Non-destructive assessment of the internal quality of watermelon using ultraviolet near-infrared spectroscopy
Watermelon is a popular tropical fruit in Malaysia especially during dry season because it considered as good thirst quenching fruit. Recently, the demand for high quality fresh watermelons has increased. In order to meet the customer’s expectation, the watermelons need to be harvested at the rig...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2019
|
| Subjects: | |
| Online Access: | http://psasir.upm.edu.my/id/eprint/77680/1/FK%202019%2045%20ir.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Watermelon is a popular tropical fruit in Malaysia especially during dry season
because it considered as good thirst quenching fruit. Recently, the demand for
high quality fresh watermelons has increased. In order to meet the customer’s
expectation, the watermelons need to be harvested at the right level of maturity
level. Fruits harvested too early have a poor quality as they are not yet ripened
properly. Therefore, optimal level of maturity level is important to preserve the
quality of the fruit for a longer shelf life. Internal quality indices of watermelon
such as soluble solids content (SSC), pH, firmness and moisture content (MC)
are the main indicators to determine the maturity level of watermelons. However,
the current practice to determine the maturity level of watermelon based on the
sound produced by tapping the fruit is subject to errors. This method is not
consistent even if it was performed by skillful and experienced farmers.
Therefore, there is a need for an accurate and reliable in-situ measurement
system to measure the quality of watermelons. Thus, the goal of this study was
to investigate the possibility of using ultraviolet near infrared (UVNIR)
spectroscopy to determine the internal quality parameters of watermelon for
maturity level prediction.
A total of 63 watermelon samples at different level of maturities (unmatured,
matured and over-matured) were used in this study. Each sample was vertically
divided into three sections namely top, middle and bottom. Then, each portion
was horizontally divided into two portions. Then, the spectral data of the skin
surfaces was collected from each sub-portion. All the spectral measurements
were carried out in the black box to minimize the influence of stray light on the spectral data. Each portion underwent laboratory analysis to measure the SSC,
pH and MC. In this study, the measurement of SSC was done using a
reftractometer while the measurement for pH was undertaken using pH meter.
The MC of the sample was measured using conventional oven dry method. The
chemical properties obtained from the samples at different maturity levels were
significantly difference at (p<0.05). The calibration and prediction models were
developed to correlate the spectral data with internal quality properties of the
samples using Partial Least Square (PLS) method. To improve the accuracy of
the PLS models, the spectral data was first pre-processed using the baseline
offset correction (BOC) method. The pre-processing methods and PLS exercises
were run using Unscrambler V. 10.3 software.
SSC prediction of watermelon samples was the highest R2 value of 0.57 for
calibration model while R2 value for prediction model was 0.50. While for pH
prediction of watermelon samples, the calibration model gave R2 value of 0.49
whereas the prediction model gave R2 value of 0.43. For MC prediction of
watermelon samples, the calibration model gave R2 value of 0.43 whereas the
prediction model gave R2 value of 0.21. From the three chemical parameters,
SSC was ` the best parameter to be predicted by UVNIR from all maturity
levels since watermelon samples achieved higher correlation coefficient as
compared to others parameters.
For the prediction of SSC for the samples at different maturity levels, it was found
that the coefficient of determination (R2) for calibration models of unmatured,
matured and over-matured were 0.65, 0.81 and 0.78, respectively. While for the
prediction models, it was found that the R2 for unmatured, matured and overmatured
were 0.60, 0.74 and 0.76, respectively. For the prediction of SSC of the
samples at different portions, it was found that the R2 for calibration models for
top, middle and bottom were 0.72, 0.71 and 0.79, respectively. For the prediction
models, it was found that the R2 for top, middle and bottom portions were 0.63,
0 62 and 0.66, respectively.
The SVM model of maturity prediction for watermelon showed an excellent result
with overall prediction of accuracy of 85%. Overall, it can be concluded that the
application of spectroscopy is a promising technique for assessing the maturity
level of watermelons. The proposed method has significant potential uses as a
tool for prediction of watermelon quality parameters in the farm. |
|---|