An Optimised Six-Wavelength Model for Predicting Kiwifruit Dry Matter

Abstract

Optimised wavelength selection is important to the development of new types of inexpensive and portable near infrared instruments that might be used on fruit in orchards. The use of discrete bandwidth devices, such as light-emitting diodes, requires preselection of a small number of discrete wavelengths. In this work, a kiwifruit data set consisting of 834 absorbance spectra and corresponding fruit dry-matter measurements, an important maturity indicator for kiwifruit, has been subjected to an exhaustive wavelength search to build optimal multiple linear regression models of up to seven wavelengths. Using a standard partial least-squares model as a benchmark, a six-wavelength model has been identified as an optimum, predicting kiwifruit dry matter with r 2 of 0.88 and root mean square error of prediction (RMSEP) of 1.22%. The sensitivity of the model to shifts in the key wavelengths was also evaluated, revealing that a 1 nm offset or a 0.25 nm random noise component would be enough to increase the RMSEP by around 0.04% in actual dry matter value or 3% in relative percentage terms.

Comparison of hand-held near infrared spectrophotometers for fruit dry matter assessment

“…When light from the lamp penetrates the fruit, some light is scattered and transmitted through the fruit flesh and is emitted back through the shadowed region into the fibre-optic probe. 9 The probe directs the light to the MMS1 spectrometer. The spectral range of the spectrometer is 310–1100 nm with a sampling interval of 3.3 nm and an optical resolution varying between 8 and13 nm across the wavelength range.…”

“…Both of these instruments use gold-coated foil as their reference standard, usually set to be measured before or after every sample measurement. The total time to record a spectrum, including reference standard measurement, is approximately 5 s. 9 Two instruments of the NIRVANA (labelled N010 and N011) and two of the F-750 type (F750-09 and F750-10) were used in our experiments.

Figure 1.Commercially manufactured hand-held, portable instruments: (a) NIRVANA (Integrated Spectronics, Baulkham Hills, Australia; discontinued), (b) F-750 Produce Quality Meter (Felix Instruments, Portland, USA), (c) SCiO spectrometer (Consumer Physics, Hod HaSharon, Israel) and (d) H-100C NIR portable spectrometer (Sunforest, Incheon, Korea).

…”

“…Both of these instruments use goldcoated foil as their reference standard, usually set to be measured before or after every sample measurement. The total time to record a spectrum, including reference standard measurement, is approximately 5 s. 9 Two instruments of the NIRVANA (labelled N010 and N011) and two of the F-750 type (F750-09 and F750-10) were used in our experiments.…”

“…Three of the hand-held instruments are relatively new (SCiO, H-100C and F-750) and not well covered in the literature, whilst the fourth (NIRVANA) is older and well established. 9,10 As a benchmark, an in-house laboratory NIR spectrometer instrument (Benchtop) of known and reported performance levels was also included in the study. 11 Fruit data sets were specifically created that involved up to three basic fruit types (apples, kiwifruit, summerfruit), each including sub-categories of each fruit type depending on what was readily available from a local retail market.…”

Light-emitting diodes as light sources for spectroscopy: Sensitivity to temperature

“…Temperature-induced wavelength shifts in LED output will impact spectroscopic performance. Shafie et al 15 optimised a six-wavelength (650, 800, 815, 885, 920 and 930 nm) multiple linear regression model for predicting kiwifruit dry matter content (R 2 ¼ 0.94 and root mean square error of prediction (RMSEP) ¼ 1.22%). The temperature-related drifts expected of LED sources were simulated, with wavelength perturbations of 5 nm resulting in a 1% increase in RMSEP.…”