Proceedings Article | 14 December 2004
KEYWORDS: Near infrared, Calibration, Optical fibers, Diffuse reflectance spectroscopy, Spectroscopy, Near infrared spectroscopy, Analytical research, Statistical analysis, Error analysis, Nondestructive evaluation
In the near-infrared (NIR) region, every component is corresponding to specific absorption spectra characteristics. So NIR spectroscopy has gained wide acceptance in many research and application fields by virtue of its advantages over other analytical techniques. This study was about nondestructive sugar content detection of “Fuji” apples by means of NIR diffuse reflectance technique using optical fiber. Variance analysis of average absorbency and root mean square (RMS) noise of NIR spectra indicated that different testing positions had little influence on spectra acquisition while different testing distances had an opposite outcome. The relationship between sugar content and NIR spectra of apples was analyzed via multi linear regression (MLR) method using software SAS. Optimal single-wavelength (1453 nm), double-wavelength (1732 nm and 1790 nm), triple-wavelength (1453 nm, 1732 nm and 1790 nm) and quaternion-wavelength (1453 nm, 1732 nm, 1790 nm and 1931 nm) calibration equations were established. Correlation coefficients (R) of the calibration set were 0.754, 0.864, 0.907 and 0.921, respectively, standard error of calibration (SEC) were 1.439ºBrix, 1.103ºBrix, 0.922ºBrix and 0.851ºBrix, respectively. Of the prediction set, correlation coefficients (R) were 0.438, 0.687, 0.746 and 0.868, respectively, standard error of prediction (SEP) were 2.342ºBrix, 1.835ºBrix, 1.171ºBrix and 0.918ºBrix, respectively. The results show that NIR diffuse reflectance technique is a feasible method for nondestructive detection of apple fruit sugar content. Furthermore, this study lays a solid foundation for setting up the sugar content forecasting model of apples.
