JP7688907B2 - Sugar content measurement method, sugar content measurement device, and sugar content measurement program - Google Patents

Description

The present invention relates to a sugar content measurement method, a sugar content measurement device, and a sugar content measurement program.

For example, the non-destructive sugar content measuring device in Patent Document 1 uses a single or multiple light sources that emit light of three wavelengths in the range of 860 nm to 960 nm, and the detector that detects the light absorption is positioned so that it is located at a position other than the position on the extension line of the straight line connecting the center point of the irradiated area on the surface of the fruit or vegetable of the incident light emitted from the light source and incident on the fruit or vegetable and the center of the fruit or vegetable, and so that the irradiated area on the surface of the incident light does not overlap with the detection area on the surface of the fruit or vegetable of the emitted light received by the detector from the fruit or vegetable.

However, in Patent Document 1, three types of light sources and explanatory variables are used: 860-890 nm, 900-920 nm, and over 920 to 960 nm. However, because each wavelength range is broad, it is necessary to further consider the effective wavelengths in each wavelength range, and there is also the problem that the accuracy of non-destructive measurement of sugar content is low when three wavelength ranges are applied.

Patent No. 3346142 JP 2000-221134 A

Iwamoto, Journal of the Japanese Society for Food Technology, 27, pp. 464-471 (1980) S. Morimoto et al. , Near Infrared Spectroscopy: Proceedings of the 10th International Conference pp155-159 (2002) Ito, Morimoto, Journal of the Illuminating Engineering Society of Japan, 98, pp581-584 (2014)

The present invention has been made in consideration of the above, and aims to provide a sugar content measurement method, sugar content measurement device, and sugar content measurement program that are capable of measuring the sugar content of fruits and vegetables with high accuracy.

In order to solve the above problems and achieve the object, the present invention provides a sugar content measurement method that irradiates fruit or vegetables with light from a light source and receives the reflected light, including diffuse reflection, to measure the sugar content of the fruit or vegetables, and is characterized by including an acquisition step of irradiating the fruit or vegetables with light in the near-infrared wavelength range from one light source or multiple light sources and acquiring light signal values of a spectral absorbance spectrum of four different wavelengths or a wavelength range including these four wavelengths from the reflected light, including diffuse reflection, and an estimation model creation step of performing multivariate analysis using the acquired absorbances (light signal values) of the four different wavelengths as explanatory variables to create an estimation model for estimating sugar content.

Furthermore, according to one aspect of the present invention, the four different wavelengths may be 856±2 nm, 876±2 nm or 884±2 nm, 900 to 918 nm, and 926±2 nm.

According to one aspect of the present invention, in the estimation model creation step, a multiple regression analysis may be performed as the multivariate analysis to create a multiple regression equation as the estimation model.

According to one aspect of the present invention, the fruits and vegetables may include the equator portion, the flower scar portion, processed products, and crushed products of the fruits and vegetables.

According to one aspect of the present invention, the fruits and vegetables may include pears, persimmons, mangoes, strawberries, paprika, mandarins (including tangerines), shiranui, tomatoes, cherries, peaches, pears, apples, plums, and melons.

According to one aspect of the present invention, the light source may be a ring-shaped light source.

According to one aspect of the present invention, in the acquisition process, reflected light, including diffuse reflection from fruits and vegetables, of the light irradiated from the ring-shaped light source may be detected at approximately the center of the ring-shaped light source.

According to one aspect of the present invention, the one light source may be a halogen lamp.

According to one aspect of the present invention, the multiple light sources may be multiple LEDs with different emission wavelengths.

According to one aspect of the present invention, the LEDs may be arranged at a predetermined angle (however, the predetermined angle is greater than 10 degrees) with respect to a plane on the side of the fruit or vegetable being measured.

In order to solve the above-mentioned problems and achieve the object, the present invention provides a sugar content measuring device that irradiates fruit or vegetables with light from a light source and receives the reflected light, including diffuse reflection, to measure the sugar content of the fruit or vegetables, and is characterized by comprising: a spectroscopic detection means that irradiates the fruit or vegetables with light in the near-infrared wavelength range from one light source or multiple light sources and acquires light signal values of a spectral absorbance spectrum of four different wavelengths or a wavelength range including these four wavelengths from the reflected light, including diffuse reflection; and an estimation model creation means that creates an estimation model for estimating sugar content by performing multivariate analysis using the acquired absorbances (light signal values) of the four different wavelengths as explanatory variables.

Furthermore, according to one aspect of the present invention, the four different wavelengths may be 856±2 nm, 876±2 nm or 884±2 nm, 900 to 918 nm, and 926±2 nm.

According to one aspect of the present invention, the estimation model creation means may perform a multiple regression analysis as the multivariate analysis and create a multiple regression equation as the estimation model.

According to one aspect of the present invention, the fruits and vegetables may include the equator portion, the flower scar portion, the processed product, and the crushed product of the fruits and vegetables.

According to one aspect of the present invention, the fruits and vegetables may include pears, persimmons, mangoes, strawberries, paprika, mandarins (including tangerines), shiranui, tomatoes, cherries, peaches, pears, apples, plums, and melons.

According to one aspect of the present invention, the light source may be a ring-shaped light source.

According to one aspect of the present invention, the spectroscopic detection means may detect reflected light, including diffuse reflection from fruits and vegetables, of the light irradiated from the ring-shaped light source at approximately the center of the ring-shaped light source.

According to one aspect of the present invention, the one light source may be a halogen lamp.

According to one aspect of the present invention, the multiple light sources may be multiple LEDs with different emission wavelengths.

According to one aspect of the present invention, the LEDs may be arranged at a predetermined angle (however, the predetermined angle is greater than 10 degrees) with respect to a plane on the side of the fruit or vegetable being measured.

In order to solve the above-mentioned problems and achieve the object, the present invention is a sugar content measurement program for measuring the sugar content of fruits and vegetables, which includes an acquisition step of acquiring light signal values of a spectral absorbance spectrum of four different wavelengths or a wavelength range including these four wavelengths from reflected light, including diffuse reflection, from fruits and vegetables in response to light in the near-infrared wavelength range irradiated from one light source or multiple light sources, and an estimation model creation step of creating an estimation model for estimating sugar content by performing multivariate analysis using the acquired four absorbances (light signal values) of different wavelengths as explanatory variables, and a sugar content measurement program for causing a computer to execute the process.

This invention has the effect of making it possible to measure the sugar content of fruits and vegetables with high accuracy.

FIG. 1 is an explanatory diagram for explaining an outline of the present invention. FIG. 2 is a flowchart for explaining the sugar content measuring method according to the present embodiment. FIG. 3 is a diagram showing an example of the external configuration of a sugar content measuring device according to this embodiment. FIG. 4 is a block diagram showing an example of the configuration of a sugar content measuring device according to this embodiment. FIG. 5 is a diagram for explaining the main parts of a high-precision practical machine. FIG. 6-A is a diagram for explaining the main parts of the LED prototype, and is a schematic plan view of the LED prototype as seen from above. FIG. 6-B is a diagram for explaining the main parts of the LED prototype, and is a schematic cross-sectional view of the LED prototype. FIG. 6-C is a diagram for explaining the main parts of the LED prototype, and is a schematic diagram for explaining the path of light irradiated from the LED when a sample is set on the sample stage of the LED prototype. FIG. 6-D is a diagram for explaining the main parts of the LED prototype, and is a schematic perspective view of the LED prototype. FIG. 7 is a diagram showing the measurement results of peaches using a high-precision practical machine. FIG. 8 is a diagram showing the measurement results of the peach LED prototype. FIG. 9 is a diagram showing the results of measurement of pears using a high-precision practical machine. FIG. 10 is a diagram showing the measurement results of the pear LED prototype. FIG. 11 is a diagram showing the measurement results of an apple using a high-precision practical machine. FIG. 12 is a diagram showing the measurement results of the LED prototype for apples. FIG. 13 is a diagram showing the measurement results of a pear using a high-precision practical machine. FIG. 14 shows the results of measurements of oysters taken using a high-precision practical machine. FIG. 15 shows the measurement results of plums using a high-precision practical machine. FIG. 16 is a diagram showing the measurement results of plums using the LED prototype. FIG. 17 shows the measurement results of mango using a high-precision practical machine. FIG. 18 is a diagram showing the measurement results of the equator of a tomato using a high-precision practical device. FIG. 19 shows the results of measurement of the flower scar of a tomato using a high-precision practical machine. FIG. 20 shows the measurement results of the tomato flower scar using the LED prototype. FIG. 21 shows the results of measuring strawberries using a high-precision practical machine. FIG. 22 is a diagram showing the measurement results of paprika using a high-precision practical machine. FIG. 23 is a diagram showing the measurement results of a melon using a high-precision practical machine. FIG. 24 is a diagram showing the measurement results of a melon using an LED prototype. FIG. 25 shows the results of measurement of the tangerine blossom scar using a high-precision practical machine. FIG. 26 shows the measurement results of the LED prototype at the equator of a mandarin orange. FIG. 27 shows the measurement results of the Shiranui high-precision practical instrument. FIG. 28 is a diagram showing the measurement results of cherries using the LED prototype. FIG. 29 shows the results of measurement of crushed tomatoes and processed tomato products using a high-precision practical machine.

Below, examples of preferred embodiments of the sugar content measurement method, sugar content measurement device, and sugar content measurement program according to the present invention will be described in detail with reference to Figs. 1 to 29. Note that the present invention is not limited to these embodiments.

[Outline of the present invention]
First, an outline of the present invention will be described with reference to Fig. 1. Fig. 1 is an explanatory diagram for explaining an outline of the present invention.

As mentioned above, Patent Document 1 uses three types of light sources, 860-890 nm, 900-920 nm, and over 920 to 960 nm, and explanatory variables (absorbance). However, because the wavelength ranges are wide, it is necessary to further consider the effective wavelengths in each wavelength range, and there is a problem that the accuracy of non-destructive measurement of sugar content is low when these three wavelength ranges are applied.

Therefore, in this invention, in addition to the three wavelength ranges, we also considered a combination that uses a fourth wavelength range.

In the present invention, it has been discovered that sugar content can be measured non-destructively with high accuracy by combining the absorbance (light signal value) at 876±2 nm or 884±2 nm, 900-918 nm, and 926±2 nm as a combination of explanatory variables with the absorbance (light signal value) at 856±2 nm as a fourth explanatory variable.

Furthermore, in this invention, it has been discovered that sugar content can be measured non-destructively with higher accuracy by using a ring-shaped light source, irradiating a ring-shaped light from the ring-shaped light source onto the fruit or vegetable, and detecting the reflected light, including diffuse reflection, from the fruit or vegetable at approximately the center of the ring-shaped light source.

On the other hand, LEDs (Light Emitting Diodes), which are becoming popular as a new light source, are characterized by their small size, long life, and low power consumption. Conventionally, near-infrared spectrophotometers capable of high-precision non-destructive measurement of sugar content (non-contact measurement using diffuse reflected light measurement mode (800-1000 nm): Kubota Corporation K-BA100R (high-precision practical model)) use a ring-shaped halogen lamp as a light source, irradiate the sample with a ring-shaped light, and detect the reflected light, including the diffuse reflection, at the center of the light source and transmit it to the main body using optical fibers.

However, with the widespread use of LEDs, halogen lamps may become scarce. Also, if optical fibers are used to guide the light, the volume of the device increases accordingly. On the other hand, MEMS (Micro Electro Mechanical Systems) technology has made it possible to manufacture ultra-compact, low-cost spectroscopes.

Therefore, we developed a compact near-infrared spectrophotometer (LED prototype) that does not use optical fiber and has a super-compact spectrometer built into the center of an LED ring light source, with the sample stage placed directly above the spectrometer.

Conventionally, in order to measure fruit sugar content nondestructively using near-infrared spectroscopy, preprocessing has been customarily performed on the nondestructively measured light absorption spectrum. Second-order differentiation has been the most widely used preprocessing method, but there are many other preprocessing methods, such as centering and first-order differentiation, and it is necessary to consider not only the differentiation conditions but also the combination of preprocessing methods. In addition, high-precision nondestructive measurement is possible by measuring small information with good reproducibility (see Non-Patent Document 1), but while preprocessing of light absorption spectra (differentiation, smoothing) has the effect of removing noise, it also removes part of the signal at the same time. However, the need to apply spectral preprocessing has decreased due to the improvement in the performance of spectrophotometers.

In the research and development of calibration curves for nondestructive measurement (also called "estimated models" or "model formulas"), methods that use a large number of explanatory variables, such as PLS regression analysis, have become mainstream, but it is difficult to see the completeness of the model formula. However, when a model formula is developed with a small number of explanatory variables, such as multiple regression analysis, the structure of the model formula is easier to explain and is highly practical.

In addition, it is known that near-infrared light absorption spectra are affected by temperature, and in practice, models are developed by non-destructively measuring fruits with different sample temperatures to reduce the effect of temperature (see Non-Patent Document 2).

Therefore, using the aforementioned high-precision practical device and LED prototype device, we discovered a new combination of explanatory variables that can non-destructively estimate the sugar content of fruits and vegetables without pre-processing such as differentiating or smoothing the light absorption spectra obtained by non-destructively measuring fruits and vegetables at different temperatures.

Except for watermelon, which was not examined, the same explanatory variables can be used for different types of fruit and vegetables, making it possible to design unified hardware and software. In addition, the signal of the measured near-infrared light absorption spectrum can be effectively used without being reduced.

The light-gathering ability of the high-precision practical machine deteriorates with small fruits (less than about 10 g), so it was necessary to develop a jig to improve the light-gathering ability (see Non-Patent Document 3). In addition, the measurement time can be as long as about 300 ms (milliseconds) for large fruits such as melons and mandarin oranges with thick skin. On the other hand, with the LED prototype machine, the installation angle of the LED is set at 10 degrees or more, preferably 32 degrees, with respect to the plane (for example, in Patent Document 2, the angle is set to 10 degrees so that the measurement light passes through the sample over a long distance), making it possible to perform non-destructive measurement of cherries, plums, peaches, pears, apples, tomatoes, melons, mandarin oranges, etc. in a shorter measurement time (5 to 30 ms) than with the high-precision practical machine.

In the present invention, the fruits and vegetables to be measured include the equator, flower scar, processed products, and crushed parts of the fruits and vegetables. The following examples are given of measuring pears, persimmons, mangoes, strawberries, peppers, mandarin oranges, shiranui, tomatoes, cherries, peaches, pears, apples, plums, and melons, but the present invention can be applied to other fruits and vegetables as well.

In FIG. 1, the present invention includes an estimation model creation process (S1) for creating an estimation model for estimating sugar content from three explanatory variables, namely, absorbance (light signal value) at 876±2 nm or 884±2 nm, 900-918 nm, and 926±2 nm of the target fruit or vegetable, as well as a fourth explanatory variable, absorbance (light signal value) at 856±2 nm, and a measurement process (S2) for measuring the light signal values at 876±2 nm or 884±2 nm, 900-918 nm, 926±2 nm, and 856±2 nm of the target fruit or vegetable, and applying the measured absorbance (light signal value) at 876±2 nm or 884±2 nm, 900-918 nm, 926±2 nm, and 856±2 nm to the estimation model to estimate sugar content.

In the estimation model creation step (S1), the light signal values of 876±2 nm or 884±2 nm, 900-918 nm, 926±2 nm, and 856±2 nm of the target fresh produce or fresh produce processed product are measured, and the absorbance (light signal value) is subjected to multivariate analysis to create an estimation model for estimating the sugar content in the fresh produce. Data preprocessing is performed on the absorbance (light signal value) as necessary. In the multivariate analysis, light signal values of 876±2 nm or 884±2 nm, 900-918 nm, 926±2 nm, and 856±2 nm are measured for known fresh produce, and a multivariate analysis is performed on the absorbance (light signal value) to create an estimation model.

In the measurement step (S2), data preprocessing is performed as necessary on the absorbance (light signal value) at 876±2 nm, 884±2 nm, 900-918 nm, 926±2 nm, and 856±2 nm for the fruits and vegetables whose sugar content is to be measured. The acquired absorbance (light signal value) is applied to an estimation model to estimate the sugar content. In this way, once an estimation model is created, it is possible to estimate (measure) the sugar content with high accuracy simply by acquiring the absorbance (light signal value) at 876±2 nm, 884±2 nm, 900-918 nm, 926±2 nm, and 856±2 nm for the fruits and vegetables to be measured.

[Sugar content measurement method]
The sugar content measuring method according to the present embodiment will be described with reference to Fig. 2. Fig. 2 is a flow chart for explaining the sugar content measuring method according to the present embodiment.

As shown in FIG. 2, the sugar content measurement method according to this embodiment is roughly divided into an estimation model creation process (S1) in which the absorbance (light signal value) of the target fruit or vegetable at 876±2 nm or 884±2 nm, 900-918 nm, 926±2 nm, and 856±2 nm is subjected to multivariate analysis to create an estimation model for estimating sugar content, and a measurement process (S2) in which the light signal values of the target fruit or vegetable at 876±2 nm or 884±2 nm, 900-918 nm, 926±2 nm, and 856±2 nm are measured, and the absorbance (light signal value) is applied to the estimation model to estimate sugar content.

Multivariate analysis includes regression analysis. Regression analysis creates a regression equation (calibration curve) that estimates a specific value, and can be used to estimate sugar content. Regression analysis includes PLS regression analysis and multiple regression analysis, and can be used to estimate sugar content values. In the multivariate analysis of this embodiment, a case where multiple regression analysis is used will be described, but the present invention is not limited to this, and discriminant analysis, principal component analysis, cluster analysis, etc. may also be used. Note that a model may be created by machine learning instead of multivariate analysis, or by incorporating machine learning into multivariate analysis.

The fruits and vegetables to be measured include fruits and vegetables, and the measurement areas include the equator, flower scar, processed products, and crushed products of the fruits and vegetables. Examples of the fruits and vegetables to be measured include pears, persimmons, mangoes, strawberries, peppers, mandarin oranges, shiranui, tomatoes, cherries, peaches, pears, apples, plums, and melons.

In the estimation model creation process, first, the fruit or vegetable to be measured is prepared in order to create the estimation model (step S11). The fruit or vegetable to be measured is set on the sample stage of the measurement unit, and near-infrared light (halogen light or multiple LED lights) is irradiated from the light source onto the fruit or vegetable, and the reflected light including the diffuse reflection is detected, and the signal value of the light of four different wavelengths in the near-infrared wavelength range (e.g., 876±2 nm or 884±2 nm, 900-918 nm, 926±2 nm, 856±2 nm) or a wavelength range including these four different wavelengths is obtained by the spectroscopic detection device (step S13). Here, "reflected light including diffuse reflected light" includes the case of only diffuse reflected light and the case of diffuse reflected light + reflected light other than diffuse reflected light (e.g., total reflected light). The signal value of the light is an electrical signal value (original data) obtained by photoelectric conversion of the reflected light. This signal value of the light can be converted into absorbance using a predetermined conversion formula. In the following pre-processing, multivariate analysis, and other processing, the light signal value may be used as is, or may be converted to absorbance before use. Reflected light includes both light reflected from the sample surface and light diffusely reflected inside the sample. The light source may be a ring-shaped light source. Reflected light, including diffuse reflection from the fruit or vegetable relative to the light irradiated from the ring-shaped light source, may be detected approximately at the center of the ring-shaped light source. The ring-shaped light source may be a halogen lamp. The ring-shaped light source may also be composed of multiple LEDs with different emission wavelengths. The multiple LEDs may be arranged on the side of the fruit or vegetable to be measured at a predetermined angle (greater than 10 degrees, preferably 32 degrees) with respect to the plane.

Next, data preprocessing is performed on the acquired absorbance (light signal value) as necessary (step S14). In the data preprocessing, signal processing calculations are performed using, for example, one or a combination of centering, standardization, normalization, and baseline correction.

On the other hand, the actual sugar content of the fruits and vegetables being measured is obtained (step S12). In step S15, a multivariate analysis is performed on the absorbance (light signal value) in the near-infrared wavelength range for unprocessed or preprocessed data to create an estimation model for estimating the sugar content in each of the fruits and vegetables.

When regression analysis is used as the multivariate analysis, for example, a regression equation (calibration curve) for estimating the sugar content value from the spectroscopic absorption spectrum is created as an estimation model. For example, multiple regression analysis or PLS regression analysis can be used as the regression analysis.

Next, in the measurement process, the fruit or vegetable (measurement target) whose sugar content is to be estimated is prepared (step S21). The fruit or vegetable to be measured is set on the sample stage of the measurement unit, and light in the near-infrared wavelength range (halogen light or multiple LED lights with different wavelength ranges) is irradiated from the light source onto the fruit or vegetable. The reflected light including the diffuse reflected light is detected, and the signal values of the light at four different wavelengths in the near-infrared wavelength range (e.g., 876±2 nm or 884±2 nm, 900-918 nm, 926±2 nm, 856±2 nm) or in a wavelength range including these four different wavelengths are measured and obtained by a spectroscopic detection device (step S22). Here, "reflected light including diffuse reflected light" includes cases where there is only diffuse reflected light, and cases where there is diffuse reflected light plus reflected light other than diffuse reflected light (e.g., total reflected light).

Next, data preprocessing is performed on the absorbance (light signal value) of the four different wavelengths as necessary (step S23). The data preprocessing is the same as in S14. The absorbance (light signal value) of the four different wavelengths or the wavelength range including these four different wavelengths that have been subjected to data preprocessing is applied to the estimation model for estimating sugar content created in step S15 to estimate the sugar content (step S24).

[Sugar content measuring device]
Next, the configuration of the sugar content measuring device of the present invention will be described by taking an embodiment as an example with reference to Figures 3 and 4. Note that the sugar content measuring device according to this embodiment can be suitably used in the sugar content measuring method described above, but the device used in the sugar content measuring method according to this embodiment is not limited to this.

Here, FIG. 3 is a schematic diagram showing an example of the external configuration of a sugar content device according to this embodiment. FIG. 4 is a block diagram showing an example of the configuration of a sugar content measuring device according to this embodiment, conceptually showing only the parts of the configuration that are related to the present invention.

As shown in Figures 3 and 4, the sugar content measuring device 1 includes a measuring unit 11 having a light source 12, a spectroscopic detection device 10, and a data processing device 20.

The data processing device 20 is a device that measures the sugar content of the fruit or vegetable to be measured from the spectral absorption spectrum of four different wavelengths in the near-infrared region or a wavelength range including these four different wavelengths acquired by the spectroscopic detection device 10. The data processing device 20 is equipped with a memory 21, a control unit 23, and a calculation processing unit 24, and the measurer inputs the measurement conditions, etc. into the data processing device 20 using the keyboard/mouse 22. The spectroscopic detection device 10 is connected to the measurement unit 11. The light source 12 is provided in the measurement unit 11. The light source 12 is a device that irradiates light of a predetermined wavelength (for example, a wavelength including a near-infrared wavelength range) to the fruit or vegetable to be measured that is set on the sample stage of the measurement unit 11. The light source 12 may be a ring-shaped light source. The reflected light, including the diffuse reflection of the fruit or vegetable against the light irradiated from the ring-shaped light source, may be detected at approximately the center of the ring-shaped light source. The ring-shaped light source may be composed of a halogen lamp. The ring-shaped light source may also be composed of multiple LEDs with different emission wavelengths. The multiple LEDs may be positioned at a predetermined angle (greater than 10 degrees, preferably 32 degrees) with respect to the plane on the side of the fruit or vegetable being measured.

The spectroscopic detection device 10 is connected to a measurement unit 11. The spectroscopic detection device 10 receives the light (including direct reflected light and diffuse reflected light) reflected by the fruit or vegetable product to be measured in response to the light irradiated from the light source 12, obtains the absorbance (light signal value) of four different wavelengths in the near-infrared light wavelength range (876±2 nm or 884±2 nm, 900-918 nm, 926±2 nm, 856±2 nm, or a wavelength range including these four different wavelengths), and transmits it to a data processing device 20. Spectroscopic analysis can be performed using various methods such as grating, filter method, pre-spectroscopic method, and post-spectroscopic method. In addition, the optical sensor can be an image sensor (polychromator), photodiode (monochromator), etc.

The measurement unit 11 is equipped with a light source 12, and irradiates the measurement object set on the sample stage with light source light input from the light source 12, and sends reflected light, including diffuse reflection, from the measurement object fruit or vegetable to the spectroscopic detection device 10.

The data processing device 20 can be configured, for example, as a personal computer, and is equipped with a memory 21, a control unit 23, and a calculation processing unit 24, and is connected to a keyboard/mouse 22, an I/O port (e.g., a USB port, etc.) 26, a display 30, etc.

The memory 21 stores the spectroscopic absorption spectrum transferred from the spectroscopic detection device 10 to the data processing device 20 and acquired by the spectroscopic absorption spectrum acquisition unit 24-1 of the calculation processing unit 24, the actual measured sugar content of fruits and vegetables used when creating an estimation model in the estimation model creation unit 24-2 of the calculation processing unit 24, and the created estimation model 27. The actual measured sugar content of fruits and vegetables used when creating the estimation model 27 can be input from the keyboard/mouse 22 or the I/O port 26.

The control unit 23 can perform control operations such as instructing the light source 12 to turn on/off the light irradiation and instructing the spectroscopic detection device 10 to start spectral detection in response to the operator's operation of the keyboard/mouse 22, and can also command the calculation processing unit 24 to perform processing.

The spectral absorption spectrum of the four different wavelengths of the fruit or vegetable transferred from the spectroscopic detection device 10, or the wavelength range including these four different wavelengths, is stored in the memory 21 of the data processing device 20. When the measurer issues a command to the calculation processing unit 24 to perform processing via the keyboard/mouse 22, the spectral absorption spectrum acquisition unit 24-1 of the calculation processing unit 24 first extracts the spectral absorption spectrum in the near-infrared wavelength range required for analysis from the spectral absorption spectrum stored in the memory 21, by data preprocessing as necessary.

The estimation model creation unit 24-2 of the calculation processing unit 24 performs multivariate analysis on the extracted spectroscopic absorption spectrum in the near-infrared light wavelength range to create an estimation model. In this embodiment, the estimation model creation unit 24-2 of the calculation processing unit 24 may create, for example, a regression equation (calibration curve) created by PLS regression analysis or multiple regression analysis as the multivariate analysis as the estimation model.

The estimation model creation unit 24-2 of the calculation processing unit 24 may store the results of the multivariate analysis (estimation model 27) in the memory 21, output them on the display 30, or print them out via a printer (not shown).

The estimation unit 24-3 of the calculation processing unit 24 estimates the sugar content of fruits and vegetables by applying the spectral absorption spectrum of the four different wavelengths in the near-infrared light wavelength range, or of a wavelength range including these four different wavelengths, extracted to the estimation model 27 stored in the memory 21. Note that by storing estimation models 27 for multiple types of fruits and vegetables in the memory 21, it becomes possible to measure the sugar content of multiple types of fruits and vegetables with a single measuring device. Note that the spectroscopic detection device 10 may be equipped with the functions of the data processing device 20, and the spectroscopic detection device 10 may execute the processing of the data processing device 20 to create the estimation model, etc.

[Example]
An embodiment will be described with reference to FIG. 6 to FIG. 29. In this embodiment, as the fruits and vegetables to be samples, pear, persimmon, mango, strawberry, paprika, mandarin orange, Shiranui, tomato, cherry, peach, pear, apple, plum, and melon were used. The spectroscopic absorption spectrum acquisition unit 24-1 acquired the spectroscopic absorption spectrum of four different wavelengths in the near-infrared wavelength range of the fruits and vegetables, or the wavelength range including these four different wavelengths. Then, the estimation model creation unit 24-2 performed multiple regression analysis using the four different absorbances (signal values) in the near-infrared wavelength range of the fruits and vegetables as explanatory variables to create multiple regression equations for estimating the sugar content of the fruits and vegetables. In addition, the multiple regression equations were evaluated. Furthermore, the estimation unit 24-3 used the created multiple regression equations to estimate the sugar content of the fruits and vegetables from the four different absorbances (signal values) in the near-infrared wavelength range of the fruits and vegetables.

(1. Sugar content measuring device)
For part or all of the sugar content measuring device 1, (1) a high-precision practical device and (2) an LED prototype device, which are near-infrared spectrophotometers using a diffuse reflection light non-contact measurement method, were used.

(1) High-precision practical instrument As the high-precision practical instrument, K-BA100R manufactured by Kubota Corporation was used. Figure 5 shows a perspective view of the main part of the high-precision practical instrument, with Figure 5(A) being a schematic perspective view of the measurement part and Figure 5(B) showing the cover. As shown in Figure 5(A), the measurement part of the high-precision practical instrument includes a main body part 41, a ring-shaped light source 12 (illuminating the sample from the halogen lamp of the main body via an optical fiber), a sample stage 42 made of a cushion on which the sample is set, and a light-receiving fiber (optical fiber) 43 that receives reflected light including diffuse reflection of the sample against the light in the near-infrared region irradiated from the light source 12 and outputs the measurement light to the main body. In addition, the measurement part of the high-precision practical instrument is provided with a cover 44 shown in Figure 5(B), which is removed when used for measurement.

The crushed tomatoes and processed tomato products were measured using a cell (e.g., Konica Minolta CR-A50) by filling the cell with the crushed tomatoes and processed tomato products and setting the cell on the sample stage 42. The absorbance was calculated from the received signal and linearly interpolated to obtain data at intervals of, for example, 2 nm.

(2) LED Prototype An LED prototype was produced. Figure 6 shows the main parts of the LED prototype. Figure 6-A shows the main parts of the LED prototype.
FIG. 6-A is a schematic plan view of the LED prototype seen from above, FIG. 6-B is a schematic cross-sectional view of the LED prototype, FIG. 6-C is a schematic diagram for explaining the path of light irradiated from the LED when a sample (e.g., a tomato) is set on the sample stage of the LED prototype, and FIG. 6-D is a schematic perspective view of the LED prototype.

In Figures 6-A to 6-D, the LED prototype is equipped with a ring-shaped main body 51, a ring-shaped light source 12 consisting of multiple LEDs provided on the upper part of the main body 51, a sample stage 52 provided inside the main body 51 on which the sample to be measured is set, and an ultra-compact spectroscope (spectroscopic detection device) 54 that receives the reflected light (diffuse reflected light) from the sample that is incident on the light entrance 53 of the main body 51.

The light source 12 is configured with three LEDs (center wavelengths: 850, 870, 940 nm) arranged alternately at 90 degrees in a ring shape, with a diameter of, for example, 36 mm. These three LEDs (center wavelengths: 850, 870, 940 nm) can produce continuous light with a peak in the wavelength range of approximately 800 nm to approximately 1000 nm, and can obtain the signal values of light with the four different spectral absorption spectra described above. Note that the central wavelengths of the three LEDs are not limited to these examples. The installation angle of each LED is set to α (for example, α = 32 degrees) degrees toward the sample stage 52 with respect to the plane (XY plane).

The sample stage 52 is configured by drilling a hole of 8.5 mm diameter in the center of a rubber plate (e.g., 3 mm thick and 24 mm diameter) so that the center of the hole is the center of the light entrance 53. For melons, measurements may be taken by placing a rubber washer (e.g., inner diameter 8 mm, outer diameter 18 mm, thickness 1.5 mm) on the sample stage 52.

The ultra-compact spectrometer 54 is a Hamamatsu Photonics C11708MA (640-1050 nm, maximum wavelength resolution of 20 nm, number of pixels of 256) and detects the measurement light incident from the light entrance port 53.

As shown in Figure 6-C, in the LED prototype configured as above, the measurement object (fruit or vegetable) is placed on the sample stage 52. Each LED of the light source 12 irradiates light from the main body 51 at a predetermined angle α (for example, α = 32 degrees) with respect to the horizontal plane (XY plane), and the reflected light (diffuse reflected light) from the measurement object is detected by the ultra-compact spectrometer 54 via the light entrance port 53.

The spectral measurement conditions were low gain, 3 or 1 average count, 1 measurement count, and a Labshere SRS-99-020 reference plate. Dark and reference were measured for 5 ms, while samples were measured for 10-30 ms when light collection was poor (plums, cherries, tomatoes). Additionally, no interpolation (no pre-processing) was performed on the optical absorption spectra obtained by non-destructive measurement of fruit and vegetables.

(2. Measurement Method)
(2-1. Sample, sample temperature, measurement location)
The sample temperature was changed to two stages (low temperature, room temperature) and at room temperature, the fruit equator (peach, pear, apple, European pear (lower part of fruit), persimmon, plum, mango, mandarin orange, Shiranui, tomato, strawberry, paprika) or flower scar (tomato including cherry tomato, melon, mandarin orange, cherry) was measured facing the light entrance of the sample stage using a high-precision practical machine or an LED prototype machine. A test was also conducted for non-destructive measurement of tomatoes by changing the product temperature to three stages.

For peaches, pears, apples, pears, persimmons, plums, mangoes, melons, peppers, and mandarins (equatorial measurements) non-destructive measurements were made; for tomatoes, strawberries, mandarins (flower scar measurements) and cherries, the whole fruit was measured; and for shiranui, the fruit was cut in half lengthwise, and the sugar content of each was measured destructively and used as the objective variable.

(2-2. Measurement of near-infrared spectrum)
Pears, persimmons, mangoes, strawberries, paprika, mandarin orange blossoms, shiranui, tomato equator (fruits weighing 10g or more), crushed tomatoes, and tomato products (tomato juice, tomato puree, tomato ketchup) were measured non-destructively using only the high-precision practical machine, while mandarin orange equator, cherries, and tomatoes including cherry tomatoes were measured non-destructively using only the LED prototype machine. Peaches, pears, apples, plums, and melons were measured non-destructively using both the high-precision practical machine and the LED prototype machine.

(2-3. Development and evaluation of calibration curves for non-destructive measurements)
The sample (with known sugar content) for creating the calibration curve is called the "nondestructive measurement method development sample", and the sample (with unknown sugar content) for evaluating the created calibration curve is called the "nondestructive measurement method evaluation sample". The data obtained from the nondestructive measurement method development sample was subjected to multiple regression analysis using Microsoft Excel to develop a nondestructive measurement calibration curve (multiple regression equation: estimation model), and the obtained calibration curve (multiple regression equation) was applied to an unknown sample (nondestructive measurement method evaluation sample) for evaluation. RMSE (Root Mean Squared Error) was calculated as an evaluation index.

The formula for calculating RMSE (nondestructive measurement accuracy of samples used for evaluating nondestructive measurement methods) is given by the following formula (1):

Where Y is the estimated sugar content by the non-destructive measurement method, X is the actual sugar content, and n is the number of samples for evaluation of the non-destructive measurement method.

(3. Measurement Results)
(3-1. Momo)
For measurements of white peaches and yellow peaches, samples were used for developing non-destructive measurement methods (high-precision practical machine n = 208, LED prototype machine n = 142) and for evaluating non-destructive measurement methods (high-precision practical machine n = 144, LED prototype machine n = 167).

Figure 7 shows the measurement results of peaches using a high-precision practical device. Figure 7 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (876, 902 nm), three (876, 902, 926 nm), and four (856, 876, 902, 926 nm). Figure 7 (B) shows the correlation (multiple regression analysis) between the actual sugar content value at the time of development and the non-destructive measurement value (estimated value). Figure 7 (C) shows the correlation between the actual sugar content value at the time of evaluation and the non-destructive measurement value (estimated value). In Figures 7 (B) and (C), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %]. Figure 7 (D) shows the results of multiple regression analysis and evaluation when the four explanatory variables (absorbance at each wavelength) are combinations of different wavelengths.

As shown in Figure 7 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 7 (B), when absorbance at 856, 876, 902, and 926 nm was used as explanatory variables for the sample for nondestructive measurement method development, the intercept was 10.9, the coefficients were 625.6, -1298.7, 886.5, and -216.4, and the correlation coefficient was 0.96. The RMSE for the sample for nondestructive measurement method evaluation shown in Figure 7 (C) was also good at 0.72. As shown in Figure 7 (D), even when the four explanatory variables (absorbance at each wavelength) were used in combination with different wavelengths, both the correlation coefficient and RMSE showed good results.

Figure 8 shows the measurement results of the LED prototype for peaches. Figure 8 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (876, 909 nm), three (876, 909, 926 nm), and four (857, 876, 909, 926 nm). Figure 8 (B) shows the correlation (multiple regression analysis) between the actual sugar content value at the time of development and the non-destructive measurement value (estimated value). Figure 8 (C) shows the correlation between the actual sugar content value at the time of evaluation and the non-destructive measurement value (estimated value). In Figures 8 (B) and (C), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %]. Figure 8 (D) shows the results of multiple regression analysis and evaluation when the four explanatory variables (absorbance at each wavelength) are combinations of different wavelengths.

As shown in Figure 8 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 8 (B), when the absorbance at 856.53 (857), 876.35 (876), 909.33 (909), and 926.42 (926) nm was used as explanatory variables for the nondestructive measurement method development sample, the intercept was 14.8, the coefficients were 519.3, -1049, 810.6, and -279.8, and the correlation coefficient was 0.94. The RMSE for the nondestructive measurement method evaluation sample shown in Figure 8 (C) was also good at 0.80. When ultimately selecting one of the multiple calibration curves developed, it is necessary to determine the X:Y = 1:1 line by taking into consideration not only the correlation coefficient and RMSE, but also the applicable sample and bias, just as with the high-precision practical device.

As shown in Figure 8 (D), even when the four explanatory variables (absorbance at each wavelength) were combined with different wavelengths, good results were obtained for both the correlation coefficient and RMSE.

(3-2. None)
For the measurements of red pears, samples (n=144) were used for developing nondestructive measurement methods, and red and green pears were used for evaluating nondestructive measurement methods (n=96).

Figure 9 shows the results of measurements of pears using a high-precision practical device. Figure 9 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (876, 910 nm), three (876, 910, 926 nm), and four (856, 876, 910, 926 nm). Figure 9 (B) shows the correlation (multiple regression analysis) between the actual sugar content measured during development and the non-destructive measurement value (estimated value). Figure 9 (C) shows the correlation between the actual sugar content measured during evaluation and the non-destructive measurement value (estimated value). In Figures 9 (B) and (C), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %].

As shown in Figure 9 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 9 (B), when the absorbance at 856, 876, 910, and 926 nm was used as an explanatory variable for the sample for nondestructive measurement method development, the intercept was 7.1, the coefficients were 502.3, -1015.4, 792.8, and -288.6, and the correlation coefficient was 0.96. As shown in Figure 9 (C), the RMSE for the sample for nondestructive measurement method evaluation was also good at 0.58.

Figure 10 shows the measurement results of the LED prototype for pears. Figure 10 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (876, 909 nm), three (876, 909, 926 nm), and four (857, 876, 909, 926 nm). Figure 10 (B) shows the correlation (multiple regression analysis) between the actual sugar content measured during development and the non-destructive measurement value (estimated value). Figure 10 (C) shows the correlation between the actual sugar content measured during evaluation and the non-destructive measurement value (estimated value). In Figures 10 (B) and (C), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %].

As shown in Figure 10 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 10 (B), when the absorbances at 856.53 (857), 876.35 (876), 909.33 (909), and 926.42 (926) nm were used as explanatory variables for the samples used in developing nondestructive measurement methods, the intercept was 12.9, the coefficients were 406.6, -877.7, 725.7, and -253.4, and the correlation coefficient was 0.86. As shown in Figure 10 (C), the RMSE for the samples used in evaluating nondestructive measurement methods was also good at 0.87.

(3-3. Apple)
For measurements of red, yellow, and green apples, samples were used for developing non-destructive measurement methods (n = 134) and for evaluating non-destructive measurement methods (high-precision practical machines, n = 164, LED prototype machines, n = 132).

Figure 11 shows the measurement results of apples using a high-precision practical device. Figure 11 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (876, 902 nm), three (876, 902, 926 nm), and four (856, 876, 902, 926 nm). Figure 11 (B) shows the correlation (multiple regression analysis) between the actual sugar content value at the time of development and the non-destructive measurement value (estimated value). Figure 11 (C) shows the correlation between the actual sugar content value at the time of evaluation and the non-destructive measurement value (estimated value). In Figures 11 (B) and (C), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %].

As shown in Figure 11 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 11 (B), when the absorbance at 856, 876, 902, and 926 nm was used as an explanatory variable for the samples for nondestructive measurement method development, the intercept was 14.0, the coefficients were 575.7, -1154.4, 780.1, -201.1, n = 134, and the correlation coefficient was 0.96. As shown in Figure 11 (C), the RMSE for the samples for nondestructive measurement method evaluation was also good at 0.48. The correlation coefficient was slightly lower at 902 nm (correlation coefficient 0.96) than at 910 nm (correlation coefficient 0.97), but as with peaches, 902 nm improved the slope (bias) of the samples for nondestructive measurement method evaluation.

Figure 12 shows the measurement results of apples using the LED prototype. Figure 12 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (876, 909 nm), three (876, 909, 926 nm), and four (857, 876, 909, 926 nm). Figure 12 (B) shows the correlation (multiple regression analysis) between the actual sugar content value at the time of development and the non-destructive measurement value (estimated value). Figure 12 (C) shows the correlation between the actual sugar content value at the time of evaluation and the non-destructive measurement value (estimated value). In Figures 12 (B) and (C), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %].

As shown in Figure 12 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 12 (B), when the absorbance at 856.53 (857), 876.35 (876), 909.33 (909), and 926.42 (926) nm was used as the explanatory variables for the nondestructive measurement method development sample, the intercept was 14.5, the coefficients were 378.1, -771.8, 609, and -214.3, and the correlation coefficient was 0.92. As shown in Figure 12 (C), the RMSE for the nondestructive measurement method evaluation sample was also good at 0.88. One of the reasons why the RMSE was slightly larger for the nondestructive measurement method evaluation sample is thought to be because the number of averages when measuring the near-infrared light absorption spectrum was reduced from three to one.

(3-4. Pear)
For the measurements of pears, n = 120 samples were used for developing the nondestructive measurement method, and n = 100 samples were used for evaluating the nondestructive measurement method.

Figure 13 shows the measurement results of pears using a high-precision practical device. Figure 13 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (874, 904 nm), three (874, 904, 926 nm), and four (856, 874, 904, 926 nm). Figure 13 (B) shows the correlation (multiple regression analysis) between the actual sugar content value at the time of development and the non-destructive measurement value (estimated value). Figure 13 (C) shows the correlation between the actual sugar content value at the time of evaluation and the non-destructive measurement value (estimated value). In Figures 13 (B) and (C), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %].

As shown in Figure 13 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 13 (B), when absorbance at 856, 874, 904, and 926 nm was used as an explanatory variable in the samples for developing the nondestructive measurement method, the intercept was 14.1, the coefficients were 609.2, -1086.6, 681.2, and -204.8, and the correlation coefficient was 0.94. As shown in Figure 13 (C), the Aurora variety is a fruit from which it is difficult to obtain juice when checking the sugar content, and samples with large errors between the nondestructive measurement value (estimated value) were found, but even in the samples for evaluating the nondestructive measurement method, the RMSE was good at 0.74.

(3-5. Oysters)
For the measurements of oysters, n = 76 samples were used for developing non-destructive measurement methods, and n = 178 samples were used for evaluating non-destructive measurement methods.

Figure 14 shows the measurement results of persimmons using a high-precision practical device. Figure 14 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (886, 918 nm), three (886, 918, 928 nm), (856, 886, 918 nm), and four (856, 886, 918, 928 nm). Figures 14 (B) and (D) show the correlation (multiple regression analysis) between the actual sugar content value and the non-destructive measurement value (estimated value) during development. Figures 14 (C) and (E) show the correlation between the actual sugar content value and the non-destructive measurement value (estimated value) during evaluation. In Figures 14 (B), (C), (D), and (E), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %].

As shown in Figure 14 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 14 (B), when the absorbance at 856, 886, 918, and 928 nm was used as an explanatory variable in the sample for developing the nondestructive measurement method, the intercept was 15.4, the coefficient was 243.3, -689.6, 885, and -439.5, and the correlation coefficient was 0.97. As shown in Figure 14 (C), a relatively large error may occur with hard fruits, but it can also be applied to astringent persimmons, and the RMSE of the sample for evaluating the nondestructive measurement method was good at 1.02. An example of how spectrum preprocessing is effective when the optical absorption spectrum can be measured is shown below. When the second derivative values at 856, 876, 906, and 926 nm after the second derivative processing of the spectrum were used as explanatory variables, the correlation coefficient during development was 0.98 (see Figure 14 (D)) and the RMSE during evaluation was 0.86 (see Figure 14 (E)), improving the nondestructive measurement values.

(3-6. Plum)
For the measurements of plums, the samples for developing the nondestructive measurement method were a high-precision practical machine (n = 84), an LED prototype machine (n = 72), and the samples for evaluating the nondestructive measurement method (n = 100).

Figure 15 shows the measurement results of plums using a high-precision practical device. Figure 15 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (876, 900 nm), three (876, 900, 926 nm), four (856, 876, 900, 926 nm), and (856, 886, 900, 926 nm). Figure 15 (B) shows the correlation (multiple regression analysis) between the actual sugar content value at the time of development and the non-destructive measurement value (estimated value). Figure 15 (C) shows the correlation between the actual sugar content value at the time of evaluation and the non-destructive measurement value (estimated value). In Figures 15 (B) and (C), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %].

As shown in Figure 15 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 15 (B), when the absorbance at 856, 876, 900, and 926 nm was used as an explanatory variable for the sample for developing the nondestructive measurement method, the intercept was 14.1, the coefficients were 1522.3, -3098.5, 1998.3, and -421.9, and the correlation coefficient was 0.97, which were good results. As shown in Figure 15 (C), the RMSE for the sample for evaluating the nondestructive measurement method was also good at 0.66.

Figure 16 shows the measurement results of plums using the LED prototype. Figure 16 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (886, 911 nm), three (886, 911, 928 nm), and four (855, 886, 911, 928 nm). Figures 16 (B) and (D) show the correlation (multiple regression analysis) between the actual sugar content measured during development and the non-destructive measurement value (estimated value). Figures 16 (C) and (E) show the correlation between the actual sugar content measured during evaluation and the non-destructive measurement value (estimated value). In Figures 16 (B), (C), (D), and (E), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %].

As shown in Figure 16 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 16 (B), when the absorbance at 854.53 (855), 886.15 (886), 911.25 (911), and 928.31 (928) nm was used as the explanatory variable for the sample for developing the nondestructive measurement method, the intercept was 9.4, the coefficient was 339.4, -1462.5, 1505.6, and -393, and the correlation coefficient was 0.91, which were good results. As shown in Figure 16 (C), the RMSE was also good at 0.97 for the sample for evaluating the nondestructive measurement method. One of the reasons for the lower correlation coefficient compared to the high-precision practical machine is thought to be that the averaging number of times when measuring the near-infrared light absorption spectrum was set to one. An example in which spectrum preprocessing is effective when the light absorption spectrum can be measured like this is shown below. For samples used for evaluating nondestructive measurement methods, the nondestructive measurement values tended to be slightly higher (bias = 0.395), but when the same explanatory variables were used after the spectrum was processed with SNV (Standard Normal Variate), the correlation coefficient during development dropped slightly to 0.90 (see Figure 16 (D)), but the RMSE during evaluation was 0.96 and the bias was 0.281 (see Figure 16 (E)), improving the tendency for nondestructive measurement values to be higher.

(3-7. Mango)
For mango measurements, n = 52 samples were used for developing the nondestructive measurement method, and n = 34 samples were used for evaluating the nondestructive measurement method.

Figure 17 shows the results of measurements of mangoes using a high-precision practical device. Figure 17 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (876, 904 nm), three (876, 904, 924 nm), and four (858, 876, 904, 924 nm). Figure 17 (B) shows the correlation (multiple regression analysis) between the actual sugar content measured during development and the non-destructive measurement value (estimated value). Figure 17 (C) shows the correlation between the actual sugar content measured during evaluation and the non-destructive measurement value (estimated value). In Figures 17 (B) and (C), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %].

As shown in Figure 17 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 17 (B), when the absorbance at 858, 876, 904, and 924 nm was used as an explanatory variable for the sample for nondestructive measurement method development, the intercept was 13.3, the coefficients were 458, -893.8, 639.5, and -208, and the correlation coefficient was 0.95. Also, as shown in Figure 17 (C), the RMSE for the sample for nondestructive measurement method evaluation was good at 0.98.

(3-8. Tomato)
(1) Equatorial Measurement For measuring the equatorial part of tomatoes, the sample temperature was set at three levels, and tomatoes weighing approximately 10 g or more were used as samples for developing the nondestructive measurement method (n=94) and for evaluating the nondestructive measurement method (n=177).

Figure 18 shows the measurement results of the tomato equator using a high-precision practical device. Figure 18 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (876, 900 nm), three (876, 900, 926 nm), and four (856, 876, 900, 926 nm). Figure 18 (B) shows the correlation (multiple regression analysis) between the actual sugar content value at the time of development and the non-destructive measurement value (estimated value). Figure 18 (C) shows the correlation between the actual sugar content value at the time of evaluation and the non-destructive measurement value (estimated value). In Figures 18 (B) and (C), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %].

As shown in Figure 18 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 18 (B), when absorbance at 856, 876, 900, and 926 nm was used as explanatory variables for the samples for nondestructive measurement method development, the intercept was 8.5, the coefficients were 1525.3, -3181.3, 2079.5, and -419.4, and the correlation coefficient was 0.97. As shown in Figure 18 (C), the RMSE for the samples for nondestructive measurement method evaluation was also good at 0.61.

(2) Measurement of the flower scar For measuring the flower scar of tomatoes, samples for non-destructive measurement development were prepared using a high-precision practical device with three sample temperature levels and tomatoes weighing approximately 10 g or more (n = 72), and an LED prototype device with two sample temperature levels and including tomatoes weighing less than 10 g (n = 73).

Figure 19 shows the results of measurements of the tomato flower scar using a high-precision practical device. Figure 19 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (876, 902 nm), three (876, 902, 926 nm), (856, 876, 902 nm), and four (856, 876, 902, 926 nm). Figure 19 (B) shows the correlation (multiple regression analysis) between the actual sugar content value during development and the non-destructive measurement value (estimated value). In Figure 19 (B), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %].

As shown in Figure 19 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 19 (B), when the absorbance at 856, 876, 902, and 926 nm was used as an explanatory variable for the sample for developing a nondestructive measurement method, the intercept was 9.1, the coefficients were 1456.6, -2917, 1881.4, and -422.3, and the correlation coefficient was 0.95, which was good.

Figure 20 shows the measurement results of the LED prototype on the tomato flower scar. Figure 20 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (882, 907 nm), three (882, 907, 925 nm), four (855, 882, 907, 925 nm), and (855, 878, 907, 925 nm). Figure 20 (B) shows the correlation (multiple regression analysis) between the actual sugar content value and the non-destructive measurement value (estimated value) during development. In Figure 20 (B), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %].

As shown in Figure 20 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 20 (B), when the absorbance at 854.53 (855), 882.24 (882), 907.42 (907), and 924.54 (925) nm was used as the explanatory variable for the sample for developing the non-destructive measurement method, the intercept was 16.5, the coefficients were 644.4, -2101.4, 2149.7, and -675.1, and the correlation coefficient was 0.90, which was good. Unlike the high-precision practical machine, tomatoes weighing less than 10 g can also be measured in the same measurement time (5 ms). When the sugar content is low, the error may be large.

(3-9. Strawberries)
For the measurements of strawberries, n = 110 samples were used for developing the non-destructive measurement method, and n = 150 samples were used for evaluating the non-destructive measurement method.

Figure 21 shows the results of measurements of strawberries using a high-precision practical device. Figure 21 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (886, 902 nm), three (886, 902, 926 nm), four (856, 886, 902, 926 nm), and (856, 876, 902, 926 nm). Figure 21 (B) shows the correlation (multiple regression analysis) between the actual sugar content measured during development and the non-destructive measurement value (estimated value). Figure 21 (C) shows the correlation between the actual sugar content measured during evaluation and the non-destructive measurement value (estimated value). In Figures 21 (B) and (C), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %].

As shown in Figure 21 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 21 (B), when absorbance at 856, 886, 902, and 926 nm was used as explanatory variables for samples used in developing nondestructive measurement methods, the intercept was 8.5, the coefficients were 1525.3, -3181.3, 2079.5, and -419.4, and the correlation coefficient was 0.94. As shown in Figure 21 (C), the RMSE for samples used in evaluating nondestructive measurement methods was also good at 0.49.

(3-10. Paprika)
For the measurements of paprika, n = 112 samples were used for developing the nondestructive measurement method, and n = 174 samples were used for evaluating the nondestructive measurement method.

Figure 22 shows the measurement results of paprika using a high-precision practical device. Figure 22 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (876, 902 nm), three (876, 902, 926 nm), and four (856, 876, 902, 926 nm). Figure 22 (B) shows the correlation (multiple regression analysis) between the actual sugar content value at the time of development and the non-destructive measurement value (estimated value). Figure 22 (C) shows the correlation between the actual sugar content value at the time of evaluation and the non-destructive measurement value (estimated value). In Figures 22 (B) and (C), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %].

As shown in Figure 22 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 22 (B), when absorbance at 856, 876, 902, and 926 nm was used as an explanatory variable for the sample for developing the nondestructive measurement method, the intercept was 7.9, the coefficients were 1797.4, -3605.4, 2311.3, and -504.2, and the correlation coefficient was 0.88. As shown in Figure 22 (C), the RMSE for the sample for evaluating the nondestructive measurement method was also good at 0.59.

(3-11. Melon)
For the melon measurements, the samples for developing the nondestructive measurement method were the high-precision practical machine n = 37 and the LED prototype machine n = 42, and the samples for evaluating the nondestructive measurement method were the high-precision practical machine n = 52.

Figure 23 shows the measurement results of melons using a high-precision practical device. Figure 23 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (876, 902 nm), three (876, 902, 926 nm), four (856, 876, 902, 926 nm), and (856, 884, 902, 926 nm). Figure 23 (B) shows the correlation (multiple regression analysis) between the actual sugar content value at the time of development and the non-destructive measurement value (estimated value). Figure 23 (C) shows the correlation between the actual sugar content value at the time of evaluation and the non-destructive measurement value (estimated value). In Figures 23 (B) and (C), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %].

As shown in Figure 23 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 23 (B), when the absorbance at 856, 876, 902, and 926 nm was used as an explanatory variable for the sample for nondestructive measurement method development, the intercept was 9.8, the coefficients were 575.8, -1223.7, 850.5, and -205.5, and the correlation coefficient was 0.83. As shown in Figure 23 (C), the RMSE for the sample for nondestructive measurement method evaluation was also good at 1.21.

Figure 24 shows the measurement results of melons using the LED prototype. Figure 24 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (886, 904 nm), three (886, 904, 926 nm), four (857, 886, 904, 926 nm), and (857, 874, 904, 926 nm). Figure 24 (B) shows the correlation (multiple regression analysis) between the actual sugar content values and non-destructive measurement values (estimated values) during development. In Figure 24 (B), the horizontal axis shows the actual sugar content values [Brix %], and the vertical axis shows the non-destructive measurement values (estimated values) of sugar content [Brix %].

As shown in Figure 24 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

Since there were melons where a gap occurred between the measurement site, the flower scar, and the sample stage, a rubber washer was placed on the sample stage 52 and measurements were taken. As a result, as shown in Figure 24 (B), when the absorbance at 856.53 (857), 886.15 (886), 903.58 (904), and 926.43 (926) nm was used as the explanatory variables, good results were obtained with an intercept of 13.1, coefficients of 234.4, -937.6, 851.9, and -143.5, and a correlation coefficient of 0.84.

(3-12. Tangerine)
(1) Measurement of the flower scar When measuring the mandarin flower scar, n = 95 samples were used to develop a non-destructive measurement method.

Figure 25 shows the results of measurements of the tangerine blossom scar using a high-precision practical device. Figure 25(A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (874, 906 nm), three (874, 906, 926 nm), and four (856, 874, 906, 926 nm). Figure 25(B) shows the correlation (multiple regression analysis) between the actual sugar content value during development and the non-destructive measurement value (estimated value). In Figure 25(B), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %].

As shown in Figure 25 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 25 (B), when the absorbance at 856, 874, 906, and 926 nm was used as an explanatory variable for the sample for developing the nondestructive measurement method, the intercept was 10.2, the coefficients were 848.4, -1570.8, 1086.8, and -368.5, and the correlation coefficient was 0.96, which was good.

(2) Equatorial measurement For the measurement of the equatorial part of mandarin oranges, the number of samples for developing a non-destructive measurement method was n=72.

Figure 26 shows the measurement results of the LED prototype at the equator of a mandarin orange. Figure 26 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (878, 907 nm), three (878, 907, 928 nm), and four (857, 878, 907, 928 nm). Figure 26 (B) shows the correlation (multiple regression analysis) between the actual sugar content value during development and the non-destructive measurement value (estimated value). In Figure 26 (B), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %].

As shown in Figure 26 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 26 (B), when the absorbances at 856.53 (857), 878.32 (878), 907.42 (907), and 928.31 (928) nm were used as explanatory variables in the samples for developing nondestructive measurement methods, the intercept was 10.8, the coefficients were 587.5, -1586.5, 1384.6, and -387.4, and the correlation coefficient was 0.93, which was good.

(3-13. Shiranui)
In the measurement of Shiranui, the number of samples for developing non-destructive measurement methods was n=36.

Figure 27 shows the measurement results using Shiranui's high-precision practical device. Figure 27(A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (876, 908 nm), three (876, 908, 926 nm), (856, 876, 908 nm), and four (856, 876, 908, 926 nm). Figure 27(B) shows the correlation (multiple regression analysis) between the actual sugar content value and the non-destructive measurement value (estimated value) during development. In Figure 27(B), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %].

As shown in Figure 27 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 27 (B), when the absorbance at 856, 876, 908, and 926 nm was used as an explanatory variable for the sample for developing a nondestructive measurement method, the intercept was 5.8, the coefficients were 183.5, -418.5, 339, and -102.7, and the correlation coefficient was 0.82, which was good.

(3-14. Cherries)
For the measurements of cherries, n = 50 samples were used for developing the nondestructive measurement method, and n = 60 samples were used for evaluating the nondestructive measurement method.

Figure 28 shows the measurement results of cherries using the LED prototype. Figure 28 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (884, 909 nm), three (884, 909, 926 nm), four (857, 884, 909, 926 nm), and (857, 874, 909, 928 nm). Figure 28 (B) shows the correlation (multiple regression analysis) between the actual sugar content measured during development and the non-destructive measurement value (estimated value). Figure 28 (C) shows the correlation between the actual sugar content measured during evaluation and the non-destructive measurement value (estimated value). In Figures 28 (B) and (C), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the non-destructive measurement value (estimated value) of sugar content [Brix %].

As shown in Figure 28 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 28 (B), when the absorbances at 856.53 (857), 884.19 (884), 909.33 (909), and 926.43 (926) nm were used as explanatory variables for the samples used in developing nondestructive measurement methods, the intercept was 17.1, the coefficients were 925.7, -3121.7, 3003.1, -819.5, and the correlation coefficient was 0.96. As shown in Figure 28 (C), the RMSE for the samples used in evaluating nondestructive measurement methods was also good at 1.46.

(3-15. Crushed tomatoes and processed tomato products)
For the measurements of crushed tomatoes and processed tomato products, the number of samples for measurement method development was n=13 and the number of samples for measurement method evaluation was n=31.

Figure 29 shows the measurement results of crushed tomatoes and processed tomato products using a high-precision practical machine. Figure 29 (A) shows the results of multiple regression analysis when the explanatory variables (absorbance at each wavelength) are two (876, 902 nm), three (876, 902, 926 nm), and four (856, 876, 902, 926 nm). Figure 29 (B) shows the correlation (multiple regression analysis) between the actual sugar content value and the measured value (estimated value) during development. Figure 29 (C) shows the correlation between the actual sugar content value and the measured value (estimated value) during evaluation. In Figures 29 (B) and (C), the horizontal axis shows the actual sugar content value [Brix %], and the vertical axis shows the measured sugar content value (estimated value) [Brix %].

As shown in Figure 29 (A), when there were four explanatory variables, the correlation coefficient was higher than when there were two or three explanatory variables.

As shown in Figure 29 (B), when the absorbance at 856, 876, 902, and 926 nm was used as an explanatory variable in the measurement method development sample, the intercept was 10.6, the coefficients were 4681.3, -10265.9, 7346.7, and -1730.2, n = 13, and the correlation coefficient was 0.994. As shown in Figure 29 (C), the RMSE was also good at 0.98 in the measurement method evaluation sample.

(4. Summary)
When the three absorbances of 876±2 nm or 884±2 nm, 900-918 nm, and 926±2 nm were used as explanatory variables in a multiple regression equation, the correlation coefficients were 0.34-0.96, including high values, but mostly low values. In addition, when these three wavelengths were used as explanatory variables, the correlation coefficients were similar to those of the two wavelengths in persimmon and tomato flower scar measurements (high-precision practical machine). Therefore, when the absorbance of 856±2 nm was used as the fourth explanatory variable, the correlation coefficient improved to 0.82-0.994.

When the dark and reference values are constant, the signal values of the dispersed light can be used directly as explanatory variables without being converted to absorbance, etc. When a multiple regression equation for non-destructively estimating fruit sugar content is developed using these four explanatory variables with narrowed-down wavelengths, the signs of the coefficients are consistently plus, minus, plus, minus, from lowest wavelength, for both the high-precision practical machine and the LED prototype machine. In particular, the coefficient for absorbance in the 900-918 nm absorption band for carbohydrates (sugar) is positive, meaning that the higher the absorbance, the higher the sugar content. In addition, similarly high estimation accuracy results were obtained for crushed tomato fruit and processed tomato products, so these explanatory variables are highly universal.

When measuring small fruits such as plums, cherries, tomatoes and strawberries containing fruits weighing less than 10g, as well as persimmons and melons, 884±2 nm was found to provide better non-destructive measurement accuracy than 876±2 nm, or both wavelengths provided similar non-destructive measurement accuracy.

As mentioned above, both the high-precision practical machine and the LED prototype machine consistently give similar results, but some differences are also observed. As shown in the results, there is a possibility that the explanatory variables used may vary slightly depending on the test sample, spectrometer, number of averages, etc., and there were cases for peaches and apples where 902 nm was used with the high-precision practical machine but 909 nm was used with the LED prototype machine, and for plums, 900 nm was used with the high-precision practical machine and 911 nm was used with the LED prototype machine. Also, 918 nm was used for persimmons. In this way, it was necessary to cover a wide range of 900 to 918 nm for the sugar absorption band. On the other hand, there is a possibility of deviation of about ±2 nm for other wavelengths (856, 876, 884, 926 nm), as shown in the results.

When using a high-precision practical device to non-destructively measure small fruits (approximately less than 10g), melons, or mandarins, the light gathering ability is poor and the measurement time is long, at approximately 300 to 350 ms. On the other hand, with the LED prototype, the LED is installed at an angle of 32 degrees to the plane as mentioned above, improving the light gathering ability and making it possible to measure not only peaches, pears, and apples, but also tomatoes, melons, and mandarins, including cherries, plums, and cherry tomatoes, in a short measurement time (5 to 30 ms).

In addition, the LED prototype can measure small fruits such as cherries and cherry tomatoes in a short time without using special tools, unlike the high-precision practical machine. In melons, gaps can sometimes occur between the sample stage and the blossom scar of the fruit due to their shape, so it is necessary to install rubber washers on the sample stage and perform standardized non-destructive measurements to reduce the effects of ambient light. The blossom scar of mandarin oranges has more severe dents than the melon blossom scar. As the results show, the high-precision practical machine is more suitable for measuring the blossom scar of mandarin oranges, but by reducing the gap between the sample stage and the sample in this way, highly accurate non-destructive measurements are also possible when measuring the blossom scar of mandarin oranges using the LED prototype.

According to the present invention, by combining absorbance at 876±2 nm, 884±2 nm, 900-918 nm, and 926±2 nm with absorbance at 856±2 nm as a fourth explanatory variable (when the dark or reference value is constant, the spectroscopic signal value can be used directly as an explanatory variable), it is possible to measure sugar content with high accuracy.

The present invention can be used not only for non-destructive measurement of fruit sugar content, but also for estimating the sugar content of crushed tomatoes and tomato products (tomato juice, tomato puree, tomato ketchup), and the applicable samples are not limited to these.

[Other embodiments]
While the embodiments of the present invention have been described above, the present invention may be embodied in various different embodiments other than those described above within the scope of the technical ideas set forth in the claims.

Furthermore, among the processes described in the embodiments, all or part of the processes described as being performed automatically can be performed manually, or all or part of the processes described as being performed manually can be performed automatically using a known method.

In addition, the processing procedures, control procedures, specific names, registered data for each process, information including search conditions and other parameters, screen examples, and database configurations shown in the above documents and drawings may be changed as desired unless otherwise specified.

In addition, with regard to the sugar content measuring device 1, each component shown in the figure is a functional concept, and does not necessarily have to be physically configured as shown in the figure.

For example, all or any part of the processing functions of each device of the sugar content measuring device 1, particularly the processing functions performed by the calculation processing unit 24, may be realized by a CPU (Central Processing Unit) and a program interpreted and executed by the CPU, or may be realized as hardware using wired logic. The program is recorded on a recording medium described below, and is mechanically read by the sugar content measuring device 1 as necessary. That is, memory 21 such as a ROM or HD records a computer program that works together as an OS (Operating System) to give instructions to the CPU and perform various processes. This computer program is executed by being loaded into RAM, and works together with the CPU to form a control unit.

This computer program may also be stored in an application program server connected to the sugar content measuring device 1 via any network, and all or part of it may be downloaded as needed.

The program according to the present invention may be stored on a computer-readable recording medium, or may be configured as a program product. Here, the term "recording medium" includes any "portable physical medium" such as a memory card, USB memory, SD card, flexible disk, magneto-optical disk, ROM, EPROM, EEPROM, CD-ROM, MO, DVD, and Blu-ray (registered trademark) Disc.

A "program" is a data processing method written in any language or description method, and may be in any format, such as source code or binary code. Note that a "program" is not necessarily limited to a single configuration, but also includes a distributed configuration consisting of multiple modules or libraries, or a program that works in conjunction with a separate program, such as an OS (Operating System), to achieve its function. Note that the specific configurations and reading procedures for reading a recording medium in each device shown in the embodiments, as well as installation procedures after reading, can use well-known configurations and procedures.

The various databases stored in memory 21 are storage devices such as memory devices such as RAM and ROM, fixed disk devices such as hard disks, flexible disks, optical disks, etc., and store various programs, tables, databases, web page files, etc. used for various processes and website provision.

The sugar content measuring device 1 may be configured as a known information processing device such as a personal computer or a workstation, or may be configured by connecting any peripheral device to the information processing device. The sugar content measuring device 1 may also be realized by implementing software (including programs, data, etc.) that realizes the method of the present invention in the information processing device.

Furthermore, the specific form of distribution and integration of the devices is not limited to that shown in the figures, and all or part of them can be functionally or physically distributed and integrated in any unit depending on various additions, etc., or depending on the functional load. In other words, the above-mentioned embodiments may be implemented in any combination, or the embodiments may be implemented selectively.

REFERENCE SIGNS LIST 1 Sugar content measuring device 10 Spectroscopic detection device 11 Measuring unit 12 Light source 20 Data processing device 21 Memory 22 Keyboard/mouse 23 Control unit 24 Calculation processing unit 24-1 Spectroscopic absorption spectrum acquisition unit 24-2 Estimation model creation unit 24-3 Estimation unit 26 I/O port 27 Estimation model 30 Display 41 Main unit 42 Sample stage 43 Light receiving fiber 44 Cover 51 Main unit 52 Sample stage 53 Light entrance 54 Ultra-compact spectrometer

Claims (21)

A method for measuring sugar content of fruits or vegetables by irradiating the fruits or vegetables with light from a light source and receiving reflected light including diffuse reflection, comprising:
an acquisition step of irradiating fruit or vegetables with light in the near-infrared wavelength range from one light source or multiple light sources, and acquiring signal values of light having only four different wavelengths of spectral absorption spectra from the reflected light, including diffuse reflection;
A process of creating an estimation model for estimating sugar content by performing multivariate analysis using only the four absorbance values (light signal values) obtained at different wavelengths as explanatory variables;
A method for measuring sugar content comprising the steps of: The sugar content measurement method according to claim 1, characterized in that the four different wavelengths are 856±2 nm, 876±2 nm or 884±2 nm, 900-918 nm, and 926±2 nm. The sugar content measurement method according to claim 1 or 2, characterized in that in the estimation model creation step, a multiple regression analysis is performed as the multivariate analysis to create a multiple regression equation as the estimation model. The method for measuring sugar content according to any one of claims 1 to 3, characterized in that the fruits and vegetables include the equator, flower scar, processed products, and crushed products of the fruits and vegetables. The method for measuring sugar content according to any one of claims 1 to 4, characterized in that the fruits and vegetables include pears, persimmons, mangoes, strawberries, paprikas, mandarin oranges (including tangerines), shiranui, tomatoes, cherries, peaches, pears, apples, plums, and melons. The sugar content measurement method according to any one of claims 1 to 5, characterized in that the light source is a ring-shaped light source. The sugar content measurement method according to claim 6, characterized in that in the acquisition process, reflected light, including diffuse reflection from fruits and vegetables, of the light irradiated from the ring-shaped light source is detected at approximately the center of the ring-shaped light source. The method for measuring sugar content according to claim 6 or 7, characterized in that the one light source is a halogen lamp. The method for measuring sugar content according to claim 6 or 7, characterized in that the multiple light sources are multiple LEDs with different emission wavelengths. The method for measuring sugar content according to claim 9, characterized in that the LEDs are arranged at a predetermined angle (however, the predetermined angle is greater than 10 degrees) with respect to a plane on the side of the fruit or vegetable to be measured. A sugar content measuring device that irradiates a fruit or vegetable with light from a light source and receives reflected light including diffuse reflection to measure the sugar content of the fruit or vegetable,
a spectroscopic detection means for irradiating fruit or vegetables with light in the near-infrared wavelength range from one light source or a plurality of light sources, and acquiring signal values of light having only four different wavelengths of spectral absorption spectra from the reflected light including diffuse reflection;
an estimation model creation means for creating an estimation model for estimating sugar content by performing multivariate analysis using only the four absorbances (light signal values) obtained at different wavelengths as explanatory variables;
A sugar content measuring device comprising: The sugar content measuring device according to claim 11, characterized in that the four different wavelengths are 856±2 nm, 876±2 nm or 884±2 nm, 900 to 918 nm, and 926±2 nm. The sugar content measuring device according to claim 11 or 12, characterized in that the estimation model creation means performs multiple regression analysis as the multivariate analysis and creates a multiple regression equation as the estimation model. The sugar content measuring device according to any one of claims 11 to 13, characterized in that the fruit or vegetable includes the equator, flower scar, processed products, and crushed products of the fruit or vegetable. The sugar content measuring device according to any one of claims 11 to 14, characterized in that the fruits and vegetables include pears, persimmons, mangoes, strawberries, paprika, mandarin oranges (including tangerines), shiranui, tomatoes, cherries, peaches, pears, apples, plums, and melons. The sugar content measuring device according to any one of claims 11 to 15, characterized in that the light source is a ring-shaped light source. The sugar content measuring device according to claim 16, characterized in that the spectroscopic detection means detects reflected light, including diffuse reflection from fruits and vegetables, of the light irradiated from the ring-shaped light source at approximately the center of the ring-shaped light source. The sugar content measuring device according to claim 16 or 17, characterized in that the one light source is a halogen lamp. The sugar content measuring device according to claim 16 or 17, characterized in that the multiple light sources are multiple LEDs with different emission wavelengths. The sugar content measuring device according to claim 19, characterized in that the LEDs are arranged at a predetermined angle (however, the predetermined angle is greater than 10 degrees) with respect to a plane on the side of the fruit or vegetable to be measured. A sugar content measurement program for measuring the sugar content of fruits and vegetables,
an acquisition step of acquiring signal values of light having only four different wavelengths of spectral absorption spectra from reflected light, including diffuse reflection, from fruits and vegetables in response to light in the near-infrared wavelength range irradiated from one light source or a plurality of light sources;
A process of creating an estimation model for estimating sugar content by performing multivariate analysis using only the four absorbance values (light signal values) obtained at different wavelengths as explanatory variables;
A sugar content measurement program to be executed by a computer.

Images