualities as hand, comfort, weight, extensibility, finished size, cover factor and most importantly fabric dimensional stability. Therefore, the single knitted loop must be of the correct size and shape for a given set of fabric performance criteria. On-line Image Analysis of Knitting If we can measure the knitted loop on-line, we believe that image analysis and processing techniques can correlate on-line size and shape with fabric properties in a matter of seconds with great accuracy. It might also provide a means for maintaining the required loop shape to achieve certain fabric properties. We have now shown that image analysis of the knitting process has great potential to provide reliable measurements to objectively evaluate of knitted fabric. Using the principles of image analysis, we developed an automatic fabric evaluation system, which enables rapid structure-analysis of knitted fabrics. We can now automatically measure fabric construction parameters by analyzing fabric images captured by a CCD camera and preprocessed by Gaussian filtering and equalization function. slow compared with that of fabric production and many experts are required, each with their own subjective decisions. We developed a quality evaluation system to detect and classify knitted fabric defects, such as hole, fly, needle run and various barré defects. From an analysis of a wide variety of defect and defect-free fabrics, our system predicted quality, thus performance, consistent with results obtained from a panel of expert inspectors (see Graph). In several cases, our system‘s classifications were better than the expert evaluations. Conclusions We developed an on-line computerized system that uses image analysis during the knitting process to predict knitted fabric defects much faster than human experts and with greater accuracy. The results of our on-line computerized analysis corresponded well with experimental values. Our system should also be able to identify the type and potential source of a defect and provide the operator with information about how to correct the problem. It is expected that the quality evaluation system can be used for automatic online quality control on the knitted machine or as a supplement to objective gradation of knitted fabric. 100 90 80 Performance % On Line Weight and Shrinkage Control of Cotton Knits 70 60 50 40 30 20 10 0 Defect free Hole Program Fly Needle run Experts Stitch length Variation Thin end Thick end Contributing Graduate Students: Safinaz Youssef, Suhasini Ram (PhilaU), Ebraheem Shady; Research Associate: James Kaufmann (Auburn). Industry Interactions: 3 [National Textiles]; Other Interactions: Non-NTC Academic: 1; Government: 2 Project Web Address: http://fibers.philau.edu/ntc/S01PH07 For Further Information: 1. Textile Institute World Conference, Cairo Egypt, March 2002. Using our new system we documented changes in knitted fabric structure as a result of different relaxation conditions and characteristics of structural changes in knitted fabrics during the manufacturing processes. Our image capturing and analysis system is capable of on-line control of the spatial characteristics of the knit structures before and after wet treatments. Predicting Knitted Fabric Defects Evaluation of knitted fabric for quality is very important for determining its commercial value. Many researchers have worked to automate fabric quality evaluation, but it is still done mainly by human inspectors who mainly just point out defects. Moreover, manual inspection is very National Textile Center Research Briefs –Management Systems Competency: June 2004 NTC Project: S01-PH07 (formerly I01-P07) Mohamed Abou-Iiana, an Asistant Professor of Textile Engineering at PhiladelphiaU since 1997, earned his Ph.D. from NC State in knitting engineering in 1995, a masters from Leicester Polytechnic (ENG) in 1987, and B.Sc. in textile engineering from Alexandria University (Egypt) in 1983. Mohamed spent about 15 years in the textile industry in Egypt and USA in knitting, dyeing and finishing and apparel industries. His research interests include knitting, online control of knitting machines, mechanical properties of fabrics and software development for the textile industry. I98-P03, S01-AE32, S01-PH07* abouiianam@philau.edu (215)-951-2680 http://faculty.philau.edu/abouiianam Claire Beevers, an Associate Professor of Textile Design at PhilaU, joined the faculty in 1985. Claire has a B.Sc. in textile design from Huddersfield Polytechnic in 1981 and a M.A. in textile/fashion design from Leicester Polytechnic in 1983. Her research interests include weft knit fabric development, particularly 3 dimensional shaping for industrial and technical applications. SO1-PH07 beeversc@philau.edu (215)-951 2511 http://faculty.philau.edu/beeversc Yasser A. Gowayed, a Professor at Auburn joined the faculty in 1992, when he received a Ph.D. in fiber and polymer science at NC State. He also earned a M.S. in materials engineering from the American University (Cairo) in 1989 after an 8-year career in industry as a structural designer and civil engineer. Yasser's research interests include modeling and analysis of textile composites, image analysis, geotextiles and re-utilization of solid wastes. F92-A02, F92-S12, F94-A08, F95-A24*, I95-A11, F98-A04, I96-A09, F00-PH05, S01-PH07, C02-AE08* ygowayed@eng.auburn.edu (334)-844-5496 http://www.eng.auburn.edu/~ygowayed Christopher M. Pastore, an Associate Professor of Textile Engineering and Technology and Director of Research of the School of Textiles and Materials Technology at PhilaU, joined the faculty in 1995. Previously he was on the Textile Materials Science faculty at NC State and the Materials Engineering faculty at Drexel Univ. Chris holds a B.A. and M.S. in mathematics and a Ph.D. in materials engineering from Drexel in 1988. His research interests include modeling of fabric and composite structures. F92-S12, F92-S09, F98-P01*, S99-PH01*, F00-PH01, F00-PH05*, S00-PH08, S01-PH07, S03-PH01 pastorec@phila.edu (215)-951-2683 http://fibers.philau.edu/stmt/cpastore.html National Textile Center Research Briefs –Management Systems Competency: June 2004