Monitoring Fruit Ripeness with Near Infrared Diffuse Reflectance Spectroscopy
Written by Miriam Mowat, Applications Scientist
Consumers demand high levels of quality and consistency in their food. For the food industry, especially with its complex international supply chains, careful monitoring methods are critical to delivering quality and consistency. One viable monitoring method is near infrared (NIR) reflectance spectroscopy. In this application note, we demonstrate the application of NIR reflectance spectroscopy to measure fruit ripening.
Monitoring the ripeness of fruit allows farmers, processors and retailers to minimize losses and maximize quality through accurate categorization of produce quality. This reduces waste throughout the supply chain and improves consistency for the user. A great advantage of NIR spectroscopy is that it requires no preparation of the sample.
In this application note we demonstrate the use of the NIRQuest, a compact and robust diode array spectrometer, for the monitoring of ripeness in bananas and other fruit to identify ripening before it becomes visible on the skin of the fruit. The NIRQuest’s combination of performance, size and ruggedness make it ideal for developing applications of NIR spectroscopy in the lab that can be directly transferred using the same equipment to the field or factory.
Experimental Setup and Method
To measure the reflected spectra of the fruit, we conducted our investigation using the NIRQuest512-1.7 (900-1700 nm) and NIRQuest256-2.5 (900-2500 nm) spectrometers, an HL-2000 tungsten halogen light source and a reflectance probe. The probe was clamped onto a reflection-transmission stage (STAGE-RTL-T), which can be adjusted to keep the probe at a fixed distance from the fruit’s surface (Figure 1). Table 1 provides details for the equipment used and the reasons for its selection.
Measurements were made on a daily basis for seven consecutive days. Bananas, apples and oranges were measured, all purchased from a supermarket on the first day of measurements. A WS-1 diffuse reflectance standard was used as a reference between each set of measurements. Six measurements were taken of each fruit at different orientations, and an average taken. This was done in order to account for the inhomogeneity of the fruit. Even though no sample preparation is required with NIR spectroscopy, the structural inhomogeneity of fruit makes it necessary to take an average at different orientations to achieve results that are representative of the fruit as a whole. For most applications, NIR spectroscopy also requires the development of sophisticated chemometric calibration models. While not addressed in this note it is an area of active applied research for the monitoring of fruit ripeness (1), (2), (3).
The process of data collection was as follows:
- Turned on the HL-2000 light source and waited 15 minutes for the source to reach thermal equilibrium and stabilize.
- Made a dark measurement with the light source shutter off and the probe facing the base of the reflection stand.
- Made a reference measurement using the diffuse reflectance standard.
- Made 6 measurements of the first fruit sample at different orientations, bringing the probe clamp to rest on the fruit’s surface, using a fixed distance of ~0.5 cm between the fruit’s surface and the probe end.
- Repeated steps 3 and 4 for each fruit sample.
Table 1: NIR Diffuse Reflectance Spectroscopy Setup
Results and Discussion
We measured the reflectance of two banana samples, as well as apples and oranges, over a period of seven days. The results for each sample showed similar trends but were clearest for the banana, since bananas tend to ripen more quickly.
The raw spectrum for a single banana as measured on different days is shown in Figure 2. The data at less than 950 nm and greater than 1650 nm was disregarded as these regions contained no significant information and are the parts of the spectrum most affected by noise since they are close to the limits of the instrument’s wavelength response. The spectra show a distinct trend with time, most noticeable at the peaks, such as at 1450 nm.
NIR spectroscopy is a vibrational technique, and relates to the absorbance of energy due to the excitation of molecules. In the NIR spectrum (750-2500 nm) these are overtones and combinations of the fundamental vibration frequencies related to specific functional groups within the sample. In organic samples these are dominated by O-H and C-H functional groups (4). Because most organic matter is majority water, changes in moisture (measured with the C-H functional group) tend to dominate the measurements.
This is shown in Figure 2 by the absorbance bands from 1400-1550 nm, at 1150 nm and at 1000 nm, which indicate changes in the fruit’s water content. The reflectance increases each day, most rapidly in the first three days. This indicates that the water content is decreasing as the fruit ripens and we can observe this as a corresponding drop in the absorbance. Correlation with moisture is a well-established technique for measuring ripeness (5).
The first derivative of the data shown in Figure 2 was calculated to show more clearly the changes taking place in the banana. This is displayed in Figure 3 and shows the rate of change of the absorbance with respect to wavelength. The trend of the fruit over time is readily observed in Figure 3, with each spectrum showing a different day’s measurements. For example, for the first two days the rate of change at the peaks was significantly greater than that of the later days, when the banana was much riper and beginning to brown. This indicates that it may be possible to use NIR to not only quantify the changes seen in fruit, but to detect differences even before it is possible to do so by sight or colorimetric measurements.
Other non-contact optical methods exist for monitoring ripeness of produce such as optical densitometry. But unlike densitometry, NIR reflectance spectroscopy gives you information at specific wavelengths, allowing for more robust, reliable results as well as more precise information about parameters such as sugar, fat and protein content (6).
The temperature of the water in the fruit has an effect on the spectrum measured and is likely to vary slightly day to day, making it difficult to obtain accurate results for one fruit sample. These measurements would require chemometric calibrations to model accurately. Such models are widely used in NIR spectroscopy, and much work has been done demonstrating the applicability of these models to NIR measurements in fruit (1), (2), (3). While our results are far too small a dataset from which to build a calibration model, they showed a clear trend in the gradient of the spectra across the peaks at 1000 nm, 1150 nm and 1400-1550 nm related to the changes in moisture as the fruit ripened. Significantly, it was possible to see this spectral change occurring in the banana before any visible change occurred.
Rotten or overripe fruit is a serious source of waste and costs producers, processors and consumers time and money. By monitoring the ripeness of the fruit throughout the supply chain it is possible to better control the ripeness of the product when it hits the supermarket shelf.
This application note has shown that it is possible to use NIR diffuse reflection spectroscopy to monitor ripeness of fruit, and that as the fruit becomes riper, there is a trend that can be characterized related to moisture content. Also, we demonstrated that taking multiple measurements across the solid fruit is important to ensure that the measurement is representative of the sample.
NIR reflectance spectroscopy is a fast and reliable way to monitor the quality of fruit and produce without any sample prep or damage to the product. While real-world application would require further development of calibration models for use in a quality control environment, we have demonstrated how NIR spectroscopy can be a powerful tool for monitoring fruit ripeness. The measurement is relatively easy and requires no sample preparation. The process is fast, can be very reliably applied and with the appropriate calibration models can be extended for many parameters. It has the potential to save time and money through non-contact monitoring of fruit quality throughout the supply chain — from the field to the factory and to the supermarket shelf. A world with fewer rotten bananas — we like the sound of that!
- Linear and non-linear regression models for classification of fruit from visible-near infrared spectra. Kim, J., Mowat, A., Poole, P., Kasabov, N. 2, July 24, 2000, Chemometrics and Intelligent Laboratory Systems, Vol. 51, pp. 201-216.
- Postharvest quality of apple predicted by NIR-spectroscopy: Study of the effect of biological variability on spectra and model performance. Bobelyn, E., Serban, A.-S., Nicu, M., Lammertyn, J., Nicolai, B., Saeys, W. 55, 20110, Postharvest Biology and Technology, pp. 133-143.
- Preliminary study on the application of near infrared spectroscopy and pattern recognition methods to classify different types of apple samples. Luo, W., Shuangyan, H., Fu, H., Wen, G., Cheng, H., Zhou, J., Wu, H., Shen, G., Yu, R. 128, 2011, Food Chemistry, pp. 555-561.
- Using NIRS spectroscopy to predict post harvest quality. Cayuela Sanchez, J. A. Sevilla : s.n., 2012.
- Non-destructive measurement of moisture content using handheld NIR. Blakey, R. J., van Rooyen, Z. 2009.
- Non-destructive measurement of moisture content in avocado’s using handheld near-infrared spectroscopy. Blakey, R. J., van Rooyn, Z. 2011, South African Avacado Growers Association Yearbook, Vol. 34.