JP2014002136A - Method for classifying meat color - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method to create a meat color classification standard and a meat color classification method capable of classifying a detailed meat color without depending on human subjective decision and being automated.SOLUTION: The present invention relates to a method to create a meat color classification standard, the method including steps of (1) extracting digital images of surfaces of plural meat samples, (2) obtaining a CIELab value of each meat sample from RGB values of the extracted digital images, (3) classifying a meat color of each meat sample, and (4) obtaining the meat color classification standard from a relationship between any two factors of the CIELab value obtained from the plural meat samples and the classification. The present invention further relates to a classification determination method, the method including a step of (5) performing the steps (1) and (2) with the meat sample to be tested to obtain the CIELab value and classifying the meat sample to be tested according to the meat color classification standard obtained in the step (4).

Description

  The present invention relates to a method for creating a meat color grade determination criterion for meat and a meat color grade determination method. More specifically, the present invention relates to a method for creating a grade determination standard and a grade determination method for meat color, particularly beef meat color, using CIELab values obtained from digital images.

  Meat color is an important factor in the quality of beef as well as fat crossing. The determination of the meat color grade is performed by the rating members of the Japan Meat Rating Association based on the beef color standard (BCS) in the slaughterhouse. Grade determination by BCS is performed based on experience by a skilled rating member using a standard model, and the grade by BCS number is discontinuous in 1 to 7 stages. In this way, the BCS number is a grade given by the rating person's subjectivity, and even in meat of the same grade (same BCS number), the meat color may differ.

  As an alternative to the method of rating the meat color directly by the rating staff in the slaughterhouse, an image read by a digital camera or image scanner is displayed on a computer monitor to determine the grade (BCS number assignment). . By using a color calibrator, it is possible to display the flesh color on a monitor with good reproducibility, and based on this display, a rating member can assign a BCS number with high accuracy (Non-patent Document 1).

Kuchida et al., Japanese Journal of Animal Science, 73 (4): 521-528

  However, there is no change in the meat color grade determination method using the read image. Therefore, more detailed meat color grades cannot be determined, and the problem of differences in meat color even with meat of the same grade remains unsolved. Furthermore, as far as the present inventors know, there is no method for objectively determining the meat color grade. Furthermore, as far as the present inventors know, there is no method for objectively and automatically determining the meat color grade.

  SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to provide a new meat color grade determination method capable of finely determining a meat color grade regardless of subjective judgment. Furthermore, an object of the present invention is to provide a new meat color grade determination method capable of automating the meat color grade determination.

  As a result of various studies to achieve the above object, the present inventors calculated L * a * b * from the RGB values of the photographed meat cross section, and L * a * b * and the conventional meat color grade and It is possible to objectively provide a meat color grade by calculating L * a * b * of the meat sample to be inspected from this correlation, and further subdivide the grade The present invention has been completed by finding that detailed grading can be performed.

  According to the present invention, a method for objectively determining a meat color grade can be provided. As a result, it is possible to mechanize the determination of the meat color class that requires skill, and it is possible to automate the determination of the meat color class of the meat sample to be inspected.

An image of the extracted cross section (a) and loin core (b) is shown. The color chart (drawing is black and white) image | photographed with the rib camera of the stationary imaging device is shown. The relationship between the L * a * b * measured value obtained from the spectrocolorimeter and the L * a * b * estimated value calculated from the RGB value obtained from the image photographed by the stationary photographing apparatus is shown. A scatter plot of rib muscle L * and muscle a * is shown. A scatter plot of muscle L * and b * is shown. A scatter plot of muscle a * and b * is shown. Based on a scatter diagram of rib muscle L * and muscle a * (Fig. 4), it is a diagram that cuts out and pastes the individual images between each numerical value in L *, a * from the loin core. . Based on Fig. 7, the minimum value (integer value) of all data is defined as the starting point (54.0, 15.0), and the straight line (y = 0.5117x-12.639) passing through the average value (63.8, 20.1) from the starting point, the average for this line A vertical line i BCS4 passing through the values, and parallel lines iBCS0 to 7 are shown at intervals such that the distance is about 2.5. Samples with i BCS from 2.12 to 6.55 as shown by flesh color detail grading by i BCS. Samples with i BCS from 2.12 to 6.55 as shown by flesh color detail grading by i BCS.

The preparation method of the meat color grade determination criteria of the present invention includes the following steps (1) to (4).
(1) A digital image is extracted from the surfaces of a plurality of meat samples.
(2) Obtain the CIELab value of each meat sample from the RGB values of the extracted digital image.
(3) Determine the meat color grade for each meat sample.
(4) A meat color grade determination criterion is obtained from the relationship between any two factors of the CIELab values obtained for the plurality of meat samples and the grade.

The meat color grade determination method of the present invention includes the following steps (11), (12) and (13).
(11) A digital image is extracted from the surface of the meat sample to be inspected,
(12) Obtain the CIELab value of the meat sample to be inspected from the RGB values of the extracted digital image,
(13) determining a meat color grade of the meat sample to be inspected in light of the meat color grade determination criteria;
However, the meat color grade determination criteria are the criteria obtained in the following (a1) to (a4) (however, (a1), (a2) and (a3) are optional).
(A1) extracting digital images from the surface of a plurality of meat samples;
(A2) Obtain the CIELab value of each meat sample from the RGB values of the extracted digital image,
(A3) Determine the meat color grade for each meat sample,
(A4) The meat color grade determination standard is determined from the relationship between any two factors of the CIELab values obtained in (a2) and the grade determined in (a3) for the plurality of meat samples described in (a1). .

  The method for creating the meat color grade determination criteria and the meat color grade determination method of the present invention can target all meats that require the meat color grade determination, such as beef, pork, and chicken. Also, the flesh color can be a muscle color or a fat color. In particular, the color of beef muscle, which is highly demanded for determining the meat color grade, is preferable. However, it is not intended to limit it.

<Method for Creating Meat Color Grade Determination Criteria of the Present Invention>
Process (1)
It is a step of extracting digital images from the surfaces of a plurality of meat samples. Steps (1) to (3) are steps for obtaining a meat color grade determination criterion in step (4). From the viewpoint of providing a highly accurate meat color grade determination method, it is preferable that the number of the plurality of meat samples is large. For example, it can be between 10 and 200, but is merely an example, and may be less than 10 or more than 200.
The surface of the meat sample is not particularly limited. In the case of beef muscle, for example, an incision can be made between the sixth and seventh ribs of the carcass left half circle, which is used in the meat color rating of the Japan Meat Rating Association. Alternatively, the surface of the beef muscle sample can be a cut surface of a beef loin core. However, it is not intended to be limited to the surface of this part, but can be appropriately determined according to the purpose of determining the meat color grade.

  Digital images can be taken using a color image photographing device such as a digital camera or an image scanner. Since a predetermined portion is extracted from the digital image and the RGB value is obtained, any digital image may be used as long as it is photographed under conditions for obtaining the RGB value. In the case of the muscle color evaluation, only the muscle portion is extracted, and in the case of the fat color, only the fat portion is extracted, and an average RGB value is obtained for each.

Process (2)
In this step, the CIELab value of each meat sample is obtained from the RGB values of the extracted digital image.
From the extracted digital image, the Red (R) average, Green (G) average, and Blue (B) average values of the muscle portion of the meat sample are obtained. This method is known and can be performed, for example, according to the method described in the following literature.
(1) Kenichiro Takahashi, Takeshi Hori, Michitaka Nami, Yasunori Honma, Hitoshi Kodaka, Satoshi Kuchita, BCS number estimation using images from high-definition carcass cross-sectional imager, Journal of the Japanese Society of Animal Science, 77 (2 ): 237-244, 2006
(2) Satoshi Kuchita, Mio Hasegawa, Michiyoshi Suzuki, Shunzo Miyoshi, Judgment of BCS number using digital images taken by carcass cross-section imaging device, Journal of the Japan Society of Animal Science, 72, 9, J321-J328, 2001

Further, the R average, G average, and B average obtained above are converted into the XYZ color system. The conversion from RGB to XYZ can be performed mathematically using the following conversion formula.

  Furthermore, the value of the XYZ color system is further converted to the L * a * b * color system. Conversion from the XYZ color system to the L * a * b * color system can be performed mathematically using the following conversion formula.

Step (3)
It is a step of determining a meat color grade for each meat sample.
In the case of beef, the meat color grade may be, for example, the beef color standard (BCS) number used in the meat color rating of the Japan Meat Rating Association. Other grades can be used as appropriate. The beef color standard (BCS) number consists of 7 levels, 1-7, and is represented in the 7-level BCS standard model. BCS number assignment (meat color rating) is usually based on the BCS standard model. The determination of the meat color grade based on the BCS standard model is based on visual comparison between the BCS standard model and the meat to be graded, and the beef color standard (BCS) closest to the meat to be determined from the BCS standard model. This is done by selecting a number.

  In addition to the beef color standard (BCS) number, for example, grades based on the US meat color standard or the Australian meat color standard can be used. US meat color standards include, for example, Beef Quality Grades (beef quality standards) (http://meat.tamu.edu/ beefgrading.html). The grade about Lean Color is described in the section of Lean Maturity. Australian meat color standards include, for example, the Beef Chiller Assessment (http://www.australian-meat.com) provided by Meat and Livestock Australia, Ltd. (MLA). / Foodservice / Proteins / Beef / Beef # Chiller # Assessment /). The grade about Meat Color (meat color) is shown in this.

  The determination of the meat color grade based on the meat color standard of the United States or the Australian meat color standard is also the closest to the meat to be determined from the standard by visually comparing the standard color etc. with the meat to be determined This is done by selecting a reference number.

The extraction of the digital image in the step (1) and the determination of the meat color grade in the step (3) are respectively performed on the same surface of the same meat sample.
Whether the step (1) is performed first or the step (3) is performed first for the same meat sample is arbitrary, and the steps after the step (1) and the step (3) are also optional. is there.

Step (4)
This is a step for obtaining a meat color grade determination standard from the relationship between any two factors of the CIELab values obtained in step (2) and the grade determined in step (3). The CIELab value has three factors L *, a *, and b *, and two of them are used. Which two combinations are used is arbitrary. The meat color grade can also be a BCS number for beef. As shown in the examples, in determining the meat color grade of the Holstein type, using L * and a * makes it easy to determine the grade with relatively high accuracy when the BCS number is used as the meat color grade. L * and a * can also be used to determine meat color grades other than Holstein.

  The case of using L * and a * will be described as an example of the meat color grade determination criteria in step (4). From the L * and a * of each meat sample and the BCS number, the BCS number of each meat sample is positioned in the L * and a * coordinates. The positioning of the BCS number in the L * and a * coordinates is based on the coordinates of the lowest value of L * and the lowest value of a * in the plurality of meat samples as one origin, and the origin and the plurality of meat samples. Using the coordinates of the average value of L * and the average value of a * as the other point, a straight line connecting the two is obtained. Next, a plurality of perpendicular lines are obtained from the straight line at equal intervals and each representing a predetermined grade. In this way, it is possible to obtain the meat color grading determination standard in which the region corresponding to each BCS number (grading) is indicated in the L * and a * coordinates.

  Therefore, if L * and a * of the meat sample to be inspected are obtained, the meat color grade determination criteria is used to determine whether the meat sample is located in the area corresponding to the BCS number (grade). A number (grade) can also be determined. Furthermore, since the distance from the perpendicular can be quantified in the coordinates, the BCS number (grade) can be displayed more finely. For example, in the case of meat having L * and a * in the middle of BCS numbers 2 and 3, BCS number 2.5 can be displayed. Based on the meat color grade determination criteria obtained in this way, the grade that is displayed by subdividing the BCS number is referred to as an iBCS number in the present specification. The iBCS number is a beef color grade calculated by image analysis, for example, measured using the relationship with L * and a *. Even if any two factors of CIELab values other than L * and a * are used, it is possible to similarly create a meat color grade determination standard, and it is also possible to determine an iBCS number. The determination of the grade based on the iBCS number will be described in more detail in the embodiment.

<Meat color grade determination method of the present invention>
In the meat color grade determination method of the present invention, the CIELab value is obtained by performing the steps (11) and (12) on the meat sample to be inspected, and the meat color of the meat sample to be inspected in light of the meat color grade determination criteria is determined. This is a method for obtaining a grade.
First, the steps (11) and (12) are performed on the meat sample to be inspected to obtain the CIELab value. Steps (11) and (12) are the steps (1) and (2) in the method for preparing the meat color grade determination criteria of the present invention, except that the meat sample is a meat sample to be inspected. Is the same. The obtained CIELab value is used to determine the meat color grade in light of the meat color grade determination criteria. The meat color grade determination standard is a standard determined by the method including the steps (a1) to (a4), which is the same as the method for creating the meat color grade determination standard of the present invention. This meat color grading standard provides a certain calibration curve (in fact, a calibration surface, not a calibration curve) as described above, and in light of this calibration curve (surface), the meat color of the meat sample to be inspected The grade of can be determined. When using L * and a *, the meat color grade of the meat sample to be inspected is obtained from the L * and a * of the meat sample to be inspected and the distance from the perpendicular determined by the meat color grade determination criteria. . Specific operations will be described in detail in Examples.

  Information on the meat color grading criteria (for example, the relationship between L * and a * and the grade) and the formula for calculating the grade from L * and a * are entered into the computer, and the L * of the meat sample to be inspected And a * can automatically determine the grade, and the grade can be an iBCS number obtained by further subdividing the BCS number.

  Hereinafter, the present invention will be described in detail with reference to examples. However, the present invention is not intended to be limited by the examples.

  Using a stationary imaging device, the 6-7 intercostal cross-section and peach of the carcass right half circle were photographed to obtain a high-definition cross-sectional image. The 6-7 intercostal cross section, loin core and peach were extracted from the captured images (FIG. 1), and image analysis traits were calculated using image analysis software (BeefAnalyzer II, Hayasaka Riko, Sapporo City). The image analysis traits to be investigated in this example are a muscle R average, a muscle G average, and a muscle B average calculated by binarizing the loin core and calculating only from the muscle portion.

  The lobe core was extracted from the image photographed by the stationary photographing apparatus using BeefAnalyzer II. The muscle R average, muscle G average, and muscle B average of the image analysis traits were replaced with the XYZ color system, and further converted to the L * a * b * color system using the values. The conversion formula is as described above.

  The above L * a * b * is a simple value obtained from the RGB of the image, and there may be a deviation from the true L * a * b *. Therefore, in order to investigate the error from the actual measurement value, the color chart (ColorChecker Passport, X-rite) L * a * b * actual measurement value, stationary imaging device using a spectrocolorimeter CM-1000 (KONICA MINOLTA) The correlation of L * a * b * obtained from the RGB values of the color chart image taken in Fig. 2 (Fig. 2: Rib camera) was examined.

  We also investigated the relationship between the two-dimensional distribution of L * vs a *, L * vs b * and a * vs b * and the BCS number obtained from the rating to determine the detailed grade of meat color. Next, we examined the flesh color grades that were closer to each other by combining the L * a * b * values, and created a continuous variable i BCS.

Result Using the color chart, L * a * b * measured with the spectrophotometer CM-1000 (KONICA MINOLTA) and L obtained from RGB of the images (rib camera, peach camera) taken with the stationary photographing device A scatter plot of * a * b * is shown in FIG. A high correlation was sought with a coefficient of determination of 0.88 or more among all traits. In L * and b *, all points were plotted almost linearly, but in a *, there were points that deviated significantly from the approximate curve. This greatly deviating number was from yellow and orange. This is considered to be a problem of the characteristics of the digital camera. By reducing this error, the accuracy can be improved.

  4 to 6 show scatter plots of rib muscle L * and muscle a *, muscle L * and muscle b *, muscle a * and muscle b *, respectively. The correlation coefficients were 0.35, 0.53, and 0.37, respectively. If the correlation was strong, only one of the traits could be used, but no strong association was found. Also, the variation in BCS numbers was large in the scatter plots of L * and a *. Therefore, in the following, it was decided to determine the meat color grade based on the combination of the muscle L * and the muscle a * that had the lowest correlation between the muscle L * and the muscle a * and the BCS variation was large.

  Based on a scatter diagram of rib muscle L * and muscle a * (FIG. 4), an image of an individual between L * and a * at each step is cut out from the loin core and shown in FIG. As a result of visual grading, the flesh color tended to fade as L * and a * increased. Based on the created diagram (Fig. 7), the X axis is L *, the Y axis is a *, the minimum value (integer value) of all data is set as the start point (54.0, 15.0), and the average value (63.8, 20.1) was obtained (y = 0.5117x-12.639). A perpendicular line passing through the average value was drawn on this straight line. Since the average BCS number of the material cattle used this time was 3.9, the vertical line passing through the average value was defined as i BCS4. With reference to this straight line, parallel lines are drawn at intervals such that the distance is about 2.5 to define iBCS0 to iBCS0 to 7 and are shown in FIG. This establishes the meat color grading criteria.

  Based on the above meat color grading criteria, i BCS for grading meat color in more detail is from the (L *, a *) point of each individual to the straight line of i BCS0 (y = -1.9543x + 164.418) The distance is calculated and the value is divided by 2.5. FIG. 9 shows samples from i BCS2.12 to 6.55, which were shown by meat color detailed grading by i BCS, which is actually a detailed color grading method.

  The present invention is useful in the meat and livestock fields.

Claims (13)

(1) Extract digital images from the surface of multiple meat samples,
(2) Obtain the CIELab value of each meat sample from the RGB values of the extracted digital image,
(3) Determine the meat color grade for each meat sample,
(4) Obtain a meat color grade determination standard from the relationship between any two of the CIELab values obtained for the plurality of meat samples and the grade.
However, (1) and (2) and (3) are optional before and after the creation method of the meat color grade determination standard. The method according to claim 1, wherein L * and a * are used as the two factors of the CIELab value in (4). The positioning of the grade in the L * and a * coordinates in (4) is based on the lowest L * value and the lowest a * value of the plurality of meat samples as the origin of the coordinates, and the origin and the plurality of meat samples. The method according to claim 2, wherein a straight line connecting the average value of L * and the coordinates of the average value of a * is obtained, and a plurality of perpendicular lines are obtained from the straight line at regular intervals so as to represent predetermined grades. The method described. The method according to any one of claims 1 to 3, wherein the meat color grade in (3) is a beef color reference (BCS) number when the meat to be inspected is beef. The meat color rating determination standard in (4) is obtained by positioning the BCS number of each meat sample in the L * and a * coordinates from the L * and a * and BCS number of each meat sample. Item 5. The method according to Item 4. The meat color grade in (3) is the grade of red color in the beef quality standard provided by the Faculty of Animal Science, Texas A & M University, or the Australian Meat Producers Association, when the meat to be inspected is beef. The method according to any one of claims 1 to 3, which is a grade for meat color in the provided frozen beef evaluation. (11) A digital image is extracted from the surface of the meat sample to be inspected,
(12) Obtain the CIELab value of the meat sample to be inspected from the RGB values of the extracted digital image,
(13) determining a meat color grade of the meat sample to be inspected in light of the meat color grade determination criteria;
However, the meat color grade determination criteria are the criteria obtained in the following (a1) to (a4) (however, (a1) and (a2) and (a3) are optional).
Meat color grade determination method.
(A1) extracting digital images from the surface of a plurality of meat samples;
(A2) Obtain the CIELab value of each meat sample from the RGB values of the extracted digital image,
(A3) Determine the meat color grade for each meat sample,
(A4) The meat color grade determination standard is determined from the relationship between any two factors of the CIELab values obtained in (a2) and the grade determined in (a3) for the plurality of meat samples described in (a1). . The method according to claim 7, wherein L * and a * are used as two factors of the CIELab value in (a4). The positioning in the L * and a * coordinates of the grade in (a4) is performed with the lowest value of L * and the lowest value of a * in the plurality of meat samples as the origin of the coordinates, and the origin and the plurality of meat samples. A straight line connecting the average value of L * and the coordinates of the average value of a * is obtained, and a plurality of perpendicular lines are obtained from the straight line at regular intervals so as to represent predetermined grades. The method described. The method according to claim 9, wherein the grade of the meat color of the meat sample to be inspected in (5) is obtained from a distance from L * and a * of the meat sample to be inspected and any one of the perpendiculars. The method according to any one of claims 7 to 10, wherein the meat color grade in (a3) is a beef color reference (BCS) number when the meat to be inspected is beef. The meat color grade determination criteria in (a4) is obtained by positioning the BCS number of each meat sample in the L * and a * coordinates from the L * and a * and BCS number of each meat sample. Item 12. The method according to Item 11. The meat color grade in (a3) is the grade of red color in the beef quality standard provided by the Faculty of Animal Science, Texas A & M University, or the Australian Meat and Livestock Producers Association. 11. The method according to any one of claims 7 to 10, which is a grade for meat color in the provided frozen beef evaluation.