Component Analysis To Cotton/Lyocell Blended Products Based On "Fiber Image Automatic Capture And Identification System"

To Cotton / Lyocell blended products, they are good moisture absorption, excellent wearing comfort and highly favored market, and there is a mandatory labeling requirement to them. In our country, composition and content of cellulose fibers are detected mostly by traditional methods, which do exist problems, such as examining timeliness, accuracy of test results, methods adaption, test cost and so on. As a result, an automatic, rapid and accurate detection method is needed.A systematic research, on cellulose fibers' identification and the blended ratio in yarns/fabrics, is being made by " computer image Automatic detection and Identification to textiles " task force which derived from Education Ministry's Special Fund for Outstanding Doctoral Thesis [200350]. At present, the automatic detection system—"Fiber image automatic capture and recognition system"has been developed, by which fiber image are captured and identified. As a breakthrough, the study on Cotton / Ramie bulk fiber identification had been made, and then the expansion of the fiber types has been achieved and the identification accuracy is well.However, the main targets of textile blended ratio test are blended yarns and fabrics, compared with the bulk fibers, whose surface morphology will have some changes, because the bulk fibers are spun, dyed or experienced other processes. there need some improvements in the system's identification core, and also the experiment plan for yarns and fabrics, before ratio test on system, is needed to determine.The main contents of this study are:1. The morphology and density differences in different processing segments Cotton / Lyocell are analysed;2. The optimization analysis to characteristic parameters in "F iber Image Automatic Capture and Identification System";3. The improved System Verification and experiment study to improve the system recognition accuracy rate: A large number of experiments had been made and verified its identification accuracy; an appropriate method of sample making has been discussed, as well as the best segment length and the minimum sample size, in order to further improve the system's recognition accuracy rate,,i.e, the methods to obtain 1 mm or so yarn-segments and dispersed those have been explored, the best segment length based on the system has been studied by parallel experiments to different material-sources, the smallest sample size based on system has been discussed from theoretical and experimental analysises. The main conclusions of this study are:1. Compared with bulk fibers, there are marked changes in fibers' diameter and density, as well as the longitudinal morphology in the fibers of yarns and fabrics, perhaps because fibers are subject to different machining /dying processes..2.Extraction principles of the original six characteristic parameters are analysed, discovered that the projections and width curves of cotton/Lyocell fibers showed no significant differences and , the original characteristic parameters can not be effectively quantified their differences. A large number of automatic identification experiments are made and obtained that to the cotton/Lyocell bulk fibers, good identification accuracy can be achieved by the original characteristic parameters. But to cotton/Lyocell blended yarns and fabrics, there is poor recognition accuracy, the accuracy rate fell by an average 5% to knitted fabrics, and 15% to woven fabrics. thereby the new identification parameters to these Cotton/Lyocell fibers are needed design.3. Pointing to two new characteristic parameters, "Inter-segment uneven" and "fragments within the uneven", are advanced by Dr. Wang Rongwu, experiments are made, from which can be conclued that the parameters have a weak association, good stability and effectiveness are showed by the probability distribution of curve characteristic parameters, which is in line with the parameters need of the classification pattern.4. Confirmatory experiments are made. It is obtained that after the system optimization, the fiber average recognition accuracy rate in yarns and knitted fabrics is an average of 95% or more; the factors ,which affected the system identification, are characteristics in fibers themselves, textiles, dyeing and finishing, sample-making methods and so on; confirmatory experiment showed that slide-making program is the biggest factor to the accuracy rate.5. the slide-making program has been Determined to further improve the system identification accuracy: Comparative analysis are made and determined the optimum cutting tool for shorter yarn-segments and the best yarn-segments dispersed liquid, that is Harrington slicer and anhydrous ethanol; the system's optimal Cut length range are choosed, from 0.6-0.7mm; as to the minimum sample size satisfying the stability of the system recognition accuracy, namely, the recognition results variance is not more than 3%, theoretical analysis showed that fiber-segments are not less than 2467, experiments revealed that woven and knitted fabrics are respectively not less than 2106, 1583, considering the established system and other factors, the minimum sample size slightly larger than the theoretical value, for 2500. The results showed that the program is valid for improving the improved system's Cotton / Lyocell fiber recognition accuracy rate.