342
B. Vincent and P. Dardenne
sample presentation. Different studies have concluded that measurement in the reflection and interaction modes is more appropriate for fruit analysis. It is essential to
report that the penetration depth for apples has been measured in the reflection mode
and is about 4 mm for the 700–900 nm range and 2–3 mm for the 900–1900 nm
range. In the transmission mode and in the 1400–1600 nm range, less than 1% of the
initial intensity of the radiation goes through a 1 mm slice. The skin definitely poses
a major barrier for the light entering the flesh of the apple, requiring a strict protocol
for presentation of the sample to the instrument. This protocol will be adapted to
the fruit analysed, the architecture of the instrument (mainly the relative position of
the source/sources and detector/detectors). Several reviews summarise the potential
of NIR for determining different parameters and criteria of fruit [36, 37]. A specific
review has been dedicated to challenges and solutions for quality inspection for
robotic fruit instrumentation [38].
Several parameters can be determined with enough precision to be used routinely.
A number of authors have reported on the use of NIR spectroscopy to determine apple
quality parameters, such as soluble solids, acidity, pulp firmness, maturity indexes,
polyphenols and vitamin C [36]. Pissard et al. has shown that NIR technique can
be used to determine sugar, vitamin C and total polyphenols contents [39]. This
study, based on large spectral databases (between 1274 and 2646 depending on the
parameter studied) built in the framework of breeding programmes and European
projects, has demonstrated the high precision of models that can be achieved. Low
standard error of prediction values, in addition to relatively high ratio to prediction
(RPD) values, has been obtained especially for total polyphenol and sugar content
(RPD values of 5.1 and 4.3 for polyphenol and sugar, respectively). These same
authors have also studied the intra-fruit variability in apples using classical and NIR
techniques [40]. This paper proposes and validates a protocol to analyse fruit based
on reference analyses of a representative sample of the apple and NIR measurements
collected at four points 45° from each other in the equatorial region of the fruit (i.e.
apple). It has been demonstrated that there was little difference between the mean
value at the four points and the mean value of the entire apple. The potential of NIR
spectroscopy on fresh apples to determine the phenolic compounds and dry matter
content in peel and flesh has been also studied [41]. In general, one of the challenges
is the online analysis of intact fruit.
More and more handheld NIR devices are commercially available and proposed
to analyse fruit. NIR uses under field conditions (i.e. orchard) have been limited
for many years due to restrictions imposed by the size and low robustness of the
instruments available. Recently, the development of new technologies used in the
construction of NIR spectrometers and data acquisition strategies has enabled a
significant reduction in size and cost of these instruments but often a decreased of
the robustness of the methods developed [42, 43]. The challenge is to set up the
right procedure to use the historical databases and calibration models, previously
developed using benchtop spectrometers.
B. Vincent and P. Dardenne
sample presentation. Different studies have concluded that measurement in the reflection and interaction modes is more appropriate for fruit analysis. It is essential to
report that the penetration depth for apples has been measured in the reflection mode
and is about 4 mm for the 700–900 nm range and 2–3 mm for the 900–1900 nm
range. In the transmission mode and in the 1400–1600 nm range, less than 1% of the
initial intensity of the radiation goes through a 1 mm slice. The skin definitely poses
a major barrier for the light entering the flesh of the apple, requiring a strict protocol
for presentation of the sample to the instrument. This protocol will be adapted to
the fruit analysed, the architecture of the instrument (mainly the relative position of
the source/sources and detector/detectors). Several reviews summarise the potential
of NIR for determining different parameters and criteria of fruit [36, 37]. A specific
review has been dedicated to challenges and solutions for quality inspection for
robotic fruit instrumentation [38].
Several parameters can be determined with enough precision to be used routinely.
A number of authors have reported on the use of NIR spectroscopy to determine apple
quality parameters, such as soluble solids, acidity, pulp firmness, maturity indexes,
polyphenols and vitamin C [36]. Pissard et al. has shown that NIR technique can
be used to determine sugar, vitamin C and total polyphenols contents [39]. This
study, based on large spectral databases (between 1274 and 2646 depending on the
parameter studied) built in the framework of breeding programmes and European
projects, has demonstrated the high precision of models that can be achieved. Low
standard error of prediction values, in addition to relatively high ratio to prediction
(RPD) values, has been obtained especially for total polyphenol and sugar content
(RPD values of 5.1 and 4.3 for polyphenol and sugar, respectively). These same
authors have also studied the intra-fruit variability in apples using classical and NIR
techniques [40]. This paper proposes and validates a protocol to analyse fruit based
on reference analyses of a representative sample of the apple and NIR measurements
collected at four points 45° from each other in the equatorial region of the fruit (i.e.
apple). It has been demonstrated that there was little difference between the mean
value at the four points and the mean value of the entire apple. The potential of NIR
spectroscopy on fresh apples to determine the phenolic compounds and dry matter
content in peel and flesh has been also studied [41]. In general, one of the challenges
is the online analysis of intact fruit.
More and more handheld NIR devices are commercially available and proposed
to analyse fruit. NIR uses under field conditions (i.e. orchard) have been limited
for many years due to restrictions imposed by the size and low robustness of the
instruments available. Recently, the development of new technologies used in the
construction of NIR spectrometers and data acquisition strategies has enabled a
significant reduction in size and cost of these instruments but often a decreased of
the robustness of the methods developed [42, 43]. The challenge is to set up the
right procedure to use the historical databases and calibration models, previously
developed using benchtop spectrometers.