Skip to main content
ARS Home » Research » Publications at this Location » Publication #261232

Title: An approach to determine calibration samples in NIR modeling for cotton fiber strength

Author
item Liu, Yongliang
item Gamble, Gary
item Thibodeaux, Devron

Submitted to: NIR news (Near Infrared Reflectance News)
Publication Type: Popular Publication
Publication Acceptance Date: 10/30/2010
Publication Date: 11/15/2010
Citation: Liu, Y., Gamble, G.R., Thibodeaux, D.P. 2010. An approach to determine calibration samples in NIR modeling for cotton fiber strength. NIR news (Near Infrared Reflectance News). 21(8):9-11,13.

Interpretive Summary: Near infrared (NIR) spectroscopy, a useful technique due to the speed, ease of use, and adaptability to on-line or off-line implementation, has been applied to perform the quantitative prediction of cotton strength. However, resultant NIR models were observed to lack the power for practical applications. Probably, this originates directly from at least one of four main factors, (1) NIR spectral absorptions do not contain the strength information, (2) strength references were inappropriately determined, (3) un-identical (or different) specimens were utilized between spectral collection and reference determination, and (4) highly heterogeneous distribution of strength property occurred in native fibers. Although both HVI and Stelometer strength readings have been adapted considerably in cotton fiber quality assessment, little effort has been attempted to unravel the relationship between the two. In the latest study, we have not only demonstrated the consistency in strength readings between two instruments if the strength readings were divided by cotton micronaire, but also observed a better NIR model by using the modified strength index as references. Nevertheless, two strength readings from two independent strength testing should be highly correlated. This concept could be implemented to determine the appropriate samples in calibration set, before NIR development on cotton strength. The resultant model demonstrated the feasibility of NIR technique in the quantitative prediction of cotton fiber strength for the purpose of quality control. The outcome provides cotton fiber/textile engineers, researchers and regulators new sight in applying both HVI and NIR technique for rapid and routine determination of cotton strength quality.

Technical Abstract: In the previous study, we have reported the consistency in strength readings between HVI and Stelometer instruments if two readings were divided by cotton micronaire. Nevertheless, two strength readings from two independent strength testing methods should be highly correlated. This concept could be implemented to determine the appropriate samples in calibration set, before developing NIR model on cotton strength. Therefore, for the first time, we proposed a pre-screening procedure to determine appropriate calibration samples for building NIR models. The resultant model demonstrated the feasibility of NIR technique in the quantitative prediction of cotton fiber strength for the purpose of quality control.