Abstract
Notwithstanding that the Kubelka–Munk function [F(R)] is the theoretically preffered treatment, relative to log 1/r, for reflectance data, it has found little favour with workers in near infrared (NIR) reflectance. The most often quoted advantage for F(R) is an improvement in linearity with concentration, which occurs when measurements are made over a wide range of reflection and concentration. However, the practice in NIR is to limit the reflectance range by standardising the method of sample preparation. Hence, the linearity of log 1/r is not an issue. However, is this restriction on sample preparation methodology a result of the use of log 1/r? This study shows that moisture changes in ground wheat fractions are linear with F(R) and curvilinear with log 1/r, and demonstrates that a calibration for moisture in wheat based on F(R) and a large reflectance range, can be successfully transferred to an instrument substantially different from that used to develop the calibration.
Get full access to this article
View all access options for this article.