Submitted to: National Cotton Council Beltwide Cotton Conference
Publication Type: Proceedings
Publication Acceptance Date: January 11, 2010
Publication Date: April 21, 2010
Citation: Liu, Y., Gamble, G.R., Thibodeaux, D.P. 2010. Comprehensive investigation of NIR technique in cotton fiber quality assessment. National Cotton Council Beltwide Cotton Conference, January 4-7, 2010, New Orleans, LA. p. 1328-1333. Interpretive Summary: Cotton is one of the most important agricultural commodities in the world, and the subsequent need for rapid and accurate determination of its quality indexes is a much discussed topic from policy makers to cotton fiber processors. Traditional methods, such as high volume instrumentation (HVI) and advanced fiber information system (AFIS), have been developed as viable tools to measure a number of cotton quality indexes. Although these methods can measure many different quality characteristics and are practiced throughout the cotton industry, the procedures are destructive, time consuming, and prone to day-to-day and location-to-location variations. The development of fast, non-destructive, accurate, and routine techniques is critical to enhance cotton fiber quality classing efficiency. Near infrared (NIR) spectroscopy, with an extension to the UV-visible region, could be an important alternative technique due to speed, ease of sampling, and low-cost. In this approach, cotton fibers were scanned in the region of 220-2200 nm and the corresponding reference values were analyzed by HVI and Shirley Analyzer. Partial least squares (PLS) regression models were developed, and the calibrations were optimized by the coefficient of determination (R2), root mean square error of validation (RMSEV), and residual predictive deviation (RPD) in an independent test set. The results indicated that this technique could be used to predict such quality indexes as micronaire and +b for quality control and other properties, including Rd, mean length, upper-half mean length, and visible trash content for screening programs. This outcome provides cotton fiber / fabric / textile engineers and researchers a new sight in applying both optical UV/visible / NIR and imaging spectroscopy for rapid and routine grading and classification of cotton qualities.
Technical Abstract: Near infrared spectroscopy, with an extension to UV and visible region, has been considerably applied for the qualitative and quantitative measurements of key color and physical characteristics in cotton fibers. However, the results have been inconsistent, mostly due to the use of different spectral regions. This work examined and compared the NIR model performances built from various regions for the prediction of cotton color (Rd and +b), physical properties (micronaire, strength, and length indexes), and trash content. On the basis of results, improvement in reference determination and consideration of other spectroscopic approach were suggested.