JP7015487B2 - Fried food texture evaluation method and fried food texture evaluation program - Google Patents

Description

Application of Article 30, Paragraph 2 of the Patent Law Proceedings of the 17th Annual Meeting of the Japanese Society of Kansei Engineering

The present invention relates to a fried food texture evaluation method and a fried food texture evaluation program.

The satisfaction of the meal is greatly contributed by the taste, and the improvement of the taste is the most important for the person who sells the food.

On the other hand, since the taste is only known after actually eating the food, how delicious it is for consumers before eating the food, specifically at the time of purchasing the food (if it is an ingredient). It is also important to make people feel how delicious they can be with their sense of smell and sight. In particular, from the viewpoint of vision, for example, in Non-Patent Document 1 below, it is reported that vision contributes not a little to the satisfaction of meals at the time of actual meals.

On the other hand, a technique for visually grasping the state of food is described in, for example, Non-Patent Document 2 below. The document discloses a technique for determining freshness using luminance distribution information in a fresh food image.

http: // www. fruitnet. com / fpj / article / 165945 / way-salad-presented-can-push-price-up-three-times-higher Tomohiro Masuda et al., "Examination of Freshness Judgment Model Using Brightness Distribution Information of Fresh Food Images", Japan Visual Society 2015 Winter Games Program, 3p21

Foods have various appearance characteristics, but as mentioned above, the appearance of foods can be one of the factors that increase consumers' purchasing motivation as a first impression. In particular, the likability of fried food is greatly affected by its texture. Therefore, the texture (texture) inferred from the appearance of fried food has a great influence on the value of the product. For example, in the case of tempura, a crispy texture is preferred, and a moist texture is not preferred. Therefore, a product with a strong crispy texture is preferred as a texture that can be inferred from the appearance. There is a need for technology for more objectively evaluating the texture of fried foods.

However, the technique described in Non-Patent Document 2 is intended only for the freshness of fresh foods, not for the texture of fried foods such as tempura, and uses luminance distribution information. However, there is a problem that the brightness alone largely depends on the state at the time of shooting such as the brightness and lacks the stability of judgment.

Therefore, in view of the above problems, it is an object of the present invention to provide a texture evaluation method for fried foods and a texture evaluation program for fried foods, which can evaluate the texture of fried foods such as tempura with excellent stability.

Regarding the above-mentioned problems, the present inventors have discovered that in fried foods such as tempura, the spatial frequency of the fried batter has a large correlation with the texture (texture) estimated from the appearance, and completes the present invention. I came to let you.

That is, the method for evaluating the texture of a fried food according to one aspect of the present invention is to perform spatial frequency analysis processing on the fried food image data.

Further, the texture evaluation program for fried food according to another aspect of the present invention is for causing a computer to perform spatial frequency analysis processing on the fried food image data.

As described above, according to the present invention, it is possible to provide a texture evaluation method for fried foods and a texture evaluation program for fried foods, which can evaluate the texture of fried foods such as tempura having excellent stability.

It is a figure which shows the schematic structure of the computer which executes this evaluation method. It is a figure which shows the example of the fried food image in the case of tempura. It is a figure which shows the spatial frequency analysis result in the state where the crispy feeling is strong and the state where the crispy feeling is weak. It is a figure which shows the spatial frequency analysis result in the state where the crunchy feeling is strong and the state where the crunchy feeling is weak. It is a figure which shows the spatial frequency analysis result in the state of strong moistness and the state of weak moistness. It is a figure which shows the summary of FIGS. 3 to 5.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. However, the present invention can be implemented in many different embodiments, and is not limited to the embodiments and examples shown below.

(Method of evaluating the texture of fried food)
First, the texture evaluation method for fried foods (hereinafter referred to as “the evaluation method”) according to the present embodiment is characterized in that spatial frequency analysis processing is performed on the fried food image data.

As described above, in this method, first, the fried food image data is processed. Here, the "fried food image data" is image data including information on the fried food obtained as a result of photographing the fried food, and is image data acquired by an image acquisition device such as a digital camera or a smartphone having a digital camera function, for example. be. When the fried image data is acquired by an image acquisition device such as a digital camera, it is a recording device such as a hard disk provided inside the device, or a flash memory, CD-ROM, DVD-ROM, etc. connected to the recording device as an external recording device. It is recorded in and processed as described below.

The fried image data is recorded and processed by a computer as electronically readable data (hereinafter referred to as "electronic data"). The computer that realizes this evaluation method is not limited, but is, for example, a central arithmetic unit (CPU), a recording device such as a hard disk or a flash memory, a recording device capable of recording electronic data for a relatively long period of time, a RAM, or the like. A memory device that temporarily records electronic data, an input device for inputting user instructions such as a keyboard and mouse, a display device such as a monitor for displaying information based on processed electronic data, and the above CPU. It is preferable to include a recording device, a memory device, a reading device for reading an external recording device such as a CD-ROM or a flash memory, a bus for electrically connecting an input device, a display device, and the like. FIG. 1 shows an outline of the configuration of a computer that realizes this evaluation method. The aspect of the computer may be a so-called desktop type, a notebook type, or a tablet type. Further, it may be a mobile phone having the above configuration, a so-called smartphone.

More specifically, in this evaluation method, a program for executing this evaluation method is recorded in advance in a recording device such as a hard disk, and the program is temporarily recorded in the memory device based on the user's operation. However, it can be realized through processing by the CPU. As is clear from the description so far, the texture evaluation program for the fried food for executing this evaluation method is for causing the computer to perform the spatial frequency analysis processing on the fried food image data.

As is clear from the above description, when the fried image data to be processed by this evaluation method is acquired by the image acquisition device, the data is transmitted / received by a wired cable or wireless communication, or is connected to an external device. The recording device or the like can be removed from the image acquisition device and recorded by the reading device of the computer. Of course, when the image acquisition device itself is a device having a function as a computer such as a smartphone, the evaluation method may be executed by the image acquisition device itself.

In this evaluation method, the fried food refers to a dish in which the ingredients are battered and fried in heated oil, and tempura is a preferable example as described above, but the fried food is not limited to this, for example, croquette, fried food, cutlet, and karaage. It is also possible to target fried chicken, fried chicken, etc. An image of a fried food image in the example of tempura is shown in FIG.

Further, in this method, "spatial frequency analysis" refers to analysis of a structure having a spatial period, and in this evaluation method, the specific spatial frequency analysis process is not limited, but for example, wavelet. Transformation processing, Fourier analysis processing, and the like can be exemplified.

Further, the spatial frequency analysis in this method is specifically performed based on the coordinate data (x coordinate data, y coordinate data) included in the fried food image data and the brightness data corresponding to the coordinate data, and the spatial frequency value data. And the power value data corresponding to the spatial frequency value data is obtained. Then, the texture evaluation value data is determined by referring to the power value data and the power value database recorded in advance.

Here, the "power value data" refers to data indicating a power value with respect to a spatial frequency, and this power value data is associated with "spatial frequency value data" indicating a spatial frequency value corresponding to this value. There is. Here, the unit of the spatial frequency value data is not particularly limited, but it is a preferable example to use cpi (cycle / image). In the case of spatial frequency evaluation by cpi, sufficient evaluation is possible if the shooting conditions are kept constant to some extent, but there is a possibility that the image is enlarged or reduced depending on the captured image. Therefore, cpd (cycle / assessment) may be used in consideration of the length of the width of the object to be imaged and the image-taking distance.

Further, in this evaluation method, the "power value database" is a set of power value data recorded in advance and serves as a reference for determining the texture evaluation value data. In this method, spatial frequencies are analyzed for each image group that has a significant difference in a specific texture, and the range where there is a significant difference by statistical processing is defined as the frequency band that represents the texture, and the impression, frequency band, and so on. By associating the power values into a database, it becomes possible to evaluate the texture. In principle, there is a difference in impression, and any texture can be evaluated as long as there is a significant difference in spatial frequency.

In this evaluation method, the power value data can be used for various textures as described above, and is not limited, but for example, in the case of fried foods, more specifically tempura, general consumers see the fried foods. In that case, the power value data when the "crispy feeling" is strongly felt and the corresponding spatial frequency value data, the power value data when the "crispy feeling" is strongly felt and the corresponding spatial frequency value data, "moist". It is equipped with power value data and corresponding spatial frequency value data when a strong feeling of "feeling" is provided, and how crispy the consumer feels when compared with the above-mentioned obtained power value data corresponding to specific spatial frequency value data. It can be exemplified to evaluate whether or not a feeling of crunchiness and moistness is felt. As a result, data indicating how much the consumer feels the texture of the fried food is created and output as texture evaluation value data. That is, the texture evaluation value data refers to data that can evaluate what kind of texture and to what extent the person who sees the fried food image feels the fried food image.

The power value data recorded in the power value database in this evaluation method is the spatial frequency analyzed so that it can be evaluated in any spatial frequency region of the power value data obtained as a result of the spatial frequency analysis of the fried food to be evaluated. The power value data of the entire region may be stored, but from the viewpoint of reducing the storage capacity and improving the calculation speed, only the power value data of a specific spatial frequency region is retained in each texture evaluation type. You may leave it alone. For example, in the case of tempura, it is preferable to hold power value data in the range of 0 to 90 cpi and 230 to 250 cpi in spatial frequency for each texture evaluation type, and more specifically, 0 to 0 to a crispy image. The spatial frequency band of about 50 cpi and the spatial frequency band of about 0 to 60 cpi are also considered for the image with a strong crunchy feeling, while the power value data of the spatial frequency band of about 0 to 90 cpi and 240 cpi is retained for the image with a strong moist feeling. It is more preferable to keep it.

As described above, the texture of fried food such as tempura can be evaluated with stability by this evaluation method. It is also possible to provide a texture evaluation program for fried foods to realize this evaluation method.

In addition, after using the above evaluation method, by making corrections to the acquired fried food image, modified fried food image data is created and used for verification of the above texture evaluation program to improve the reliability of the database. Is also possible.

Specifically, this method performs spatial frequency analysis processing on the fried food image data to obtain power value data, corrects the power value data, performs reverse spatial frequency analysis processing, and creates modified fried food image data. As a result, as described above, the reliability of the accumulated power value database can be further improved by verifying the modified fried image data.

In addition, the program that realizes the above method performs spatial frequency analysis processing on the fried food image data to obtain power value data, corrects the power value data, performs reverse spatial frequency analysis processing, and obtains the modified fried food image data. It is for creating.

As a method of this correction, only the power value data in a specific spatial frequency is set in the desired value range. For example, the power value data in the spatial frequency range in which "moist feeling" is strongly felt before processing is "crispy". The texture can be improved by modifying the power value data to strongly feel. Further, in order to suppress the discomfort of the entire fried food image, the power value data may be increased or decreased over the entire spatial frequency as described above.

That is, as described above, by using this method and this program, by creating a modified fried food image in which the acquired fried food image data is modified to a more desired feeling, the database is enriched and further evaluation verification is performed. It will be possible to contribute and improve the reliability of evaluation.

(Principle confirmation)
Here, we actually created a texture evaluation program for fried foods and confirmed the effect of the texture evaluation method for fried foods.

First, we prepared a total of 29 types of shrimp tempura images. These images were evenly distributed on the two axes of unevenness and roundness of clothing details. At the time of shooting, about 15 cm of shrimp tempura was taken from a distance of about 25 cm, trimmed to a size of 512 x 512 pixel (resolution 72 dpi) at 3 points each, and 87 sheets were excluded, excluding those with very similar appearance. Fried food image data was created.

Next, from these images, 5 images each with strong and weak "crispy", "crispy", and "moist" images were selected (including duplication). Then, using the 10 images selected for each texture, a sorting task was performed by the normalization ranking method for 13 university students as general consumers, and between the top 5 and bottom 5 images. A statistical advantage was recognized. Specifically, in the case of a crispy feeling, t (24) = 5.498, p <0.05, in the case of a crispy feeling, t (24) = 3.354, p <0.05, in the case of a moist feeling, T (24) = 3.836 and p <0.05, respectively, were observed.

Then, the spatial frequency analysis, specifically, the fast Fourier transform process was performed on these fried food image data, and the power value data at the spatial frequency was calculated using MATLAB.

FIG. 3 shows the results of spatial frequency analysis in a state of strong crispness and a state of weak crispness in the confirmation of this principle. When the t-test was performed between the upper and lower power values in this spatial frequency, it was 0 to 1, 3 to 6, 10 to 20, 24 to 26, and 29 to 31 in the strong and weak crispness states. , 33-35, 52-53, 61-62 cycle / image showed a dominant difference.

Further, as in the above, the spatial frequency analysis results of the state where the feeling of crunchiness is strong and the state where the feeling of crunchiness is weak, and the state where the feeling of moistness is strong and the state where the feeling of moistness is weak are shown. The results are shown in FIGS. 4 and 5, respectively. Further, FIG. 6 is a diagram showing a summary of FIGS. 3 to 5 above.

As a result, as a result of the spatial frequency analysis, the amplitude is large in the spatial frequency band of about 0 to 50 cpi in the image with a strong crispy feeling, and the amplitude is large in the spatial frequency band of about 0 to 60 cpi in the image with a strong crispy feeling, and the feeling of moistness is high. In a strong image, it was confirmed that the amplitude was large in the spatial frequency band of about 0 to 90 cpi and 240 cpi. This result is still speculative, but the low to medium frequency components of about 0 to 50 cpi affected the impression of density and continuity of the clothing surface, which strengthened the crispy and crunchy feeling, and 50 to 50. It is probable that the slightly high frequency band of about 60 cpi contributed to the enhancement of the contour, and the clothes gave a firm impression and strengthened the crunchy feeling. On the other hand, it is considered that the moist feeling was created by the small change in brightness of the low to medium frequency components of about 0 to 50 cpi.

Claims (2)

Spatial frequency analysis processing is performed on the fried food image data of tempura including the coordinate data and the brightness data corresponding to the coordinate data .
Power value data is obtained by the spatial frequency analysis based on the coordinate data and the luminance data .
A method for evaluating the appearance of tempura, which determines the appearance of texture evaluation value data by referring to the power value data and a power value database recorded in advance. On the computer
Spatial frequency analysis processing is performed on the tempura fried image data including the coordinate data and the brightness data corresponding to the coordinate data .
Power value data is obtained by spatial frequency analysis based on the coordinate data and the brightness data .
Tempura's appearance texture evaluation program that determines the appearance texture evaluation value data by referring to the power value data and the power value database recorded in advance.

Images