JP7654223B2 - Texture evaluation method, texture standard model - Google Patents

Description

The present invention relates to a texture evaluation method capable of evaluating the texture of various foods with high accuracy, and a standard texture model for performing texture evaluation.

The deliciousness of food is said to be mainly composed of three elements: flavor, aroma, and texture. Of these, texture is made up of extremely delicate and complex information. For example, texture, which is determined by a complex of factors such as stickiness and melt-in-the-mouth feel, is particularly difficult to quantify. For this reason, sensory evaluation of this type of texture is conducted by multiple evaluators.

In recent years, a method for evaluating food texture has been used, using a device with the same structure as the inside of the human mouth. For example, the tactile sensor in Patent Document 1 below has a flexible layer and a substrate layer laminated together, with a magnet embedded inside the flexible layer. In addition, on both sides of the substrate of the substrate layer, a magnetoresistance element (GMR) and an inductor are arranged on the element arrangement surface opposite the flexible layer arrangement surface that comes into contact with the flexible layer, and a circuit is formed on the element arrangement surface of the substrate.

Furthermore, a tooth-shaped hard member can be provided via a soft layer, and when a tooth-shaped member is provided, the tactile sensation applied to the teeth can be detected. In this way, the texture can be evaluated using a device that uses a tactile sensor (paragraphs 0010, 0012, 0019, 0025, Figure 1).

Patent No. 5187856

The tactile sensor in Patent Document 1 is a device that compares food texture based on the load and vibration generated by the element, and is capable of evaluating food texture such as softness and hardness. However, there is a problem that such a device alone does not provide an evaluation close to human senses and does not have sufficient accuracy in judgment.

The present invention has been made in consideration of these circumstances, and aims to provide a food texture evaluation method that can accurately evaluate the food texture of various evaluation foods.

In order to achieve the above-mentioned object, the food texture evaluation method of the present invention is a food texture evaluation method for performing a food texture evaluation on an evaluation food, and is characterized by comprising the steps of: creating a template consisting of pattern data, which is a plurality of virtual data serving as an index for the food texture evaluation; performing a sensory evaluation on a model food for the food texture evaluation and obtaining a sensory evaluation value; obtaining measurement data (A) corresponding to the pattern data of the model food and comparing the measurement data (A) with each of the pattern data of the template to calculate a similarity (S _A ); creating a standard texture model according to the type of texture from the sensory evaluation value and the similarity (S _A ); and performing a food texture evaluation of the evaluation food using the standard texture model.

In the food texture evaluation method of the present invention, a template is first created that is made up of a plurality of pattern data that serve as indices for food texture evaluation. Then, the measurement data (A) of a model food is compared with the pattern data to calculate the similarity (S _A ). This makes it possible to obtain information about which pattern data the model food is similar to.

In addition, a sensory evaluation is also performed on the model foods, and a standard texture model is created from the sensory evaluation value and the similarity (S _A ). Since the standard texture model combines the sensory evaluation value and the similarity (S _A ), the standard texture model is used to evaluate the texture of the evaluation foods. This method allows the evaluation of the texture of various evaluation foods with high accuracy.

In the food texture evaluation method of the present invention, it is preferable that the method comprises the steps of acquiring measurement data (B) corresponding to the pattern data of the food under evaluation, and comparing the measurement data (B) with each of the pattern data of the template to calculate a similarity (S _B ), and evaluating the similarity (S _B ) using the food texture standard model to perform a food texture evaluation of the food under evaluation.

According to this configuration, measurement data (B) is obtained for the evaluation food, and the measurement data (B) is compared with the pattern data to calculate the similarity (S _B ). This makes it possible to obtain information as to which pattern data the evaluation food is similar to. Then, the evaluation food having the similarity (S _B ) is evaluated using the standard food texture model. This method makes it possible to perform a food texture evaluation of the evaluation food.

In addition, in the food texture evaluation method of the present invention, it is preferable that the pattern data is obtained by arbitrarily varying the intensity of a specified parameter over time.

The template pattern data is, for example, virtual data obtained by using a program to arbitrarily vary the strength of parameters over time. By preparing multiple pieces of such pattern data, pattern data with a high degree of similarity can be extracted when comparing the measurement data with each pattern data. This makes it possible to use this method for evaluating the texture of foods under evaluation.

In addition, in the food texture evaluation method of the present invention, the step of calculating the similarity is preferably carried out by calculating the cumulative value of the distance between corresponding elements of the pattern data obtained by changing over time and the measurement data (A) or the measurement data (B).

The waveforms of the pattern data are obtained by changing the parameters over time, so the data are graphed. For example, a comparison between the measurement data (B) of the evaluation food and the pattern data can be obtained by calculating the cumulative distance between corresponding elements of the two. The smaller the distance and the more the two waveforms overlap, the higher the similarity can be evaluated.

In addition, in the food texture evaluation method of the present invention, it is preferable that the measurement data (A) includes load change data, which is the load change obtained when the model food is pressed. In that case, it is even more preferable that the measurement data (B) includes load change data, which is the load change obtained when the evaluation food is pressed.

Measurement data (A) and (B) include data on the change in load when the model food and the evaluation food are pressed, respectively. For example, if the evaluation food is pressed twice, the load changes between the first and second presses depending on the type of food. As the change in load is data specific to that food, this method can be used to improve the accuracy of texture evaluation.

In addition, in the food texture evaluation method of the present invention, it is preferable that the measurement data (A) includes vibration change data, which is a vibration change obtained when the model food is pressed. In that case, it is even more preferable that the measurement data (B) includes vibration change data, which is a vibration change obtained when the evaluation food is pressed.

The measurement data (A) and (B) may include vibration change data when the model food and the evaluation food are pressed, respectively. Since the vibration change data is also specific to the evaluation food being evaluated, this method can be used to improve the accuracy of food texture evaluation.

In the food texture evaluation method of the present invention, it is preferable that the standard food texture model is an evaluation prediction formula obtained by regression analysis with the similarity (S _A ) as an explanatory variable and the sensory evaluation value as a response variable.

The standard food texture model is an evaluation prediction formula obtained by regression analysis using the similarity (S _A ) and the sensory evaluation value. Therefore, in this method, a food model can be easily created and updated by inputting the similarity (S _A ) and the sensory evaluation value into the evaluation prediction formula using a program or the like.

In addition, in the texture evaluation method of the present invention, the regression analysis can be performed using any linear model, but it is particularly preferable to use a generalized linear mixed model (GLMM).

Regression analysis is a linear model including generalized linear model (GLM), generalized linear mixed model (GLMM), etc., and this technique enables the present method to easily and accurately evaluate the texture of the food being evaluated.

The standard texture model of the present invention is a standard texture model for conducting texture evaluation of evaluation foods, and is characterized by comprising: a template consisting of a plurality of pattern data serving as an index for the texture evaluation; a sensory evaluation value from a sensory evaluation conducted on a model food for the texture evaluation; a similarity (S _A ) obtained by comparing measurement data (A) corresponding to the pattern data of the model food with each of the pattern data of the template; and an evaluation prediction means for making an evaluation prediction according to the type of texture obtained from the sensory evaluation value and the similarity (S _A ).

The standard food texture model of the present invention has a template consisting of multiple pattern data serving as indices for food texture evaluation, and calculates the similarity ( _SA ) by comparing the measurement data (A) of a model food with the pattern data, thereby making it possible to obtain information as to which pattern data the model food is similar to.

The evaluation prediction means performs a sensory evaluation on the model foods and predicts the evaluation from the sensory evaluation value and the similarity (S _A ). This allows the standard food texture model to link the sensory evaluation value and the similarity (S _A ) and perform a food texture evaluation of the evaluation foods.

According to the present invention, it is possible to accurately evaluate the texture of various foods.

FIG. 1 is a diagram illustrating a food texture evaluation device used in the food texture evaluation method of the present invention. FIG. 2 is a bottom view of the circuit board of the food texture evaluation device. Load and vibration measurement results for sample S1. Load and vibration measurement results for sample S2. Load and vibration measurement results for sample S3. Load and vibration measurement results for sample S4. Load and vibration measurement results for sample S5. FIG. 1 is a diagram for explaining an outline of the food texture evaluation method of the present invention. FIG. 13 is a diagram showing a template (load pattern data). FIG. 13 is a diagram showing a template (vibration pattern data). FIG. 1 shows the average scores of the sensory evaluation of seven types of samples. Histogram of the average scores of the sensory evaluation of the seven samples. Measurement results (area) using a texture analyzer. Measurement results (number of peaks) using a texture analyzer. Texture analyzer measurements (average drop-off). Measurement results (peak value) using a texture analyzer. Loading of seven types of samples by the texture evaluation device. Vibration of seven types of samples using a texture evaluation device. The cumulative vibration sum of seven types of samples measured by the texture evaluation device.

Below, one embodiment of the food texture evaluation method of the present invention is described with reference to the drawings.

Figures 1 and 2 show a food texture evaluation device 10 used in the food texture evaluation method of the present invention. The food texture evaluation device 10 shown in Figure 1 has a configuration equivalent to human teeth, gums, etc. The food texture evaluation device 10 measures the occlusion of the evaluation food using various sensors in order to evaluate the texture of the evaluation food, which is the subject of evaluation.

The food texture evaluation device 10 is mainly composed of a contact part 11 corresponding to the teeth, an intermediate layer part 15 corresponding to the gums (periodontal ligament), and a base part 17 corresponding to the alveolar bone. The base part 17 is placed on a circuit board 19.

A hard material is used for the contact portion 11, and the contact portion 11 is, for example, wedge-shaped. Making the contact portion 11 wedge-shaped is preferable because it makes it easier to detect vibration. A permanent magnet 13 is embedded inside the contact portion 11. The change in magnetism when the evaluation food presses the contact portion 11 is acquired by a magnetic resistance element (GMR: Giant Magneto Resistive effect) 21 and an inductor element 23, which are texture sensors provided on the circuit board 19. The upper surface of the contact portion 11 may be processed to be flat so that it is easier to press the evaluation food.

The permanent magnet 13 may be a neodymium magnet (e.g., 15 mm in diameter, 1.5 mm in thickness, and 115 mT in magnetic flux density at the surface). The magnetic resistance element 21 can detect the change in position (amount of displacement) of the permanent magnet 13, and the change when the food under evaluation presses against the contact portion 11 is recorded as a "load."

When the position of the permanent magnet 13 is displaced, an induced electromotive force is generated in response to the change in magnetic flux density. The inductor element 23 detects this induced electromotive force. The magnitude of the induced electromotive force depends more on the speed of change in magnetic flux density than on the amount of change. When the food under evaluation presses against the contact portion 11, the change detected by the inductor element 23 is recorded as "vibration."

The intermediate layer 15 is made of elastic elastomer (natural rubber, synthetic rubber, etc.) or springs, while the base 17 is made of hard materials to replicate human teeth. For this reason, the values of "load" and "vibration" change depending on the type of food being evaluated.

Figure 2 also shows a view of the circuit board 19 from the underside.

As shown in the figure, eight magnetic resistance elements 21 are arranged on the underside of the base portion 17, indicated by the circular dashed line, with the center of the circle left empty. Since the permanent magnet 13 (not shown) of the contact portion 11 is located in the center of the circle, when the food under evaluation presses against the contact portion 11, the position change (load) of the permanent magnet 13 can be detected by any one of the magnetic resistance elements 21.

Furthermore, one inductor element 23 is placed in the center of the circle. This makes it possible to detect changes (vibrations) in the magnetic flux density of the permanent magnet 13 located almost directly above the inductor element 23. The number of magnetic resistance elements 21 and inductor elements 23 can be changed as appropriate. Furthermore, signals detected by the magnetic resistance elements 21 and inductor elements 23 are relayed by a connector 25 on the circuit board 19 and transmitted to a control unit (not shown).

Next, with reference to Figures 3A to 3E, an example of measuring load and vibration using a food texture sensor will be described.

The texture evaluator (hereafter referred to as "evaluator") used shrimp tempura (five types, samples S1 to S5), which are considered to have a high contribution of mechanical properties to texture, as the food to be evaluated, and performed eight measurements for each sample. Samples S1 to S5 are shrimp tempura with different batters, each cooked under different conditions.

In Figures 3A to 3E, the left vertical axis is Force (N), the right vertical axis is Voltage (V), and the horizontal axis is Time (s). Samples S1 to S5 corresponding to Figures 3A to 3E each have a different load waveform (bold line: load waveform data) and vibration waveform (solid line: vibration change data); for example, the measurement data for sample S1 in Figure 3A, sample S3 in Figure 3C, and sample S5 in Figure 3E show prominent load peaks at specified times.

As will be described in more detail below, the evaluator compares such measurement data with a template consisting of multiple time-series data prepared in advance to calculate the degree of similarity. Furthermore, the evaluator evaluates the texture of each sample (crispy, crunchy, crunchy, etc.) by comparing it with a standard texture model created from the measurement data and sensory evaluation values of the model foods.

Next, the food texture evaluation method of the present invention will be described in detail with reference to Figure 4. In addition, the content of the template and its creation will be further explained with reference to Figures 5A and 5B.

First, the evaluator uses model foods F _M (usually 6 to 7 types) to create a standard food texture model M. It is advisable to select multiple foods that have a texture similar to that of the evaluation food as the model foods F _M.

First, a sensory evaluation value λ is obtained for the model food F _M through a sensory evaluation by a sensory evaluator. For example, when performing a sensory evaluation of chicken nuggets, the same chicken nuggets are used as the model food F _M. The sensory evaluator then chews the model food F _M and evaluates whether it is closest to one of the following categories: "crispy,""crunchy,""crunchy,""crunchy," and "crunchy." The sensory evaluator may assign a "1" to the closest texture, a "2" to the second closest texture, and a "3" to the third closest texture, and then later convert them into numerical values.

Next, the evaluator measures the model food F _M with the food texture sensor. For this measurement, the above-mentioned food texture evaluation device 10 is used. Specifically, the evaluator presses the model food F _M against the contact part 11 of the food texture evaluation device 10 twice to obtain measurement data A of the load and vibration. The measurement data A is waveform data of the load and vibration (load change data and vibration change data) as shown in Figs. 3A to 3E.

Next, the evaluator compares the measurement data A with a previously prepared template T. Template T is composed of pattern data with different waveforms that serve as indices for food texture evaluation. The pattern data is virtual data obtained by generating random numbers to vary the intensity of a specified parameter arbitrarily (randomly) and over time.

There are templates T for load (see FIG. 5A) and vibration (see FIG. 5B), and each waveform is one pattern data. The evaluator compares the "load" of the measurement data A of the model food F _M with each of the pattern data in FIG. 5A, and determines the similarity based on the shape of the waveform. The evaluator also compares the "vibration" of the measurement data A of the model food F _M with each of the pattern data in FIG. 5B, and similarly determines the similarity based on the shape of the waveform. The similarity S _A obtained by this operation is used to create a standard food texture model M.

Here, in the measurement of the model food F 1 _M with the texture sensor, the model food F 1 _M is pressed against the contact portion 11 twice, and the pattern data for load shown in Fig. 5A has a waveform that changes over time accordingly. Specifically, peaks P1 and P2 of the pattern data correspond to the first pressing, and the pressure (Force [N]) is changed for each. Peaks P3 and P4 of the pattern data correspond to the second pressing, and the pressure is changed in the same way as for peaks P1 and P2. In this way, the evaluator creates 50 types of pattern data, each with a different peak p1 to p4.

5B shows pattern data for vibration. The vibrations detected by the food texture sensor (inductor element 23) have different occurrence timings even for the same type of model food F _M. For this reason, the evaluator creates 50 types of pattern data that change over time, with the cumulative sum of the number of times positive induced electromotive force of vibration (cumulative sum of vibration) on the vertical axis.

The similarity between the measurement data A of the model food F _M and the template T can be quantitatively evaluated using dynamic time warping (DTW). In this evaluation, the evaluator measures the distance between the two waveforms (DTW distance), and the more similar the two data are, the smaller the DTW distance (the higher the similarity) and the closer the waveform shapes are. Here, each waveform is composed of a time series of measurement values (load, etc.) sampled at regular intervals, and each point (element of the present invention) is continuously shown. To measure the DTW distance, the distance between the elements of each waveform is compared in a brute-force manner, and the path with the shortest distance between the points is found. Then, the cumulative value of the distances found is the DTW distance.

In addition, the similarity between the above-mentioned measurement data A and the template T can also be evaluated by measuring, for example, the Euclidean distance, the Fourier transform distance, the autoregressive coefficient distance or the EDR (Edit Distance on Real sequences) distance instead of the above-mentioned DTW distance.

Then, by linking the sensory evaluation value λ with the similarity S _A , for example, the pattern data of the texture evaluated by the sensory evaluator as “crispy” is determined. By accumulating such information, a standard texture model M that can be used to evaluate various textures is completed.

In fact, the standard food texture model M is given by the following evaluation prediction formula, where the explanatory variable is the similarity S _A (the comparison results with 50 types of pattern data S _a1 , S _a2 , ..., S _a50 ) and the objective variable is the sensory evaluation value λ.
log {λ/(1-λ)}=β ₀ +β ₁ s _a1 +β ₂ s _a2 +...+β ₅₀ s _a50 +r _i ... (Formula 1)
Here, β ₀ , β ₁ , β ₂ , ..., β ₅₀ are coefficients, and r _i are parameters representing individual differences.

Specifically, an evaluator uses a program or the like to substitute the similarity S _A and the sensory evaluation value λ into the evaluation prediction formula (Formula 1), and determines the coefficients β ₀ , β ₁ , β ₂ , ..., β ₅₀ (as well as the standard deviation of r _i ) through regression analysis, completing the standard food texture model M. Note that this regression analysis can be performed by any method as long as it can obtain the values of each parameter. Any linear model, such as a generalized linear mixed model (GLMM) or a generalized linear model (GLM), a nonlinear model, a hierarchical Bayesian model, etc., can be applied.

In addition, in order to evaluate six types of textures, "crispy", "crunchy", "crunchy", "crunchy", and "crunchy", there are six sensory evaluation values λ. In other words, there are six types of evaluation prediction formulas (Formula 1), and the evaluator creates a standard texture model M according to the texture by substituting the similarity S _A for each of them.

Next, in order to evaluate the evaluation food F _E , the evaluator uses the food texture evaluation device 10 to obtain measurement data B of the load and vibration of the evaluation food F _E. The load and vibration are measured under the same conditions as when measuring the model food F _M.

Next, the evaluator compares the measurement data B with the template T. Specifically, the evaluator compares the "load" and "vibration" of the measurement data B of the evaluation food F _E with each of the pattern data (50 types) of Figures 5A and 5B, respectively, and calculates the similarity based on the shape of the waveform. This operation obtains the similarity S _B (comparison results s _b1 , s _b2 , ... s _b50 with the 50 types of pattern data). Note that the similarity between the measurement data B of the evaluation food F _E and the template T can also be calculated using, for example, the dynamic time warping method (DTW).

Finally, the evaluator compares the similarity S _B of the evaluation food F _E with the previously prepared standard food texture model M. Specifically, the evaluator uses the following evaluation prediction formula to obtain the model evaluation value λ'.
log {λ'/(1-λ')}=β ₀ +β ₁ s _b1 +β ₂ s _b2 +...+β ₅₀ s _b50 +r _i ... (Formula 2)

This time, the evaluator substitutes the similarity S _B (s _b1 , s _b2 , ... s _b50 ) of the evaluation food into the evaluation prediction formula (Formula 2) to obtain the model evaluation value λ' (the coefficients β ₀ , β ₁ , β ₂ , ... , β ₅₀ , _ri use the values obtained when creating the standard food texture model M). A specific example of the model evaluation value λ' will be described later, but by the above series of operations, the texture of the evaluation food F _E can be evaluated.

Next, referring to Figures 6A and 6B, the results of a sensory evaluation using seven types of chicken nugget samples will be described.

First, Figure 6A shows the average scores (sensory evaluation values) of the sensory evaluations given by the sensory evaluators for Samples 1 to 7. There are six textures: "crispy," "quick," "crunchy," "crunchy," and "crunchy," and Samples 1 to 7 were each evaluated three times by five sensory evaluators (on a scale of 10). For example, Sample 2 had the highest score for "crunchy," and so it can be said that this was the sample that the sensory evaluators felt had that texture.

Here, a rough definition of each texture is given: "Crunchy" is the sensation of "the coating crumbling with light force, which persists even as you continue to chew". Additionally, "Sakuto" is the sensation of "the coating crumbling with light force during the first chew", and "Zakuzak" is the sensation of "when biting into with your back teeth, it crumbles into small pieces repeatedly". Furthermore, "Karuki" is the sensation of "something hard and firm crumbling, which persists even as you continue to chew", "Kari" is "the sensation of something hard crumbling with light force between your front teeth", and "Garigiri" is "the sensation of biting into ice cream".

In the figure, "Proba" is the test value used in one-way analysis of variance (ANOVA) to determine whether there is a possibility of a difference between each item, and "Significant" is the risk ratio of the sample with the most significant difference in the multiple comparison test (Tukey Test).

Next, Figure 6B shows a histogram of the average sensory evaluation scores for each texture for Samples 1 to 7. For Sample 1, the average scores for the "crispy" and "quick" textures were high, but the overall score was low, and it was judged to have "no distinctive features." A low absolute score means that the texture is weak; for example, if the score for "crispy" is low, the examiner would have judged the texture to be weak.

Based on the average scores, it was determined that Sample 2 had a strong "crunchy" texture, Sample 3 had a strong "crunchy" texture, and Sample 4 had a strong "crunchy" texture. Similarly, it was determined that Sample 5 had a strong "crispy" texture, Sample 6 had a strong "crunchy" texture, and Sample 7 had a strong "crispy" texture. In Sample 4, the "crispy" and "crunchy" textures received the same score, but since the "crispy" texture of Sample 7 was clearly evident, Sample 4 was determined to have a "crunchy" texture.

In addition, different symbols are assigned to samples that show significant differences (P<0.05) in multiple comparison tests. For example, there is a significant difference between "a," "b," and "c," but there is no significant difference between "a" and "ab," or between "ab" and "b."

Next, referring to Figures 7A to 8B, the results of an evaluator measuring the same samples 1 to 7 using a commercially available texture analyzer (TA.XTplus: manufactured by Stable Micro Systems) will be described. A texture analyzer is a device that measures numerical data related to the chewiness, texture, mouthfeel, etc. of food. The measurement of the chicken nuggets using the texture analyzer was repeated four times for each of samples 1 to 7.

First, Figure 7A shows the results of measuring the "area" of each sample. "Area" is the measurement of the surface area S, which is the product of time (Time [sec]) and stress (Force [g]), and the vertical axis of Figure 7A is S [g·sec]. The "area" value is the total rebound force when the texture analyzer presses into the sample, and is therefore generally an index of the elasticity and hardness of the sample. The line at the end of the histogram indicates the standard deviation of the measured values within the sample.

In the "area" measurements, the textures of "Sample 3 (crunchy)," "Sample 6 (crispy)," and "Sample 7 (crispy)" were high at about 1,000 [g・sec], but the other textures were also around 800 [g・sec], with no significant differences observed. Therefore, this was not very useful in evaluating the texture of chicken nuggets.

Next, Figure 7B shows the results of measuring the "number of peaks" for each sample. When the sample is pressed into the texture analyzer and breaks, a phenomenon occurs in which stress is released (a peak). The number of times this occurs is counted to give the "number of peaks." In measuring the "number of peaks," "Sample 6 (crispy)" and "Sample 7 (crunchy)" had high textures. Generally, a higher "number of peaks" value indicates a stronger "crispy" texture, so this is an indicator of this.

Next, Figure 8A shows the results of measuring the "average drop-off" of each sample. "Drop-off" is the value indicating how much stress (Force [g]) was released at the peak. Furthermore, "average drop-off" is the average value of all the drop-offs in one measurement. In measuring the "average drop-off," "Sample 7 (crisp)" had a high texture. Generally, "average drop-off" is an index of brittleness, and the higher the value, the more force is required to crush it.

Next, Figure 8B shows the results of measuring the "peak value" of each sample. The "peak value" is the average value of the maximum stress of each peak in one measurement. No significant differences were observed between textures in the "peak value," but it is generally an indicator of the hardness of the sample.

The sensory evaluation revealed clear differences between Sample 4 (crunchy) and Sample 6 (crispy) (see Figures 6A and 6B). However, in the evaluation using a texture analyzer, the evaluators performed a one-tailed, unpaired t-test on both samples at a significance level of 5% and found no significant differences in any of the following: Area (significance probability p = 0.058%), Number of peaks (significance probability p = 0.501%), Mean drop-off (significance probability p = 0.444%), and Peak value (significance probability p = 0.231%). As mentioned above, there was no clear difference in the measured values of both samples, and the difference in texture could not be fully evaluated.

Next, with reference to Figures 9A to 9C, the results of measuring the same samples 1 to 7 using the food texture evaluation device 10 of an embodiment of the present invention will be described.

Figures 9A to 9C show the results of measuring the "load," "vibration," and "cumulative vibration sum" of samples 1 to 7, respectively, using the food texture evaluation device 10. Specifically, the evaluator pressed each sample twice against the contact part 11 of the food texture evaluation device 10, and the load measurement data (load waveform data) and vibration measurement data (vibration waveform data) were obtained.

According to the "Load" data in Figure 9A, almost all of the waveforms were similar except for samples 5 and 7. However, according to the "Cumulative Vibration Sum" data in Figure 9C, although the waveforms of samples 2, 3 and 5 were similar, the waveforms of samples 1, 4, 6 and 7 could be distinguished from those of samples 2, 3 and 5, resulting in distinctive waveforms. The results showed that sample 7 had the largest "cumulative vibration sum" and sample 1 had the smallest.

Thereafter, the evaluator compared this load waveform data and vibration waveform data (measurement data A in FIG. 4) with each pattern data (50 types) of template T to obtain the similarity (similarity S _A in FIG. 4). Specifically, the evaluator measured the "load" and "cumulative vibration sum" 10 times for each of samples 1 to 7 using the food texture evaluation device 10, compared it with each pattern data (50 types) of template T to obtain the similarity S _A , and calculated the average value of the similarities S _A for the 10 times.

Next, the evaluators created six types of evaluation prediction equations (Equation 1) using a generalized linear mixed model (GLMM) from the obtained average similarity _SA and the average score of the previous sensory evaluation (see Figure 6A), and calculated the values of parameters _β0 , _β1 , _β2 , ..., _β50 , and r _i .

Next, the evaluator used another chicken nugget sample 3', which was produced and cooked through the same process as sample 3, and obtained load and vibration measurement data (measurement data B in Figure 4) using the texture evaluation device 10 under the same conditions as when measuring sample 3. The measurement of sample 3' was repeated 10 times.

After that, the evaluator compared these measurement data B with the respective pattern data (50 types) of the template T to obtain the similarity (similarity S _B in FIG. 4).

Furthermore, the evaluator calculated a model evaluation value λ' for sample 3' using the obtained similarity S _B and the values of parameters β ₀ , β ₁ , β ₂ , ..., β ₅₀ , _ri calculated from the evaluation prediction formula (Formula 1) for samples 1 to 7. Since the measurement of sample 3' was repeated 10 times, the evaluator calculated the average value of the model evaluation value λ' and used this as the evaluation result based on the food texture evaluation model.

Next, with reference to Table 1, we will explain the correlation between the evaluation results using the food texture evaluation model of an embodiment of the present invention and the sensory evaluation results.

Table 1 shows the evaluation value (sensory evaluation value λ) when sample 3 was subjected to a sensory evaluation and the evaluation value (model evaluation value λ') when sample 3' was evaluated using the food texture evaluation model M of this embodiment.

The evaluation values using the texture evaluation model M showed the same tendency as the sensory evaluation value λ, and the correlation coefficient between the sensory evaluation value λ and the model evaluation value λ' was 0.97. In general, a correlation coefficient with an absolute value exceeding 0.4 is considered to be correlated, and a correlation with an absolute value exceeding 0.7 is considered to be strong. Therefore, it can be said that there is a very strong correlation between sample 3 and sample 3'.

Similarly to sample 3', model evaluation values λ' were also obtained for samples 4' and 6' corresponding to samples 4 and 6. The correlation coefficient between the sensory evaluation value λ and the model evaluation value λ' for sample 4 was 0.86. Similarly, the correlation coefficient between the sensory evaluation value λ and the model evaluation value λ' for sample 6 was 0.87. Therefore, the model evaluation value λ' for both sample 5 and sample 6 was consistent with the sensory evaluation value λ, resulting in good results.

As mentioned above, measurements using the texture analyzer did not allow a clear distinction between Samples 4 and 6, but evaluation using the standard texture model M showed a significant difference in the evaluation values of the two samples, indicating that the differences in texture could be evaluated as distinguishable.

In this way, the food texture evaluation device 10 according to the present invention is able to measure the "load" and "vibration" of the evaluation food product F _E and combine these results to evaluate the food texture with high accuracy.

It is also possible to use the template T created this time to evaluate foods that have a texture relatively similar to that of the above-mentioned chicken nuggets (fried chicken, minced meat cutlets, spring rolls, croquettes, etc.). Furthermore, by using a different template, it is also possible to evaluate meat buns, fruit, etc., which have a completely different texture from chicken nuggets, making it possible to evaluate the texture of a wider range of different evaluation foods.

This time, 50 types of pattern data for template T were prepared, but the number is not limited to 50, as it depends on the food to be evaluated. Conventionally, for example, waveform data of a typical "crunchy" texture was created, and the waveform data was compared with the measurement data of the evaluation food F _E to evaluate the texture. However, with the food texture evaluation device 10, the degree of "crunchiness" can be obtained by preparing multiple pattern data corresponding to textures. In other words, a more detailed evaluation can be performed, such as "crunchy" close to "crispy" or "crunchy" close to "crunchy."

In addition, in the above embodiment, the evaluator presses the model food F _M and the evaluation food F _E against the contact portion 11 of the food texture evaluation device 10 twice to measure the data, but the number of presses may be just once, or may be three or more times.

Furthermore, in the above embodiment, a food texture evaluation device 10 was used that acquires measurement data through changes in magnetism, but, for example, a device that acquires measurement data through changes in vibration or sound may also be used, and a texture tester, rheometer, creep meter, etc. may also be applied.

S1 to S5, 1 to 7...samples, 10...food texture evaluation device, 11...contact part, 13...permanent magnet, 15...intermediate layer part, 17...base part, 19...circuit board, 21...magnetic resistance element, 23...inductor element, 25...connector.

Claims (11)

A food texture evaluation method for evaluating the food texture of an evaluation food, comprising:
creating a template consisting of a plurality of pattern data, the plurality of virtual data being indexes for the food texture evaluation;
a step of allocating a model food for the food texture evaluation using each of a plurality of samples , conducting a sensory evaluation of each model food , and obtaining a sensory evaluation value (λ) for each sample and for each type of food texture ;
acquiring measurement data (A) corresponding to the pattern data for each of the model foods for each sample , and comparing each of the measurement data (A) for each sample with each of the pattern data of the template to calculate a similarity (S _A ) for each sample ;
creating a standard texture model for each texture type from the sensory evaluation value (λ) for each sample and each texture type and the similarity (S _A ) for each sample ;
A step of evaluating the texture of the food product for evaluation using the standard texture model for each texture type ;
Equipped with
A food texture evaluation method, characterized in that the standard food texture model is an evaluation prediction formula obtained by regression analysis in which the similarity (S _A ) is included in explanatory variables and the sensory evaluation value (λ) is used as a response variable . acquiring measurement data (B) of the food product to be evaluated, and comparing the measurement data (B) with each of the pattern data of the template to calculate a similarity (S _B );
a step of evaluating the similarity (S _B ) using the standard texture model for each texture type and performing a texture evaluation of the evaluation food products for each texture type ;
The food texture evaluation method according to claim 1, further comprising: 3. The food texture evaluation method according to claim 2 , wherein the pattern data is obtained by arbitrarily varying the intensity of a predetermined parameter over time. The food texture evaluation method according to claim 3, characterized in that the step of calculating the similarity is performed by calculating a cumulative value of the distance between corresponding elements of the pattern data obtained by changing over time and the measurement data (A) or the measurement data (B). The food texture evaluation method according to any one of claims 1 to 4, characterized in that the measurement data (A) includes load change data, which is the load change obtained when the model food is pressed. The food texture evaluation method according to claim 5, characterized in that the measurement data (B) includes load change data, which is the load change obtained when the food for evaluation is pressed. The food texture evaluation method according to any one of claims 1 to 6, characterized in that the measurement data (A) includes vibration change data, which is a vibration change obtained when the model food is pressed. The food texture evaluation method according to claim 7, characterized in that the measurement data (B) includes vibration change data, which is a vibration change obtained when the food for evaluation is pressed. 9. The food texture evaluation method according to any one of claims 1 to 8 , wherein a linear model is applied in the regression analysis. The food texture evaluation method according to claim 9 , wherein a generalized linear mixed model (GLMM) is used for the regression analysis. A method for creating a standard texture model for evaluating the texture of a food product to be evaluated, comprising the steps of:
creating a template consisting of a plurality of pattern data, the plurality of virtual data being indexes for the food texture evaluation;
a step of allocating a model food for the food texture evaluation using each of a plurality of samples , conducting a sensory evaluation of each model food , and obtaining a sensory evaluation value (λ) for each sample and for each type of food texture ;
acquiring measurement data (A) corresponding to the pattern data for each of the model foods for each sample , and comparing each of the measurement data (A) for each sample with each of the pattern data of the template to calculate a similarity (S _A ) for each sample;
a step of creating a model for each type of texture from the sensory evaluation value (λ) for each sample and for each type of texture and the similarity (S _A ) for each sample, the model including the similarity (S _A ) as an explanatory variable and an evaluation prediction formula obtained by regression analysis with the sensory evaluation value (λ) as a target variable as the standard texture model;
A method for creating a standard food texture model, comprising:

Images