Author
Liu, Yongliang | |
Gamble, Gary | |
Thibodeaux, Devron |
Submitted to: ASABE Annual International Meeting
Publication Type: Proceedings Publication Acceptance Date: 6/15/2010 Publication Date: 7/14/2010 Citation: Liu, Y., Gamble, G.R., Thibodeaux, D.P. 2010. Evaluation of 3 cotton trash measurement methods by visible/near-infrared reflectance spectroscopy. 2010 ASABE Annual International Meeting, June 20-23, 2010, Pittsburgh, PA. p. 1-8. Interpretive Summary: Presence of non-lint material (trash) in lint cotton fibers at various degrees influences the market values, and also hurts the end-use qualities. Currently, three types of instrumentals have been developed to assess the trash content in lint cotton fibers, namely, Shirley analyzer (SA), advanced fiber information system (AFIS), and high volume instrumentation (HVI). Each of these instruments has its unique advantages, and comprehensive comparison on the trash content measurements among them is not available. Near infrared (NIR) spectroscopy has been considerably applied for the quantitative prediction of key color and physical indices in cotton fibers, and could be a potential and independent tool for the evaluation of three trash measurement techniques. Partial least squares (PLS) regression models, relating the spectral response in 400-2500 nm visible/NIR region to the respective SA, AFIS, and HVI readings, were developed in various spectral ranges from two chemometric packages. 2 HVI readings were observed to have better model performances and are acceptable for quantitative determination of cotton trash content. The outcome provides cotton fiber / textile engineers, researchers and regulators a new sight in applying both HVI and visible/NIR spectroscopy for rapid and routine determination of cotton trash content. Technical Abstract: Currently, three types of instrumentals have been developed to assess the trash content in lint cotton fibers, namely, Shirley analyzer (SA), advanced fiber information system (AFIS), and high volume instrumentation (HVI). Each of these devices has its unique advantages, and comprehensive comparison between them has not been attempted. Visible / near infrared (NIR) spectroscopy could be a potential and independent tool for their evaluation. Partial least squares (PLS) regression models, relating the spectral response in 400-2500 nm visible/NIR region to the respective SA, AFIS, and HVI readings, were developed in various spectral ranges from two chemometric packages (Foss Vision and Thermo Grams/AI) and then compared. Because of different strategies to select calibration and validation samples between two commercial packages, resultant NIR trash models were inconsistent. 2 HVI and AFIS readings were observed to have better model performances in Thermo Grams/AI approach than in Foss Vision, and are acceptable for quantitative determination of cotton trash content. Furthermore, considering different sampling species between reference and spectral measurement, 90% confidential interval was applied to remove outlier samples. The recalibrated models verified that HVI trash module might be the most promising one in precise and accurate determination of trash content in lint cottons. |