MXPA99001702A - Method and apparatus for using image analysis to determine meat and carcass characteristics - Google Patents

Abstract

In a process and apparatus for determining grading parameters of a carcass, the outline of an image of the carcass is traced and reference points representing anatomical features of the carcass are identified. Second reference points being located at pre-determined positions relative to the first reference points are then identified. The carcass image is divided into a plurality of sections, the boundaries of each section being determined as a function of the position of the first and second reference points, and the area of each section is determined. A grading parameter predictive equation is determined wherein the grading parameter is included as a dependent variable, and at least one area of a carcass image section is included as an independent variable. Solving the predictive equation provides a value for the grading parameter of the carcass. Other measurements which can be obtained from the carcass image and used as independent variables in predictive equations include distances from dorsal and ventral regions of the carcass image outline to a carcass mid-line, carcass widths, angular measurements between reference points, and measurements of curvature of the carcass image outline. Improved rib eye tracing techniques permit accurate measurement of rib eye parameters. The measured rib eye parameters may be used to determine a quality grade for the carcass or as independent variables in a carcass grading parameter predictive equation, alone, or in conjunction with measurements taken from the carcass image.

Description

METHOD AND APPARATUS FOR USING THE IMAGE ANALYSIS TO DETERMINE THE CHARACTERISTICS OF THE MEAT AND RESES BACKGROUND OF THE INVENTION FIELD OF THE INVENTION The invention relates to the processing of images and the statistical analysis of digitized images of reees of meat-producing animals to determine the degree and performance parameters of the beef.

DESCRIPTION OF THE RELATED TECHNIQUE The evaluation of meat-producing animals, both in live animals and in cattle, was typically carried out by human qualifiers, introducing considerable subjectivity to the evaluation process. There are two main aspects with respect to the evaluation of the meat, the degree of quality and the degree of performance. The degree of quality of the young animals is determined by the amount of intramuscular (marble) fat of the REF. 29512 meat. The degree of yield describes the proportion of lean tissue in the beef. In the res, the evaluation is usually done by observing and measuring a cross section of the sysimus dorsi ongi (described in butchery as "the rib eye muscle" and in pigs as "loin muscle"). The degree of quality or marbling is typically determined by comparing the appearance of the rib eye with reference to beef eye photographs of known quality grades. The qualifier can evaluate the grade of quality by comparing the amount of marbling of the examined rib eye with the amount of marbling observed in the reference photographs. The proportion of leaved lean beef (degree of yield) is typically estimated by considering the area of the rib eye and the thickness of the subcutaneous fat at various points around the rib eye. The calculations of the degree of performance can also involve measurements of the fat of the viscera and the weight of the hot beef. As will be explained in more detail below, there are several definitions of a possible "degree of performance", since these may depend on the specific processing rules of the cattle. A particularly useful measure of the degree of yield is the "yield for sale" of the beef, which reflects the proportion of the live weight of the animal constituted by the sum of the weight of salable cuts plus the weight of the cuts. Typically, the yield for sale is determined by cutting the beef into standard cuts of meat. A number of automated meat processing arrangements have made use of the different reflection properties of muscle tissue light in contrast to adipose tissue. U.S. Patent No. 5,324,228 (Vegely, issued June 28, 1994) discloses a method and apparatus for illuminating a fish fillet with a light strip when taken by a pair of video cameras. The light signals emitted by the cameras are converted by a computer to electrical digital signals that represent the luminosity of the illumination. The computer compares the digital signals with a preselected threshold of gray scale levels to locate the peripheral fat areas. Then, the computer controls the operation of a cutting mechanism to eliminate fat areas. In U.S. Patent No. 3,800,364 (Lapeyre, issued April 2, 1974) and U.S. Patent No. 4,738,004 (Lapeyre, issued April 19, 1988) describe similar provisions to distinguish meat from light colored edible loin of non-edible waste meat from dark-colored tuna slices. U.S. Patent No. 4,154,625 (Kail, issued October 27, 1964) describes a method for determining the marbling of the rib eye of a res by measuring the average reflectivity of a rib eye with respect to the reflectivity of a rib eye. colored sample plate of fat, using a photometer. United States Patent No. 4,413,279 (Gorl, issued November 1, 1983) describes an improved method for calculating a brightness threshold to distinguish fat from lean tissue in order to overcome problems in identifying fabrics of intermediate luminosity, such as blood-stained fat, for use in meat evaluation arrangements in which the relative luminosity of various tissues is recorded with a video camera.

United States Patent No. ,352,153 (Burch et al., Issued October 4, 1994) describes an apparatus for illuminating and obtaining video images of fish cuts during processing. U.S. Patent No. 4,226,540 (Barten et al., Issued October 7, 1980) discloses a method for determining meat quality characteristics, in which the ratio of adipose to lean tissue is determined by scanning a product meat with a beam of light moving and discriminating fat from lean tissues based on the different values of clarity of fat and tissues. A number of evaluation arrangements have been described by video images in which a series of images of live animals is taken. U.S. Patent No. 5,483,441 (Scofield et al., Issued January 9, 1996) discloses an arrangement for obtaining and analyzing video images in which a series of video images is obtained and evaluated as the alive animal crosses successive visual fields. U.S. Patent No. 4,745,472 (Hayes et al., Issued May 17, 1988) discloses an arrangement for obtaining and analyzing video images in which markers are placed at various anatomical reference points on the body of the patient. a living animal. The animal is then placed in an inclined passageway having upper and lateral walls comprising measuring rods. Recordings of video tapes of the animals in the catwalk are made, and the information is analyzed in video with a computer to determine the distances between the markers adhered by hand on the body of the animal. Other provisions have information by video images combined with other acquired information, for example, inserting a probe in the res, to produce the evaluative data. U.S. Patent No. 4,939,574 (Peterser et al., Issued July 3, 1990) discloses a camera with an illuminated screen in which the silhouette of an animal with an electronic camera is recorded and the outline of the image is determined. res with a data processing device. The contour information of the beef is used in conjunction with a previous assessment of the color of the beef and the information on the thickness of the meat and fat determined by inserting a probe into the beef, to determine a classification of the latter. United States Patent No.

No. 4,439,037 (Northeved et al., Issued March 27, 1984) discloses an optical probe for insertion into a beef to evaluate the ratio of meat to fat of beef. Ultrasound images of live animals have been analyzed in order to estimate the marbling or thickness of subcutaneous fat of the animal. U.S. Patent No. 4,785,817 (Stouffer, issued November 22, 1988) discloses an apparatus and method for employing ultrasound to determine the thickness of the fat in different parts of a beef from which determinations can be made. evaluative Similarly, U.S. Patent No. 5,339,815 (Liu et al., Issued August 23, 1994) concerning the ultrasonic imaging of cattle meat product, describes the association of autocorrelation property of noise. punctual ultrasonic with the qualification of the marbling of the meat. International Patent Application WO 93/21597 (Benn et al., International filing date - April 13, 1993) describes a method for tracking the contour of a digital image of a rib eye muscle of a res in which they are defined. links between pairs of concavities of the rib eye sketch to be able to cut sections of external images with respect to the rib eye. International Patent Application WO 92/00523 (Ne man, Date of International Presentation - June 24, 1991) describes a method to evaluate cattle after killing, which includes the steps of detecting the presence of a beef in a field of view of a camera, checking that the beef is correctly oriented with respect to to the camera, obtaining images of the res from a plurality of points of view, determining a plurality of dimensions of the res with respect to the images and comparing the dimensions with the stored values to determine a rating for the res. However, there is no description of how to determine the dimensions of the beef or determine how they can be related to the beef rating. International Patent Application WO 91/14180 (Benn, Date of International Presentation - March 14, 1991) describes a method for evaluating cattle by means of objective images comprising the steps of recording an image of a background, recording a second image. of a res placed against the bottom, analyze the first and second images to differentiate the res of the background subtracting the first or second image of the other for each color component in order to produce a series of differential images of the components that recombine to produce an image of absolute difference. The request expresses that anatomical points of the res can be identified comparing the surface of the profile of the res with a series of reference profiles, and matching the anatomical points of the images that have the highest proportion of similar area. It is stated that the quantitative measurements of the dimensions can be taken from the anatomical points to estimate the composition, but it is not described how to make those quantitative measurements, which ones can be useful, or how to make a prediction based on the measurements. In conclusion, the provisions described above do not allow continuous calculations of the grade or yield of the cattle during the killing process. Techniques are needed to reliably take accurate and reproducible measurements of the dimensions of the cattle without manual identification of the animal's anatomical characteristics, and to develop yield predictions based on these beef measurements. This requires the identification of precise and reproducible specific measurements of the beef, which are closely linked to the parameters of grade or yield of interest. Also necessary are refined rib eye tracking techniques that can also be employed in grade and performance determinations.

SYNTHESIS OF THE INVENTION The inventors have developed a fast and accurate process and apparatus for online qualification of cattle processed in a slaughterhouse. Visible video spectral images of skinned half-carcasses suspended from a conditioning rail are obtained, digitized and stored in a computer memory for image processing and analysis. The image of the contour of the res is traced and the anatomical features of the res, such as the tail, are identified as entrances or protrusions in the image of the contour of the res by the methods described herein. Although preferably a side view of a half-beef is used, other views of the beef, or images of intact cattle, may be used. Once one or more anatomical characteristics have been identified in the contour of the res, other reference points can be located in the image of the res at predetermined positions with respect to the original anatomical features identified in the first place. For example, an additional reference point can be placed at a given percentage of the length along a line joining two anatomical features. In this way, by locating a small number of anatomical features of the beef, it is possible to quickly, accurately and reproducibly identify any number of additional reference points on or within the contour of the beef image. These additional reference points may reflect the anatomical characteristics of the res that are not easily identified as indicator marks in the outline of the beef image. On the other hand, the additional reference points can be arbitrarily assigned points that are useful to provide a multitude of defined, reproducible points, from which single- and two-dimensional measurements can be made. Using the various identified reference points, a plurality of measurements are made by images of the res. These may include, among other things, the linear distances between reference points, the areas limited by reference points, the angular measurements between selected series of three reference points, and the calculations of the curvature along the contour of the image. of the res. Using known statistical techniques, such as step-by-step regression, predictive equations have been developed in which a qualification parameter of the res selected as a dependent variable is included, and various measurements by res images are included as independent variables. In the exemplified case, related to the vaccine cattle, it has been found that the shortest distance from each of a plurality of reference points along the contour of the beef image to a determined midline parallel to the longitudinal axis of the animal. outline of the image of the beef, which divides the image into portions that could be called dorsal and ventral, the width of the outline of the beef image, and the areas of parts of the beef image that have approximate limits to the of the main normal cuts of the res are especially useful independent variants. The main cuts are the cross sections in which a beef is initially cut during the meating process and on which the remaining cuts depend during the manufacturing process. The inventors have developed a method to quickly approximate the main standard cuts in the beef image, which involves dividing the beef image into sections delimited by lines that link predetermined reference points on and within the outline of the image of the animal. beef. The area of certain major cuts, and the ratio of the surface area of these major cuts to the total area of the beef image, have proved to be especially useful as independent variables in the predictive equations used to predict such things as yield for the sale of the res.

Stepwise regression techniques are used to determine the degree of linear association between each of the measurements obtained from the beef image and the selected qualification parameter of the beef, and to determine the best model to predict the value of the grading parameter. selected from the res in which a plurality of measurements by res images form the independent variables. Once a predictive equation has been developed, the disposition to take measurements by images of other cattle can be used, and the predictive equation can be solved by means of these measurements, to provide a result of the value of the selected qualification parameter of the beef. Although predictive equations can be developed to estimate the value of a wide variety of beef rating parameters, a particularly useful application of the invention is the prediction of the yield for the sale of a beef. The definition of "yield for sale" varies between different markets for meat products that are slaughtered. In general, it reflects the sum of the weight of the initial cuts of the beef at a defined fat cover level plus the weight of piles of cuttings at various percentages of thickness. In the examples set forth herein, "yield for sale" was defined as the total weight of all cuts with 1/4"(6.25 mm) of adipose cover, in which all cuts are derived from the eight major cuts of the cut. rump, loin, skirt, rib, roast beef, shoulder, shoulder, and breast, plus piles of cutouts with 50%, 75%, and 85% magura.Therefore, briefly expressed, in a preferred embodiment, the invention presents a process for determining a qualification parameter of a res, comprising the steps of: (a) obtaining an image of a view of the res, image that is composed of a series of pixels that provide data representative of the color information in the corresponding part of the image; (b) tracing the profile of the image to produce a contour image of the res; (c) locating a plurality of first reference points on the contour of the image of the res, first reference points representing traits anatomical of the res, anatomical features that are identified as prominences or depressions in the image contour of the res; (d) locating at least one second reference point on or within the contour of the image of the res, these second reference points being located at predetermined positions with respect to the first reference points; (e) dividing the image of the beef into a plurality of sections, the boundaries of each section being determined as a function of the position of the first and second reference points, and determining those of the area of each section; (f) present a predictive equation of the qualification parameters in which the qualification parameter is included as a dependent variable, and at least one area of a given section in step (e) is included as an independent variable; and (g) solving the predictive equation of the rating parameter to produce a value for the res rating parameter. The invention is extended to the analysis of the rib eye of the res. During the cutting process, the beef is cut in the transverse direction between the ribs, usually between the 12th and 13th ribs. The cut does not completely cross the beef, so it remains intact, in one piece, hanging on the conditioning rail. The weight of the beef opens the cut, allowing the visualization of the cross section of the muscle l ongi ssimus dorsi, which, in cattle producing meat, is called "rib eye" (ribeye), and in pigs it is called "loin eye". In the present and in the claims, the term "rib eye" is used to include the ongi ssimus dorsi ongi muscle of cattle and swine., as typically seen in cross-section during the res rating. The rib eye represents the most valuable cut in red meat animals such as pigs, lambs and cattle. Qualification techniques around the world have emerged from the measurements obtained from the rib eye. In accordance with what has been described above in reference to the entire res, a visible video image of the rib eye is obtained, digitized and stored in the memory of a computer for the processing and analysis of images. The pixels representing the muscle tissue are distinguished from the pixels that represent the fat based on a threshold of color characteristics of the pixels, such as level of cxarity. The pixels that represent the cartilage can be identified by their low level of color saturation. The contour of the rib eye is tracked, and the value of the variables such as the percentage of intramuscular fat, the surface of the rib eye, and the thickness of the subcutaneous fat at various points of the rib eye contour are determined. These variables can be included as independent variables in the predictive equations to predict the beef rating parameters based on the imaging measurements of the cattle described above. On the other hand, information about the rib eye can be used independently of the measurements taken from the images of the res to develop predictive equations to predict the res rating parameters based on the rib eye image measurements only. . Accurate scoring predictions based on rib eye imaging require accurate tracing of the rib eye contour. The muscular tissue that rests on the ongi ssim us dorsi (rib eye) but which is not part of the sysimus dorsi ongi must be differentiated to allow accurate measurements of the rib eye of the image being tracked. The present invention is extended to novel rib eye tracking techniques in which the external sections of images representing the muscle tissue in contact with the rib eye but not part of the eye of the rib eye are identified and cleaved with precision. rib, to produce superior rib eye tracking results. In accordance with the foregoing, the information obtained from the contour of the tracked rib eye can be used independently of the information derived from the image of the animal to predict a rating parameter of a beef such as the yield for sale or the grade. of quality (marbling). In a preferred embodiment, in relation to the prediction of marbling, the invention includes the obtaining of images of reference photographs of standard quality grades of rib eyes that are commonly used as a guide in the refrigerators by human qualifiers to determine the degree of quality. The photographs illustrate rib eyes with a degree of marbling at the maximum level for a given grade. Traditionally, the human qualifier compares the rib eye under study with reference photographs and assigns a grade based on the maximum marbling levels between which the rib eye under examination appears to fall. In one embodiment of the present invention, the actual percentage of intramuscular fat in the rib eye samples illustrated in the reference photographs is determined by scan analysis of rib eye images. After the rib eye tracking analysis, yield grades can be assigned to the res as a function of the percentage of marbling calculated for the rib eye image.

BRIEF DESCRIPTION OF THE DRAWINGS In the drawings illustrating embodiments of the present invention: Figure 1 is a side elevational view of the image arrangement of a res versus the background screen. There is a res suspended from a top conveyor between the bottom screen and the res image arrangement; Figure 2 is a side elevational view of the rib eye chamber connected to the CPU; Figure 3 is a bottom plane view of the rib eye chamber, showing the positioning jig; Figure 4 illustrates an image of a white cross-linked panel used to calibrate the camera; Figures 5-7 illustrate different image search masks useful in the present invention; Figure 8 illustrates the division of the beef image into a plurality of separate regions for analysis; Figure 9 illustrates another search mask useful in the present invention; Figures 10-14 illustrate successive steps of image analysis of the res according to the present invention; Figures 15-18 illustrate the identification of the anatomical landmarks, the estimation of the main cuts, the definition of the linear measurements and the definition of the angular measurements, respectively; Figure 19 illustrates the general appearance of the rib eye in an unprocessed digital image; Figures 20-30 illustrate successive steps of image analysis of the res according to the present invention; Figure 31 illustrates other details of the rib eye image; and Figures 32-35 illustrate steps in the performance of qualifying measurements from the contour of the tracked eye.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT The invention will be better understood with reference to the attached figures.

I. General Comments on Disposition A. Apparatus First images are obtained of intact beef half-carcasses hanging from a conditioning rail. You can take pictures of the intact beef within a few hours of slaughter, when the beef is still at a temperature close to body temperature, or once the beef has been refrigerated before continuing processing. The moment of the shape of the image of the intact res is not essence. According to that illustrated in Figure 1, a background screen 10 comprises four panels folded bottom 12 mounted on a frame 14 and supported on wheels stainless steel, it may be positioned about two feet (60 cm) behind the carcass 16 transported by a top conveyor and ten feet (3 m) in front of a CCD camera. One of the combs 12 can be reversible, and has a trellis design on one side and a black surface on the other. The grid serves to calibrate a camera. Five squares are presented in the center column of the grid, each one in black, white, red, green and blue, for color calibration. The bottom honeycombs 12 have a non-reflective blue background (or other color that has a tone different from the tone of any significant portion of the animal, with blue or green preferred) behind the cattle 16 to facilitate tracing res contour 16. One or more sliding bars 18 fitted to the frame 14 and oriented parallel to the direction of travel of the carcass 16, maintain the carcass 16 at a desired distance from the combs 12, in a plane perpendicular to the camera. When a res 16 moves towards the center of the background screen 10, the arrangement activates a camera to capture its image. The arrangement for obtaining images of cattle 20 includes a CCD camera, two sets of illumination by reflectors 24 (only one is seen in Figure 1) and CPU 26 mounted on a mobile base 28. The arrangement for obtaining images of warm cattle 20 it is located at a stipulated distance in front of the res 16 which travels along the upper conveyor and the bottom screen 10 is positioned behind the res 16 in line with the arrangement for obtaining images of hot cattle 20. The reflector assemblies 24 they are directed towards the bottom screen 10. Each set of reflectors 24 includes two halogen lamps of 110 V and 250 W, mounted in water-tight stainless steel boxes. Preferably, the lamps 30 are placed approximately five feet (1.5 m) in front of the curtain 10. Although the reflector lamp assemblies 24 are mounted on a movable base 28, these may be mounted on the ceiling, or they may be lamps standing. The entire arrangement for obtaining images of hot cattle 20 is preferably foldable for easy transportation and storage. The CCD camera 22 can be a Panasonic 3-CCD industrial color camera (model GP - US502) mounted in a stainless steel case and connected to a 12 V DC power supply. The RGB video signal generated by the camera 22 is fed to a Matrox image digitizing panel (Matrox Electronic Systems Ltd., Dorval, Quebec, Canada) (not illustrated) at PCU 26. The UCP is connected in a stainless steel case (which is not illustrated) to comply with the sanitary regulations of slaughterhouses. The UCP 26 can be a 150 MHz Pentium computer that operates programs under the Microsoft Windows 95 operating systems. The digitizing panel captures an image of a res 16 when it moves to the position in front of the background screen 10. The contour is traced of the res 16, certain anatomical points are determined, one takes a series of linear, two-dimensional and angular measurements, and the results and the image are saved. Generally, images of the rib eye area of the res 16 are taken once the res 16 has been refrigerated for 24 hours. While images of the rib eye of warm cattle could be taken, it would be more difficult to distinguish intramuscular fat from lean muscle tissue. As illustrated in Figures 2 and 3, the disposition for taking pictures of the rib eye 32 includes a manual rib eye chamber 34 and a CPU 36. Images of the rib eye are taken from a partial cross section of the rib eye. res 16 between the 12th and 13th ribs. This is the rib eye muscle site normally qualified by government inspectors. Therefore, the chamber 34 must be portable so that it can be placed inside a cut made in a V shape between the 12th and 13th ribs of the beef. The manual camera 34 can be a Panasonic 3-CCD camera, similar to the CCD camera 24, mounted inside a stainless steel case 38. There are two EXN low voltage exposure lamps of 12 V, 50 W 40 mounted on each side of the chamber 34. A switch key 42 on the handle of the chamber 44 is connected to an input / output panel of the CPU 36, which in turn informs the CPU 36 to record an image. The manual camera 34 is mounted on a guide 46 having a positioning tab 48 which places the camera 34 correctly in relation to the muscular area of the rib eye of the res 16. On the other hand, a small camera support can be used with a small camera and lens unit connected to a remote control unit (not illustrated). The camera support could be mounted directly on a small guide of light weight similar to the guide 46 for easy handling. The CPU 36 can be a 150 MHz Pentium computer mounted in a stainless steel case 38. The CPU 16 processes programs under the Microsoft Windows 95 operating systems. The software records an image of a rib eye. Trace the contour of the rib eye muscle. Calculate the surface, as well as the length and width of the rib eye, measure the thickness of the subcutaneous fat, the color of the muscle and the percentage of intramuscular fat. Then the software saves the results and the image. The following additional hardware and software can be used with the hot-dog imagery arrangement 20 and the arrangement for obtaining rib eye images 32 described above: A Matrox MGA Millenium video display adapter (Matrox Electronic Systems, Ltd.) for live video display (30 fps in 32-bit color) on screen; A Matrox Meteor image digitizing panel (Matrox Electronic Systems, Ltd.) to capture images; A PC-TIO-10 Input / Output panel (National Instruments Corporation, Austin, Texas) used to accept an external indicator signal for the software to obtain an image from the video digitizer; A communications port FASTCOM / IG232 (Industrial Computer Source, San Diego, California) used for communication in series with communication systems between refrigerators; - A Minolta Spectrophotometer for the calibration and initial verification of the program; A 3-chip JVC RGB camera (model GP-U5502) used as a video input source; A Microsoft Windows 95 operating system (Microsoft, Redmona, WA) under which the breakthrough occurred; - Microsoft Visual C ++ V4.0 used as main development environment including the C ++ language, antivirus, interface trainer, and Microsoft Bases Classes; A Matrox Imaging Library (MIL) Lite base unit interface sensor (Matrox Electronic Systems, Ltd.) to acquire RGB images from a video source; An I / O panel interface of NI-DAQ Function Library (National Instruments Corporation), which results in activation of the software by an external button connected to a hardware interface of a computer; and Programming Code JPEG library version 6 (Independent JPEG Group) for reading and writing JPEG formatted images for secondary storage devices.

B. Image Processing The arrangement comprises two subsystems, a subsystem of image processing of cattle and a subsystem of image processing of rib eyes. The measurements of the beef can be taken from the hot beef immediately during the slaughter process or after some cooling period. Rib eye measurements are typically taken after 24 hours of cooling. 1. Processing of Res Images The arrangement for obtaining images of cattle 20 is used to capture a visible spectrum image of a half full-length vaccine that travels along a conditioning rail during the killing process. The beef has been sectioned into two symmetrical halves along its dorsal axis. Two orientations of the res (visualized side of the res) are possible for each of the left and right halves of the res, resulting in four possible views: (a) face of the left bone; (b) each of the right bone; (c) face of the left skin; and (d) face of the right skin. The face of the left skin of the beef is used in the meat qualification method of Canada. The other views are not used in the Canadian method. As the stockings are lowered, the face of the bone or face of the skin can face the camera. An operator standing before the arrangement to obtain images of the res 20, manually rotates the half cattle in the orientation of the face of the skin. The arrangement for obtaining images of the res 20 establishes the difference between the left and right halves of the half cattle, and retains the images of the selected half of the beef for analysis. Various linear, two-dimensional and curvature measurements of the beef are made, as will be explained in detail herein. A total of approximately 400 measurements is made. Prediction equations can be developed based on these measurements to (i) predict the conformation of the beef, (ii) classify the cattle into groups by size (iii) produce precise coordinates for the automated manufacture of main cuts by robotic machinery. 2. Rib eye processing. After 24 hours of cooling, a cut is made between the 12th and 13th ribs of the animal to expose the muscle ongi ssim us dorsi, also called the rib eye. The manual observation of a cross section of the rib eye constitutes a conventional classification technique. The weight of the beef opens the cut so that the manual chamber 34 can be inserted into the resulting slit and take an image. There are two fundamental reasons why the rib eye is examined for meat rating purposes. In the first place, it is known that the qualification measurements of the rib eye are intimately related to the degree of the rest of the res. Secondly, the rib eye is the most expensive meat cut of the beef and, therefore, of the greatest importance for qualification purposes. In general, the analysis of the rib eye area includes the following steps: a. The lower edge of the rib eye area is detected. The lower edge of the rib eye muscle area is defined by a fat / air limit. Represents the outer edge of a cross section of the res. This step is given to determine if there is something in the image that interferes with the step of determining a threshold of the image. This step is optional. b. Threshold clarity levels are determined over the entire surface of the image of the res, to distinguish lean tissue from fat. It is presumed that the bottom is the entire area of the image below the lower edge of the rib eye area determined in step a. All the pixels below this edge are adjusted to black. c. The rib eye muscle limit is tracked. d. The area and percentage of marbling is determined within the limit of the rib eye muscle. and. The longitudinal axis of the rib eye muscle surface is determined. f. The maximum width of the rib eye muscle surface is determined, perpendicular to the maximum longitudinal axis. g. The approximate location of the subcutaneous fat layer is determined. Subcutaneous fat is found in the area between the outer limit of the hind and the edge of the rib eye muscle. h. The thickness of the subcutaneous fat is measured by dividing the longitudinal axis of the rib eye muscle area into four equal quadrants and measuring the thickness of the fat in the boundary between each of the quadrants.

II. Color Analysis of Meat Tissues The analysis of the beef of the present invention includes three main types of tissues, meat (muscle), fat and cartilage. Each of these types of fabric has chromatic characteristics that distinguish it.

The color of light is a function of its electromagnetic wavelength. You can distinguish seven colors or tones expressly designated in the spectrum of visible light, each of which represents a different wavelength: red, orange, yellow, green, blue, indigo and violet. The colors of light, which differ from pigmented colors, work in the form of additives. The absence of light of any color produces black. A combination of the three primary light colors, red, green and blue, is added to produce white light. There are three dimensions or attributes of color: tone, value or clarity; and intensity of color, or saturation. "Tone" is the specific name of a color. The tone depends on the dominant visible wavelength of a color. The wavelength associated with the specific tone indicates the position of that tone within the spectral range. The seven tones described above are considered pure tones, and can not be separated into simpler colors. Other visible tones are combinations of several different wavelengths of light (see, for example, Wal lschlaeger C, and C. Busic-Snyder, Concepts and Basic Visual Principles for Artists, Architects, and Designers, 1992, Wm C. Brown Publishers). "Value" or "Clarity" is the relative lightness or darkness of a color, as it would appear in a black and white photograph. The value of a color depends on how much light the color reflects. The values of the colors are called "tonalities" or "shades." The tonalities have values of greater clarity. The nuances have darker values. "Color intensity" or "saturation" refers to the amount of tone in a given color or the relative purity of a color on a color-to-gray scale. It is said that a color that has a high intensity of color is saturated with a monochromatic tone (a color). The tone, the value and the color intensity (tone, clarity, saturation) are interrelated. For example, at extremely high or low values, it is difficult to determine the tone or intensity of a color, since all colors appear respectively, very light (white) or very dark (black). In addition, the human eye different tones can reach their maximum intensity at different values. For example, yellow reaches its maximum intensity at a higher value (clarity) than violet blue. An excellent explanation of color analysis, representation and reproduction can be found in Benson's Manual of Televising Engineering, K. Blair edit. Edition Rev. McGraw-Hill Inc., New York, 1992. When these chromatic principles are applied to the tissues of an animal's beef, the different chromatic characteristics of muscle tissue, fat and cartilage can be distinguished in a digitized image of the area of the animal. rib eye. The muscle tissue, which has a value (clarity) and intensity (saturation) means, appears with a reddish tone. The cartilage, which has a low value at medium and low intensity, looks like an achromatic gray because the saturation or intensity of the tone (reddish) is low. The fat, which has a high value, seems bright white because it is difficult to discern the tone (reddish) at very high values. Accordingly, cartilage can be distinguished from muscle tissue or fat based on the low chromatic intensity of cartilage, and fat can be differentiated from muscle or cartilage due to the high value of fat. A number of systems can be used to define color for computer / video applications. In the RGB system, the individual amount of red, green and blue is defined in the color of a pixel. An RGB value can be represented in hexadecimal form (ie bright red FF000, bright red 00FF00, bright blue 0000FFR, black 000000, white FFFFFF). You can determine all the necessary information regarding the colors from the RGB color value. The tonality or tone is determined by the relative proportions between the red, green and blue values. Thus, CC33FF has a different tone of CC3300, the latter being a bluer tone. The clarity and saturation of the color are also determined by the relationship between the red, green and blue values. For example, 66FF99 has the same tone as 33CC66 (medium green) but is brighter. The RGB values can be converted directly to other color systems such as HSL (Hue, Saturation, Clarity). The HSL system is intuitively suitable for chromatic analysis of cattle images, since it makes direct reference to the distinctive characteristics of tone, chromatic intensity (saturation) and value (clarity) described above. It has been determined that muscle tissue, fat and cartilage have the HSL color characteristics set forth in Table 1.

Table 1 tone saturation clarity reddish muscle tissue medium average reddish fat (difficult to measure high measure) reddish cartilage (difficult low low-average to measure) The color of an object depends on the lighting. To objectively evaluate the color of the meat, the color of the image is corrected to known standard lighting conditions. First, the clarity of the image is equalized to compensate for differences in the level of illumination across the entire surface of the rib eye muscle or eye. The image is then adjusted to standard lighting conditions. To adjust the image in its clarity, an image of a uniform 18% gray card with a known RGB color value is recorded. The standard chromatic value of each pixel of the gray card image is used to adjust the chromatic value of each corresponding pixel in the res or rib eye muscle image. The RGB value of each pixel of the res or muscle image of the rib eye is adjusted to the known lighting condition as follows: Pixel (c) i 1"= xGri s true) 7- Camera gri (c) ± j where pixel (c) * ij is the adjusted value of the pixel and pixel (c) ij is the pixel to be adjusted located in row i and column j and c is a color component R, G (green), or B (blue) . True gray (C) ÍJ is the gray value known under standard lighting conditions and the gray camera (c) ij is the value perceived by the camera. The (RGB) ÍJ is transformed into (RGB) ij * after the clarity setting. The corrected clarity image is subjected to color adjustment by means of a calibration matrix A as follows: ccn 12 a13 (rgb) = (rgb) - - Ct? l Ct22 &23 &31 &32 CC33 The calibration matrix is obtained by comparing five colors (black, white, red, green and blue) to be calibrated with the same five colors under known standard lighting according to the following relationship : XA = B where X is a 5 x 3 matrix that represents values of r, g and b of the five color samples, A is a 3 x 3 calibration matrix to solve, and B is a 5 x 3 matrix of known constants, which represent values r, g and b of the five color samples under standard illumination. A solution of minimum squares for the calibration matrix A is obtained as follows: A = (X'X) ~ X'B where X 'is the transposition of the matrix X y (X'X) "1 is the inverse of the matrix (X'X).

III. Calibration of the Chamber. A white cross-linked panel with a grid of black lines spaced at 10 cm intervals is used as a background screen to scale the image of the res. Before starting a cattle analysis session, the background screen is displayed to determine the number of pixels in the frame of the image (scale). A threshold value is established to distinguish the white background from the lines of the grid. Since the panel is bright white and the grid lines are dark black, that is, they are the ends of the light range, the threshold value can be set arbitrarily as a medium clarity level (in a HSL color measurement method). In accordance with that illustrated in Figure 4, the digital image illustrates the white cross-linked panel 100 against a background 102 of empty space. To find the area (number of pixels) in each square of grid 104, the boundaries of each square 104 should be detected. Since squares have regular geometric shapes, the boundaries of each square are defined by the regular intersections 105 of the squares. lines of the grid 106. Halar the intersections of the lines of the grid 105 involves searching uniformly along each row and column of pixels of the image using a masking to locate a desired shape determined by the mask used. Once the intersections of the lines along the lines of the grid that define the right edge and the lower edge of the grid have been located, the whole grid can be extrapolated, since it is defined by a uniform geometric pattern of the grid. lines that intersect at right angles. The intersections 105 of the lines of the grid 106 of the cross-linked background describe a number of different right-angled shapes I. These could be described, likewise, as the forms that define the four vertices of a square. You can define the masks as identifying each of the possible shapes of the intersections. For example, if the screen illustrated in Figure 4 is analyzed by looking pixel by pixel, vertically in an upward direction and from right to left, along lines 108, the first intersection shape to be detected along the line of the grid that defines the lower edge of the grid will be a lower right vertex shape. A lower right vertex can be identified by a 5x5 pixel mask 100 according to what is illustrated in Figure 5, in which the H represent high intensity pixels and the L represent low intensity pixels. The mask 110 is defined so that the sum of the values of clarity of the positions of the pixels L is subtracted from the sum of the values of clarity of the positions of the pixels H. The difference obtained is indicative of the adjustment of the mask . Since a threshold determination process is used, the clarity levels are arbitrarily reduced to two numerical designations. For example, in a range of 0 - 255 levels of clarity, all intermediate levels of clarity can be reduced to a value of 10 (low clarity - black line area) or 200 (high clarity - white area) depending on if the pixel clarity value falls above or below a predetermined threshold value. Accordingly, a lower right vertex shape is detected when (sum of H) - (sum of L) = maximum difference for mask 110 illustrated in Figure 6. In this example, the maximum difference of mask 110 would be (9x200 ) - (16x10) = 1640. This difference is reached when the mask is located in such a way that each pixel H is located on a clear pixel (white background) of the digital image and each pixel L of the mask is located on a pixel dark (black line of grid 106) of the digital image. A better adjustment of the mask is indicated by a greater difference between (sum of H) - (sum of L).

Conversely, as illustrated in Figure 7, when the mask 110 is not aligned over the lower right corner shape, the difference between (sum of H) - (sum of L) is smaller. For the ill-fitting mask 110 illustrated in Figure 7, the difference would be ((3x200) + (6x20)) - ((10x200) + (6x20)) = -1400. Therefore, it is evident that mask 110 is not centered on a lower right vertex shape. The first form of the lower right vertex is located by looking through the bottom-up image, one column of pixels at a time, working from right to left. Once the first lower right vertex form (lower right vertex of the grid) is located, the mask 110 can be moved from right to left in the same row of pixels of the image to locate the lower right vertex shapes of the same row (the horizontal line of the lower end grid). By moving from right to left along the X axis to determine the positions of the first and second lower right vertex shapes, it is possible to determine the number of pixels between the positions of the lower right vertex shapes.

Other intersecting shapes 105 of lines 106, such as the upper right vertex shapes, can be detected by using correctly designed masks 110. Repeating this process along the Y axis to detect the lower right vertex shapes, intersections can be detected of the lines in the vertical line of the grid on the far right. By determining the number of pixels between the lower right vertex shapes of the vertical line at the right end of the grid, it is possible to calculate the number of pixels in each of the 10 cm x 10 cm squares. By calculating the number of pixels of the digital image of each square of 10 cm x 10 cm of the grid in the reticulated background, the arrangement can be calibrated to determine the real distance represented by each pixel. This calibration becomes especially useful when using digital images to take real measurements of the characteristics of the res.

B. Analysis of Reserved Images. Once the arrangement for obtaining images of hot cattle 20 is calibrated, the reticulated bottom panel 12 is replaced by a bottom panel 12 which has a great contrast with the beef. The preferred colors of the panel are blue and green. Preferably, panels 12 are substantially non-reflective. The first media res is placed in position in front of the background screen 10 and a digital image is recorded. As illustrated in Figure 8, analysis of the image begins by inserting a rough processing rectangle 112 around the portion of the image that includes the res (the image encompasses both the area in which the res is located, and an area surrounding of blue color). The processing rectangle 112 is positioned so that it is somewhat larger than the area of the res. The processing rectangle is divided into twelve areas of equal size 114 for image processing purposes. The clarity threshold determination is performed separately in each zone 114 to take into account the variation of illumination of the surface of the res. On the other hand, and preferably, during calibration, the camera 22 can be adjusted so that the frame of the image is not larger than the blue background, thus excluding any object that may interfere with the background. 1. Establishing Threshold on the Edge of the Res Image Five equally spaced points 116 are selected along the lower edge of the lower middle zone to continue the analysis using a masking technique. The mask 118 of 10 pixels in length is used in the Y coordinate and of a pixel in width in the X coordinate, according to what is illustrated in Figure 9. The mask 118 moves in ascending direction pixel by pixel in each of the columns designated by one of the five selected points 116. In each increment of one pixel, the sum of the values of the pixels L of the mask 118 is subtracted from the sum of the values of the pixels H. The values measured in the mask 118 are the values of clarity on a scale of arbitrary clarity such as "0". To -255, designating the clearest values with a larger number The limit between the background and the res is detected when the (sum of H) - (the sum of L) = greater value. H pixels are centered on a pixel of the image of the res and each of the pixels L of the mask is centered on a pixel of the background image The average of the low pixels (L) of the mask is determined.

This represents the measured clarity of the background image. Of the five samples taken, the one with the highest average background clarity is that used as a reference model. The level of clarity of the background is used to set a threshold of clarity in order to distinguish the background of the image (blue panels 12) from the image of the res (clear - covered in grease). A small constant is added to the level of background clarity (average of L pixels of the mask) to provide an error image. It is presumed that any pixel of the image that has a level of clarity higher than the threshold is a pixel of the res. On the other hand, and preferably, the edge of the beef image is determined based on the tone, instead of clarity. That is, the beef that has a reddish / yellowish hue can easily be distinguished from the background screen that has a blue hue. In order to establish the threshold of tone, the average tone of the background of the image is determined (blue panels 12), preferably from an image taken from the background without the presence of a res. Since the background is of a substantially uniform color, the approximate tone can be achieved by taking a very small portion of the background. With respect to the image of the res, if the tone of a given pixel of the image differs from the average tone of the blue background by a considerable predetermined amount (for example - more than 40 degrees on the HDL color wheel), in that case it is presumed that the pixel represents a portion of the image containing the res. 2. Tracking the Edge of the Res Image Once a position has been detected at the edge of the res (boundary bottom / res), the outline of the res is traced. As the background of the image (blue panels 12) is dark and the outer surface of the res is clear (adipose tissue), the tracking procedure traces the boundary between dark and light pixels. Working from the first pixel of the res (first pixel above a threshold established in the previous step) detected in the lower right edge of the boundary of the res, the contour trace of the res continues using a chain linkage algorithm of basic edge to draw the internal limit of the pixels of the res in the image in the direction of the movement of the hands of the clock. The edge chain linkage algorithms are known and the basic theory and practice of these algorithms have been described by González, Rafael C. and others, in Digital Image Processing, .Addison - Wesley, USA, 1993. Working in an upward direction in each column from the lower right edge of the frame of the image, the first pixel of the res (clear-fat) identified with the mask 118 illustrated in Figure 9 is detected. Starting with the first pixel of the res, the tracing clockwise through the nine contiguous adjacent pixels of the first pixel of the res to find the next pixel of the res (the next pixel has a clarity value greater than the threshold). Then the tracking process is repeated until the entire outline of the res has been drawn. In this way, each edge pixel of the image of the res is detected, thus tracing the contour of the image of the res. 3. Differentiation of the Left Exterior Image of Res. A. Find the tail. Once the contour of the res is drawn, the left and right images of the media res distinguish. According to the above, in the Canadian qualification technique, the left side of the res is analyzed. For use in Canada, the invention discards the image taken from the right half of the cattle. The left and right images of the media res appear in general as illustrated in Figures 10 and 11, respectively. It is possible to detect certain differentiated anatomical characteristics in each beef image. To distinguish the right and left halves of the res, the position of the tail 120 is first detected. To detect the position of the tail 120 in the image, the middle point of the list of the X coordinates in the outline is selected. of the res (starting from the furthest point in the bottom right). The remainder of the list of the X coordinates in the contour of the res is searched until the same X coordinate is found. Connecting these two X coordinates, a line 122 is defined that cuts the res in left and right halves. Since it is known that tail 120 is not in the left half of the image, the left half of the image can be ignored and only the right half of the image needs to be analyzed. According to what is illustrated in Figure 12, for each X coordinate there are at least two Y coordinates, one defining the upper edge 124 of the res, and the other defining the lower edge 126 of the res. The change of the values of Y along the line that defines the edge of the res is larger at the edge of the res where the tail 120 is located. b. Differentiation of the bony face of the face of the skin of the skin Once the position of the tail 120 is determined, the image is analyzed to determine if the image shows the face of the bones or the face of the skin. As illustrated in Figures 10 and 11, the surface of the skin face of the beef is covered by a relatively homogeneous layer of fat (light) 128. The surface of the face of the bones of the beef is characterized by alternating dark and light bands 130, representing, respectively, the lean tissue between the ribs and the rib bones. To identify the ribs, a horizontal line 132 (X axis) is defined at a selected distance within the boundaries of the side of the tail. The clarity value of each pixel is determined along the horizontal line 132. The rapid significant variation in the level of clarity along the horizontal line 132 indicates that the image represents the internal surface of the area where the lines are exposed. ribs. The level of clarity along the outer surface of the beef does not vary considerably since the outer surface of the beef is covered with a relatively uniform fat layer. Using the information regarding the position of the tail and whether the ribs are visible or not, it is possible to determine which half of the res and orientation is being visualized. The image of the left half of the face of the skin is used for the qualification in Canada, and for the qualification in the United States the image of the right half of the face of the skin is used. The image of the left half of the face of the skin is characterized by not exposing the ribs, and by the fact that the tail 120 does not appear in the upper part of the image. 4. Measurements of the Res As explained in more detail below, a variety of measurements are made on the contour of the image of the res. These measurements are useful as independent variables in predictive equations to estimate various parameters of beef qualification. In every predictive equation, not all possible types of measurements are used. For example, as described in the Examples herein, a highly accurate equation for predicting yield for sale in cattle has been developed without using any of the oblique angle or curvature measurements described herein as independent variables. The measurement techniques of cattle described here are especially useful, since they do not require any measurement or manual analysis of the cattle. On the contrary, certain distinctive anatomical characteristics of the res can be identified by image analysis techniques. The rest of the measurements can be made depending on the position of the anatomical features initially identified. to. Locating Anatomical Reference Points A number of anatomical characteristics of the animal can be distinguished in each image contour of the animal. As illustrated in Figure 13, the tail 120, the lower ankle 134, the dorsal rear leg 136, the lower armpit 138, and the upper armpit 140 are characterized by prominences or depressions along the contour of the res and can therefore be localized by image analysis techniques. To locate one of the anatomical characteristics referred to above, the analysis of the contour of the image of the res is restricted to a short segment of the contour of the image of the res within which it is known that the anatomical feature is located. . Within each selected short segment of the image contour, the anatomical feature of interest typically appears as a protrusion or depression of the contour of the res image. As illustrated in Figure 14, (a representation of the lower armature 138) it can be estimated that the protrusion or depression is an apex 142 defined by two intersecting lines 144, 146. Although the apex 142 shown in Figure 14 defines the position of the lower armature 138 along the linear segment delimited by 148, 150 relatively precisely, the actual position of the apex 142 and the direction of the lines (vectors) 144, 146 are not essential. A line 152 is projected perpendicular to line 154, bisector of the angle formed by the apex 142. The point furthest along the contour segment of the image of the car analyzed from the perpendicular line 152, as defined by line 156, represents the lower armpit 138. This method can be used for any anatomical point of reference of the contour of the res, which can be defined as protrusion or depression of the contour of the image of the res. The projection of the virtual lines 144, 146 and the bisection of the angle formed between the lines 144, 146 is not of critical importance. This is simply a graphical representation of a method by which to place the position line 152 approximately perpendicular to the apex 142, so that the apex 142 is the most distant point along the line segment bounded by 148, 150, with respect to to line 152. b. Locate Other Anatomical Points Other anatomical points, described in the present as the back, rib, quad, neck, ventral hindquarters, palette, 12th rib, and chest, can be located as proportions of the distance between certain anatomical reference points previously determined . According to what is illustrated in Figure 15, the distance in the X coordinate between the tail 120, located at the bl point of the contour of the res, and the upper armpit 140, located at the point b9 of the contour of the res, is define as di. The distance at the X coordinate between the dorsal rear quarter 136, located at point b3 of the contour of the res, and the lower armpit 138, located at point bd of the contour of the res, is defined as d2. The anatomical points can be located in the manner shown in Figure 15 and Table 2.

Table 2 Anatomical point proportion of di proportion of d2 loin (bl3) 0.19 rib (bl2) 0, 60 rump (bll) 0.74 cogote (blO) 1.00 ventral hind quarter (b4 0.00 pallet (b5) 0.19 12th rib (b6) 0.52 kill (b7) 0.73 c. Define Main Cuts Once the anatomical points have been determined, the main cuts can be calculated. The main cuts are the first cuts of meat during the meating process and that include the neck, the loin, the rib, the rump, the shoulder, the matambre and the breast. The main cuts are illustrated in Figure 16. The lines of the main cuts are identified by their anatomical end points. The point pl is located at 42% of the distance along the line bl2-b6. Point p2 is located at the intersection of a line drawn in an upward direction from pl, parallel to line bl2-bl3 and a line drawn to the right of bl3, parallel to the Y axis (note that the axes in Figure 16 they are inverted with respect to their usual positions). Point p3 is located at the intersection of line p2-b5 with a line drawn to the right of bl, parallel to the axis Y. Point p4 is located at 45% of the distance along line bl2-b6. Point p5 is located at the intersection of line bll-b7 with a line drawn in a downward direction with respect to p4, parallel to line b6-b7.

The front main cuts are separated from the main cuts of the hindquarters by the line b6-bl2. The main cut of the quad is delimited by the edge contour of the res and the lines bl-p3 and p3-b5. The main cut of the skirt is delimited by the outline of the edge of the res and the lines bl-p3, p3-p2 and p2-bl3. The main cut of the spine is delimited by the outline of the edge of the beef and the lines bl3-p2, p2-pl and pl-bl2. The main cut of the rib is delimited by the contour of the edge of the rib and the lines bl2-p4, p4-p5 and p5-bll. The main cut of the back is delimited by the contour of the edge of the res and the lines b5-p2, p2-pl and p5-b6. The main cut of the matambre is delimited by the outline of the edge of the beef and the lines b6-p4, p4-p5 and p5-b7. The main cut of the chest is delimited by the outline of the edge of the beef and the lines b6-p4, p4-p5 and p5-b7. \ ^ Record the length of the res (length in pixels of the line b3-b9) and determine the surface area of the eight main sections. The sum of the surfaces of the eight main cuts gives the total area of the res. d. Linear Measurements Linear measurements are made to divide the res into six linear regions, the hind quarter, the lower quad, the upper haunch, the lower loin, the middle loin and the shoulder. According to what is illustrated in Figure 17, line b3-b9 defines the length of the res. The longitudinal line of the res b3-b9 also divides the res in the dorsal / ventral direction. Before making the following measurements, rotate the image so that the line b3-b9 is parallel to the X axis (as before, the axes of Figure 17 are inverted). The lines separate and define the linear regions. The line b2-cl, which separates the region of the hind leg from the region of the lower haunch, projects from b2 perpendicular to the longitudinal line of the res b3-b9, to intersect the opposite side of the contour of the res in cl.

The line bl-c2, which separates the lower anca from the upper anca, projects from the tail (bl) perpendicular to the longitudinal line of the res b3-b9, to intersect the opposite side of the contour of the res in c2. The line c8-c3, which separates the region of the upper haunch from the lower rear region, is drawn perpendicular to the longitudinal line of the res b3-b9, at 33% distance along the line b3-b9. The line c7-c4, which separates the lower rear region from the central rear region, is drawn perpendicular to the longitudinal line of the res b3-b9, at 53% distance along the line b3-b9. The line c6-c5, which separates the central rear region from the shoulder region, is drawn perpendicular to the longitudinal line of the res b3-b9, at 80% distance along the line b3-b9. The line b3-b9, which defines the length of the res, forms the basis for a series of measurements that can be useful as independent variables in equations to estimate the res rating parameters. Each of the six linear regions (the hind leg, the lower haunch, the upper haunch, the lower dorsal part, the central dorsal part and the shoulder) is divided into a plurality of fine divisions, for example 10 divisions, perpendicular to the line b3-b9. The distance of the line b3-b9 to the ventral or dorsal edge of the contour of the image of the res in each fine division can be used as an independent variable. Arithmetically dividing each of the linear regions into many equal divisions is a convenient method to produce a large number of reproducible measurements of the beef image, any of which can serve as an independent variable in prediction equations for the qualification parameters of the cattle. and. Oblique Angle Measurements As illustrated in Figure 18, the distance of b9 is determined to each of cl, c2, c3, c4, c5, c6, c7, c8, bl, and b2. The numbering and the letters in Figure 18 correspond to those in Figure 17. The angles Zc6-b9-c5, Zc7-b9-c4, Zc8-b9-c3, Zbl-b9-c2 and Zb2-b9-cl are measured. .

F. Curvature Measurements The cubic curvature functions of the contour segments of the res b2-b3 are estimated by known methods according to what González and others described.

IV. Image Analysis of the Rib Eye Muscle A. Calibration of the camera Using the disposition to obtain images of the rib eye 32, a digital image of a gray card is obtained in 18% to calibrate the camera for the level of clarity. The image is pre-processed with a low-pass filter that allows only minute variations of clarity to pass through. This eliminates variations in the level of clarity throughout the image. The average clarity of the uniform gray image is determined using a spectrophotometer and each pixel of the image is corrected in its clarity in the manner described above. These data are used during the image analysis of the res to compensate for the lack of uniformity of the light source, and to improve the accuracy of the color measurements. 2. Color Balance and Image Scale The color balance and image scale information is determined from an image acquired from a standard color calibration sheet marketed by Kodak and used for color correction in the manner described above. A standard color calibration sheet consists of calibration squares of black, white, red, green and blue, and a large calibration square of the white area of the image, each side of which has a length of eight mm. The RGB values of the color calibration squares are recorded for use in color correction. The scale of the image is determined by calculating the surface of the calibration square of the image area on the color calibration sheet. A row of pixels is searched approximately at a midpoint of the image from left to right. As the background of the color calibration sheet is dark, and the area calibration square of the image is clear, it is presumed that any pixel along the searched horizontal row that has a value of clarity lower than the average value of the clarity scale is a background pixel, and the pixels above the average brightness value fall within the calibration square. The number of clear pixels per row is added. This process is repeated for each row above and below the central row of pixels of the image until finding the first row of both directions (upper and lower) that does not contain any clear pixels. When the limit of the entire calibration square has been determined, the area of the calibration square (in number of pixels) is calculated for use in verifying the accuracy of the next image area calculation based on a matrix solution. During the determination of the image area, the upper left (SI), the lower left (II) and the lower right (ID) vertices of the calibration square are recorded. The length of the pixels of the vectors II-SI and II-ID is calculated and correlated with their real lengths (8 cm). Solve a simple matrix to find kX (the scale of X) and kY (the scale of Y) and compare the results with those determined in the previous step. If the results differ beyond an allowable limit (ie 50 square pixels), the calibration step of the image is repeated. The camera is rotated slightly in each direction between each step to ensure a calibration of image size that does not vary with rotation.

B. Obtaining the Image of the Rib Eye Muscle After 24 hours of cooling, a cut is made between the 12th and 13th ribs of the beef. The weight of the beef opens the cut so that the rib eye chamber 34 can be inserted into the resulting slit and an image can be taken. The arrangement for obtaining images of the rib eye 32 is used to take a digital image of the rib eye area. According to the above, the rib eye is the vulgar name of the ongi muscle ssimus dorsi. There are two main reasons to examine the rib eye muscle for meat qualification purposes. In the first place, it is known that the qualification measurements of the rib eye muscle are intimately related to the degree of the rest of the res. Secondly, the rib eye is the most expensive cut of beef and, therefore, its qualification is of the highest importance. The digital image of the area of the rib eye muscle is corrected for any deviation of clarity by adding the deviation of clarity of each pixel previously calculated by each pixel in the uniform gray clarity calibration image to the clarity value of each pixel. in the image of the rib eye muscle area.

C. Previous Processing of the Image of the Eye of the Rib Eye Muscle. 1. Tracing of the Outer Fat Rim The image of the rib eye muscle generally appears as illustrated in Figure 19. The rib eye (muscle ongi ssimus dorsi) 200 appears as a generally elliptical dark region in the center of the eye. image. The subcutaneous fat 202 appears as a clear band under the rib eye muscle 200. The bottom 204 is the open space that surrounds the pendant and has a dark appearance. Muscle tissue 206 may be present, which is in contact, but is not part of the rib eye 200. As illustrated in Figure 20, the image is scanned from bottom to top along a plurality of spaced-apart columns. short distance of pixels 208, spaced apart at a distance of about 5 mm (actual distance in the rib eye muscle) to locate clear elevations of the clarity of the pixels. In this step, the image is analyzed in increments of 5 pixels to quickly approximate the outer fat edge 210 of the image. Each clear elevation of pixel clarity of a column of pixels 208 identifies a point 212 located approximately along the outer edge of the grease. Points 212 are joined along a minimum energy path. A minimum energy path has very little internal energy. In other words, it is the most straight line possible. A straight line has no internal energy, while a zigzag line has high internal energy. This path defines the outer fat edge 210. If the outer fat edge 210 does not cover the entire image in the horizontal direction, it extends horizontally as a straight line from left to right towards the boundaries of the image. The outer fat edge 210 of the beef may be damaged by small cuts or notches. These will appear as irregularities or depressions in the outer edge of approximate fat. Therefore, an isolation process is used to match the outer edge of rough fat and be able to ignore small irregularities. As illustrated in Figure 21, deterioration of the outer fat rim 210 probably appears as a small indentation 214 in the outer rim of the approximate grease. Most depressions can be represented as three "turns" of the approximate line 216. Normally, the approximate line forms a 180 ° angle around a point 212. In a depression 214, line 216 forms at least three turns 218, 220, 222. It is recognized that a rotation occurs when the angle of the approximate line 210 around a point 212 differs substantially from 180 °. The first and last turn in a small area indicate the beginning and end of a deviation of the approximate fat outer edge line 210. The points 212 in the first turn 218 and the last turn 222 join to eliminate the depression 214 The points within the depression are transferred to the new leveled line. All the pixels below the line of the approximate grease outer edge 210 are adjusted to a minimum clarity level (black) to avoid strange interference with the image processing of any of the objects appearing at the bottom of the image. 2. Differentiation of the Images of the Left and Right Side of the Eye Rib Eye If there were images of the rib eye muscle on the right side they will be turned horizontally and processed as left side images. All algorithms can be designed to process the rib eye muscle images on the left side, if the provision is intended for the Canadian qualification standards. As illustrated in Figure 20, the lower edge of the rib eye muscle is tuned upwardly on one side. In the right half of the beef, the rib eye is tuned upwards on the right. The value of the two-point Y coordinate 224, 226 is determined along the line of the outer fat edge 210, spaced approximately 20% inward from the left and right edges of the image. If the furthest point to the left 224 is closer to the top of the image frame than the furthest point 226, the image represents the right side of the res. If the point at the right end 226 is closer to the top of the image frame than the leftmost point 224, the image represents the left side of the res. 3. Establishing an image threshold A clearness threshold of the whole image is established to distinguish the pixels representing the muscle tissue (medium clarity value) of the pixels representing the fat (high clarity value) and a saturation threshold to distinguish the pixels representing the muscle tissue (medium saturation) of the pixels that represent the cartilage (low saturation). to. Increase of Image Contrast Each pixel of the rib eye muscle is remapped over a high contrast image ("IMap"). The level of clarity of each pixel of the IMap is calculated by the formula: IMap (i) = 255 (1.0 - (Y (i) / 255)) 2/3 in which Y (i) is the clarity component (0: 255) of the pixel value in the "i" position of the image . This function enhances or exaggerates this difference in clarity between the muscle tissue pixels and the fat pixels, thus increasing the contrast of the image. This function also results in an inversion of the color so that the pixels of the muscle tissue, which normally appear and have a low clarity number (0: 255), appear clear and have a high number of clarity (0: 255) in the IMap, and the fat pixels, which normally appear clear and have a high clarity number (0: 255), appear dark and have a low clarity number (0: 255) in the IMap. b. Determination of Meat Color Clarity Threshold The IMap image of the rib eye muscle is divided into six zones of equal size before performing the threshold analysis. The determination of the color threshold is carried out separately in each zone. Any portion of the image that appears below the outer edge of grease is discarded. This increases accuracy, since the color of the middle muscle tissue can vary in different parts of the imigen.

In each section, the threshold of clarity of all the rib eye muscle pixels is determined. An arbitrary value of 55 is established on a scale of 0: 255 as the cut-off value between the clarity of the muscle tissue and fat pixels. Any pixel that has a value of clarity lower than 55 is presumed to represent fat (clarity is inverted in the IMap). It is presumed that the rest of the pixels represent muscle tissue, or cartilage. c. Determination of the Color Saturation Threshold of the Meat In each section, the color saturation of the average muscle tissue pixel (the pixels that have not been defined as fat in the determination of the previous step) is determined. A threshold saturation level is established by subtracting a constant value from the color saturation level of the average muscle tissue pixel, thus producing an error image. If a pixel has a saturation level lower than the threshold value, it is presumed that it represents cartilage. If a pixel has a saturation level higher than the threshold value, it is presumed that it represents muscle tissue. For the rest of the steps, the cartilage is treated as an equivalent of fat. Accordingly, the pixels representing muscle tissue are distinguished from the pixels representing any other tissue, either fat or cartilage. 4. IMap in reduced scale. An IMap is created on a reduced scale, designated "QIMap" to reduce the amount of data to analyze in some of the following steps. The IMap is preferably reduced by a ratio of 8: 1, although other proportions can be used to form the QIMap. In case, for example, of using a scale reduction ratio of 8: 1, the color information of 64 pixels of the IMap is averaged (eight in the dimension x and eight in the dimension y) and becomes the information of the color average for a pixel of the QIMap.

. Location of the Eye of the Rib Eye Inside the Image by Bubble Analysis. Bubble analysis is used to distinguish the rib eye from the foreign muscle tissue that appears in contact with the rib eye muscle, but which is not an anatomical part of the rib eye muscle. In bubble analysis, the muscle tissue pixels of the QIMap are grouped into four connected objects. In a grouping of four connected objects, it is considered that the pixel of an image is in the same group or bubble as a pixel immediately adjacent to the left, right, top or bottom. The pixels of the adjacent vertices are not considered part of the same bubble. This can be compared to the grouping of eight connected objects, in which a pixel is considered to be part of the same object as a pixel that occupies any of the eight pixels of the surrounding positions. There are known techniques for analyzing four connected objects and eight connected objects, which have been described by González and others, upra. In the QIMap bubble analysis, only those pixels that have QIMaps that exceed the threshold value of muscle tissue are considered part of a bubble. The threshold of clarity is established by subtracting an arbitrary constant of the IMap value from the average non-fat pixel of the entire image. The analysis is done from left to right, from top to bottom. However, other ordered analysis patterns can be used, such as from left to right, from bottom to top. The analysis continues until the first pixel of muscle tissue in the image is detected. The pixels immediately above and to the left of the first pixel of muscle tissue are examined to determine if one or both are also pixels of muscle tissue. If the analysis proceeds from top to bottom, from left to right, the pixels above and to the left of the first pixel of localized muscle tissue do not constitute pixels of muscle tissue. The process continues following the same pattern until detecting the next pixel of muscle tissue. Once again, the pixels that are immediately above and to the left of the first pixel of muscle tissue are examined to determine if one or both are also pixels of muscle tissue. If so, then it is known that the pixel that is currently being examined is part of the same bubble as the pixel that is immediately above and / or to the left. If it is determined that the pixel is part of a first bubble, and it is subsequently determined that it is also connected to a second bubble, it is concluded that what originally appeared to be two bubbles, actually forms a single bubble.

The larger bubble detected during the bubble analysis is indicative of the position and size of the rib eye muscle. 6. Estimating the Pointing of the Rib Eye Muscle in the Image. A small area is marked within the rib eye image in which the rib eye muscle itself is actually located defining a processing area that denotes the rib eye box around the rib eye muscle. The precision of the color discrimination of the different tissues increases as the foreign tissue of the rib eye muscle area of the analysis is eliminated as much as possible. To define the box of the rib eye, the approximate position of each of the upper, lower, left and right edges of the rib eye muscle is determined. to. Location of the Right Edge of the Rib Eye Muscle. Starting at the center of the bottom edge of the QIMap and working to the right, each column of pixels is examined from the bottom up, stopping when the first pixel of muscle tissue is located. As illustrated in Figure 22, the length of the column of the non-muscle tissue pixels 228 is substantially prolonged when the right edge 230 of the rib eye muscle is reached. Therefore, when the length of the column of non-muscle tissue pixels grows, it is known that the position of the right edge of the rib eye muscle has been detected. This column is defined as the right side of the rib eye box. b. Location of the Lower Edge of the Rib Eye Muscle. Starting at the center of the right edge of the QIMap and working towards the lower edge, each row of pixels is examined from right to left, stopping when the first pixel of muscle tissue is located. As illustrated in Figure 23, the length of the rows 232 of the non-muscle tissue pixels is substantially prolonged when the lower edge of the rib eye muscle is reached. Therefore, when the length of the row of non-muscle tissue pixels grows, it is known that the position of the lower edge of the rib eye muscle has been detected. This row 234 defines the underside of the rib eye box. c. Location of the Upper Edge of the Rib Eye Muscle. According to what is illustrated in Figure 24, a row of pixels 236 of the QIMap is selected at approximately 5 cm (actual measurement on the rib eye muscle) above the underside of the rib eye box. The total number of muscle tissue pixels in the row is tabulated to the column where the right edge of the rib eye muscle has been located. Moving upwards towards the top of the image, one row of pixels at a time, the tabulation of the total number of muscle tissue pixels continues until the sum of the lengths of the pixels in the row is less than 2.5 cm, in based on the calibration of the image size. The first row 238 detected in case the sum of the lengths of the pixels is less than 2.5 cm is presumed to constitute the upper edge of the rib eye muscle and defines the upper side of the rib eye box. By adding up the total number of muscle tissue pixels in a row, instead of using only continuous examinations of the muscle tissue pixels, the lack of continuity of the edge of the rib eye muscle is ignored for these purposes of approximation of the location of the muscle. upper edge of rib eye muscle. d. Location of the Left Edge of the Rib Eye Muscle. As illustrated in Figure 25, a pixel column 240 of the QIMap that is approximately 13 cm (actual measurement on the rib eye muscle) is selected to the left of the right side of the rib eye box and is determines the total length of the number of pixels of muscle tissue in the portion of the column limited by the upper and lower sides of the rib eye box. This measurement is taken in each column, working to the left, until the total length represented by. the muscle tissue pixels of the column are less than 1.5 cm. This column 242 indicates the position of the left edge of the rib eye muscle and is defined as the left side of the rib eye box. 7. Determination of Secondary Threshold to Improve Discrimination of Muscle / Fat Tissue. The QIMap rib eye box is projected onto the IMap and the threshold is set for the pixels within the rib eye box over which it has not been determined to represent fat or cartilage in step IV (C) (3) in each of the six zones calculated in step IV (C) (3). Since many pixels representing clear cartilage or fat have already been identified by the determination of saturation threshold and clarity of step IV (C) (3), and the area outside the rib eye box can be discarded, the threshold of clarity of the rib eye muscle can be obtained with greater sensitivity. This makes it possible to distinguish fat that has a slightly dark or reddish tint from lean muscle tissue. 8. Tracing the Edge of the Rib Eye Muscle. The tracing of the edge of the rib eye muscle is performed on the IMap. The rib eye box is divided into upper and lower halves and three vertical columns to form six equal zones. Tracking starts in the pixel column in the center of the bottom side of the rib eye box. Working from the bottom up, the pixel column is examined until the first pixel of muscle tissue is located. It is presumed that this is a pixel of the edge of the rib eye muscle. The rib eye muscle is tracked using an eight-way edge tracking technique described above, now tracing in the direction opposite to the direction of clockwise movement. The last ten positions of the edge pixels are temporarily saved. The tracking around the rib eye muscle proceeds until once again the first edge pixel of the rib eye muscle is reached. If the area defined by the edge of the rib eye box is too small, ie below an arbitrary cut-off value, the scan is discarded and the edge scan is repeated from a new starting point upwards with respect to the first edge pixel and above the tracked area rejected. This avoids errors produced as a result of starting the tracking of a pixel of muscle tissue that is not at the edge of the rib eye muscle. to. Identification of Strange Muscle Tissue Attached to the Rib Eye Muscle. Anyway, strange muscle tissue may appear attached to the rib eye muscle that is not part of the rib eye muscle based on the bubble analysis. Foreign muscle tissue can be identified and removed during the tracking of the rib eye muscle.

Removal of foreign muscle tissue involves tracking inside the foreign muscle, thus severing it from the rib eye muscle. At the moment, "cutting", "cutting" or "cutting" means the process of excluding the extraneous tissue from the rib eye muscle by estimating the actual edge of the rib eye muscle and tracking within the foreign muscle tissue. As illustrated in Figure 19, the rib eye muscle edge 205, while generally relatively straight, typically abruptly rotates to the outside where there is foreign muscle tissue adhered. Therefore, those sections of muscle can be identified by abrupt rotations at the edge of the rib eye muscle. According to the one illustrated in Figure 26, during scanning of the rib eye muscle, a group of 10 pixels at a time is examined, a group of 10 pixels moving in increments along the tracked edge 242. The ZABC angle in each incremented step. It is known that the edge of the tracked rib eye muscle rotates abruptly when ZABC falls below a selected acute angle such as 45 °. As illustrated in Figure 26, ZABC will only be less than 45 ° when point B is near vertex 244 of the rotation. A rotation is defined as an area along the edge of the rib eye muscle from the point along the edge (position of B) where ZABC falls for the first time below 45 ° (or other prescribed angle) and the point along the edge where ZABC exceeds 45 ° for the first time. Figure 27 is a symbolic representation of the contour of rib eye 200, illustrating abrupt rotation 246. When abrupt rotation 246 has been identified, a line 248 is drawn, which divides the angle formed by rotation 246.

A somewhat elliptical or oval figure 250 is drawn around the rib eye muscle 200. The ellipse 250 approximates somewhat to the general shape of the rib eye muscle and fits intimately into the interior of the rib eye box, touching the box of the rib eye on each of its four sides. A line 252 is drawn from the center 254 of the ellipse through the vertex of the rotation until it intersects a point 256 of the ellipse. A line 258 is drawn through the point of the tangent ellipse thereof. Lines 248 and 268 are compared. If they are to a certain extent parallel, rotation 246 is also considered a site for cutting. Further consideration is given to turn 246 to cut because the direction of some parallel way of lines 248 and 258 suggests that the proposed cut (near line 248) would follow the edge of the estimated rib eye muscle, instead of cutting toward the center of the rib eye muscle along an atypical path of a rib eye muscle contour. If lines 248 and 258 are not generally parallel, rotation 246 is rejected as a potential cut site, and tracking is continued since the proposed cut could not follow the estimated edge of the rib eye muscle, but would instead be directed towards the rib eye. rib eye muscle center. b. Determination of the Path of a Cut In accordance with what is illustrated in Figure 28, if the cut in an abrupt rotation 246 identified in the previous step has not been rejected, a series of lines 260 are drawn starting from the vertex of the rotation 246 approximately 60 ° on each side of the line that cuts the rotation. The sum of the values of clarity is recorded along each line 260. The recording ends when the line reaches a significant adipose body (ie, 4 or more consecutive pixels). This may represent rib eye muscle edge fat 262 or intramuscular fat 264. As the colors are inverted in the IMap, the high clarity values represent muscle tissue. A line that has a high sum of clarity constitutes a long line that crosses the muscle tissue, or is interrupted by a few pixels of fat. A line with a low sum of clarity is probably a short line through the muscle tissue, which ends up in the layer of fat surrounding the rib eye muscle or an intramuscular fat bubble. The line that represents the most probable cut path is, therefore, the line that has the lowest sum of clarity. If the line that has the lowest clarity sum still has a relatively high sum of clarity (above a selected threshold), the cut will be made, since the proposed cut line is long, and consequently is likely to project substantially into the rib eye muscle instead of toward the edge. If the proposed cut line is below the sum of threshold clarity, the possibility of cutting shall be considered. As illustrated in Figure 27, a small ellipse 266, substantially smaller than the ellipse 250, is traced within the ellipse 250. The small ellipse 255 is substantially small, so it always falls within the contour of the rib eye muscle. . If the starting point or end point of a proposed cut falls within the small ellipse, the cut is rejected. If neither the starting point nor the end point of the proposed cut falls within the small ellipse, the cut is made. Using this process, cuts can be made to exclude foreign muscle tissue from the tracked contour of the rib eye muscle. As illustrated in Figure 19, the boundary between a segment of foreign muscle 206 and rib eye muscle 2000 is often dotted with a chain of small fat globules that form an interstitial seam of fat 268. The above-described slice analysis selects a series of short cuts from one fat globule to the next, which, in most cases, accurately describes the edge of the rib eye muscle. c. Confirmation of Edge Tracking. In certain cases, especially where there is no interstitial seam of fat between the rib eye muscle and a piece of foreign muscle tissue, the process described in the previous step fails to successfully cut a piece of foreign muscle tissue. Therefore, a technique described herein is employed as a progressive angle analysis used as a guard against failed cuts. As illustrated in Figure 29, a line 270 is projected from the center of the ellipse 250 (illustrated in Figure 27) to each point along the edge of the tracked rib eye muscle, progressing consecutively in the opposite direction to the direction of displacement of the hands of the clock, starting from an arbitrary point of origin 274. Since the edge of the tracked rib eye muscle is of a generally regular shape, with few abrupt turns, line 270 appears to sweep counterclockwise through points 274-282. Consequently, angle 284 continues increasing as the sweep of line 270 continues. Between points 282 and 286, the sweep of line 270 reverses and moves in a clockwise direction. Therefore, the angle 284 becomes smaller. Between points 286 and 288, angle 284 is enlarged once again. Point 288 is collinear with points 272 and 282. The angle change 284 indicates the location of the abrupt rotation at 282. As illustrated in Figure 30, as in step IV (C) (8) (a ), abrupt rotation at point 282 is divided, and a series of radiated lines 290 is projected. Unlike step IV (C) (8) (a), lines 290 do not end when reaching fat, but rather at otherwise, when the rib eye muscle contour 205 is reached. The shortest line 290 is selected, line 290 which runs to point 292, for possible cutting. The cut along line 290 from point 282 to point 292 would cut muscle segment 294. The longer axis 296 of muscle segment 294 is determined. If line 290 is less than about half the length of line 295, the following cut criteria are examined. If line 290 is greater than about half the length of 296, cutting is not performed. The approach of the edge of muscle segment 294 (along the contour of the rib eye muscle between points 282 and 292) is made as a series of short straight lines, each approximately 20 pixels long, to define a polygon of many sides. The polygon is divided into triangles, and a surface formula of the triangle is used to determine the surface of each triangle of the polygon. The sum of the areas of the triangles that make up the polygon is equal to the area of the polygon and is used as an estimate of the segment area of muscle 294. It is determined at the ratio of the area of the muscle segment 294 that falls within the projected ellipse 250 The proposed cut will be rejected if any of the following criteria is met: the area of the muscle segment 294 that falls outside the projected ellipse 250 is greater than 15 cm2 (on a scale of 1: 1 with the real rib eye); the area of the contour of the rib eye that will remain after excision of muscle segment 294 would be less than 50 cm2 (on a scale of 1: 1 with the real rib eye); or less than half of the total area of the muscle segment 294 falls outside the projected ellipse 250. If neither of these criteria is met, the cut is made. 9. Determination of the area of the Trapped Eyeball Muscle. After completing the tracing of the rib eye muscle contour, the number of pixels in each row is tabulated within the contour of the rib eye muscle tracked in the IMap.

These data are easily related to the image size information obtained during the calibration of the camera. It is determined that each pixel within the contour of the rib eye muscle is a pixel of fat or a pixel of muscle tissue based on the determination of the threshold of clarity. All the color information of the pixels in the original acquired image corresponding to the muscle tissue pixels identified in the IMap is recorded and used to calculate a precise average color of the rib eye muscle tissue. The middle color of the muscle tissue of the rib eye is corrected with the color information obtained during the calibration of the camera. The medium color information of the rib eye muscle tissue can be used for meat rating purposes. As the number of fat pixels within the contour of the rib eye muscle is known, the proportion of intramuscular, or marbled, fat can be determined by dividing the number of fat pixels within the contour of the rib eye muscle by the number total of pixels within the contour of the rib eye muscle. All the intramuscular fat pixels are subjected to bubble analysis, according to the above described. Larger fat bubbles than a selected threshold size are removed from the marble percentage calculation.

. Determination of the Thickness of the Subcutaneous Fat Layer. As illustrated in Figure 31, it may be difficult to distinguish the subcutaneous fat layer 202 employed for the purpose of rating other fats 298. The subcutaneous fat layer 202 is limited by the contour of the rib eye muscle 205, and the border of fat 210 traced in step IV (C) (1). Generally, a thin dark line 300 can be detected which separates the subcutaneous fat layer 202 from the other fat 298. As illustrated in Figure 32, starting from the rightmost extreme column of pixels 302 on the contour of the muscle of the rib eye 205, the length of line 304 is determined between fat edge 210 and the contour of rib eye muscle 205. The length of line 304 defines the thickness of subcutaneous fat layer 202 in the spine column. pixels of the right end 302 on the contour of the rib eye muscle 205.

Using the same method, the thickness of the subcutaneous fat layer 202 is measured at intervals of approximately 3 mm (measurement on the rib eye muscle), working to the left. Typically, the other fat 298 is present primarily on the left side of the rib eye muscle image (rib eye muscle on the left side of the image) and is not located on the right side of the muscle image of the rib eye. rib eye. Therefore, the thickness of the subcutaneous fat layer 202 is very easily determined on the right side of the rib eye muscle image. Using a LLHHHHH mask, an attempt is made to locate the thin dark line 300 by scanning upwards along the columns at intervals of five columns, and moving to the left. If a thin dark line 300 is found, it is used to define the inner edge of the subcutaneous fat layer 202 by connecting the points found along the thin dark line 300. This defined inner edge is rejected if it turns out that the thickness of The subcutaneous fat layer to the left of the image is substantially different from the average thickness of the subcutaneous fat layer to the right of the image. If no thin dark line 300 can be detected, or it is too weak to be reliably detected, it is estimated that the position of the inner edge of the subcutaneous fat layer 202 where it meets another fat 298 (which would be in the same position as the thin dark line 300) is equal to the average thickness of the subcutaneous fat layer at to the right of the image, where there is little chance of finding another type of fat 298. Therefore, a line is drawn at a distance upward from the fat edge 210 which represents the average thickness of the subcutaneous fat layer 202 a the right of the image. The inner fat edge is then smoothed using the process described in step IV (C) (1). 11. Determination of the Length of the Rib Eye Muscle. According to what is illustrated in the Figures 33A to 33C, the points of the left end 306 and the right end 308 of the contour of the rib eye muscle are selected. A series of lines is projected from point 306 to a series of points 310 along the contour of the rib eye muscle near point 308. The length of each line is compared from point 306 to point 310 with the length of the line from point 306 to point 308 to define which line is the longest. Similarly, a series of lines is projected from point 308 to a series of points 312 along the contour of the rib eye muscle near point 306. The length of each line is compared from point 308 to the point 312 with the length of the line having the ends 308 and 306 to determine which of the lines is the longest. 12. Determining the Rib Eye Muscle Width According to what is illustrated in Figure 34, line 314 defining the rib eye muscle length is divided into numerous divisions (about 200) equal 316. The distance from each point-318 of the rib eye muscle contour 205 above line 314 of the pixel column in each division 316 is added to the distance from each point 320 about the contour. of rib eye muscle 205 below line 314 of the pixel column in each division 316. The column that has the largest total length from point 318 to line 314 plus line 314 to point 320 defines the width of the rib eye muscle. As a backup, the width measurement is rejected if it falls in a column outside the center at 50% of the rib eye muscle length. 13. Measurement of the Thickness of the Subcutaneous Fat Layer for Rating purposes. Once the rib eye muscle length has been determined, this information can be used, together with the data collected in Step IV (C) (10) on the thickness of the subcutaneous fat layer for the qualification analysis. The qualification analysis varies according to meat qualification practices in various jurisdictions. In Canada, for example as illustrated in Figure 35, the line describing the rib eye muscle length 314 is divided into four equal sections, thus defining points 322, 324 and 326. From each of the points 322 , 324 and 326, a descending line 328 is drawn, perpendicular to line 314, to intersect the inner edge of subcutaneous fat layer 202 (this being the outline of the rib eye 205). The thickness of the subcutaneous fat layer 202 is determined along each line 328 through points 322, 324 and 326, by calculating the distance between the inner edge of the subcutaneous fat layer 202 (this being the outline of the eye). rib 205) and the outer edge 210 of the subcutaneous fat layer 202, perpendicular to the outer edge 210. The minimum thickness of the subcutaneous fat layer in the right quadrant, which represents the "fat degree" measurement, is also determined. standard. This information is used to determine the degree of beef yield.

V. Analysis of the Data Collected by the Provision to Obtain Images of Hot Residents and the Disposition to Obtain Images of the Rib Eye Muscle. The measurements obtained from the disposition to obtain images of cattle and to obtain images of rib eyes are useful as independent variables in predictive equations to estimate different parameters of the res. Predictive equations may include independent variables that consist of measurements taken only from the whole res (one face only), only from the rib eye, or may include measurements taken from both the whole res and the rib eye image. The parameters of the beef that are of particular interest include those that are indicative of the commercial value of the beef. In accordance with what is expressed in the Examples of the present, predictive equations have been developed to estimate the yield for the sale of beef cattle. Other parameters of the animal of interest include, without limitation, lean body mass and beef yield. Lean body mass can be defined as the total skeletal muscle mass of an animal or as the proportion of the total weight of the living animal represented by skeletal muscle, bone and associated fat as a proportion of the weight of the living animal. This value is usually expressed in terms of percentages (for example, 60% of beef yield) or in terms of weight ratio (for example, 600 g / kg of live weight).

To develop a predictive equation, the image analysis of the res and the image of the rib eye described above in a sample of the group of cattle is performed and the value of the parameter of the res in question for each res is measured. The sample population contains a sufficient number of cattle to be able to determine a statistically significant relationship or correlation between one or more selected independent variables and the parameter of the res (dependent variable) of interest. The sample population may contain only three cattle, and more preferably more than ten cattle, and even more preferably more than 100 cattle. The relationship between independent variables and dependent variables can be determined by any of a number of known statistical methods such as multiple linear regression, Artificial Neural Network examination, discrimination analysis. In a preferred embodiment, the multiple regression procedure of SAS is used (Sas Institute Inc., Cary, North Carolina). Where there are multiple independent variables, a solution using matrix algebra can be used. For example, where the independent variables are being analyzed, and the dependent variable is the yield for sale, the multiple regression model can be: y, = a + j jCy + -12X21 + i- X-jj < . + J - 5X57 + -fc-sX © + byXr, * bs¡ + + e, 9 a + S -bi and + e, where: i = the yield for the sale of the i-avo animal, j = 1, 2 9, a = the total mean, bi = the coefficient i-th of regression, i = 1, 2 ... 9, x: = the j-ava variable of prediction, j = 1, 2 .. .9, e. = random error associated with i-ava observation. The following matrices and vectors are defined: 1 X12 X22 X-s: y:? b2 x = r y = - e = y b 1 Xj n X? N X? N yn e "bn The complete series of equations is y = Xb + e with E (y) E (e) = O and var (e) = o2 1, where E denotes the estimated operator and o2 is a constant.

The regression coefficient b is estimated as b = (X'X) The prediction is The information taken from the rib eye analysis can also be used only to obtain a predictive equation for a beef parameter such as the above-described yield for sale. On the other hand, the rib eye analysis information can be used to determine the quality grade of the beef. In accordance with the above, the qualification grade of the cattle in North America as degrees such as A, AA and AAA generally depends on the percentage of intramuscular (marbling) fat of the beef, estimated taking into account the rib eye. Since the muscle / fat tissue screening and discrimination techniques of the present invention allow accurate evaluation of the intramuscular fat percentage of the rib eye, the rib eye analysis techniques of the present invention are useful for determining the degrees of quality of the cattle. The invention is further illustrated by the following non-limiting examples.

EXAMPLE 1 An apparatus and process of the present invention was examined to evaluate the accuracy of the arrangement for obtaining rib eye images in the determination of rib eye parameters. For convenience, reference is made to the embodiment of the present invention analyzed in the following examples with the designation of the inventors "CVS Lacombe". Images of the rib eye are obtained and analyzed with CVS Lacombe of 40 cattle of cattle. Then the rib eyes were tracked manually by human qualifiers and measured by means of a rating grid. As shown in Table 3, the CVS Lacombe tracing of the rib eye area was intimately related to manual tracking. Finding the square of the relationship gives the value R2.

Table 3 Correlation of the rib eye area tracked by CVS and estimates of the qualifiers Area tracked by grid area of computer-tracked area qualifiers qualifiers 0.89 0.87 EXAMPLE 2 A total of 65 cattle were sampled and ran to the bottom of the grading platform, representing 19, 19 and 22 grains of grade A, AA and AAA, respectively, to obtain more measurements (Table 5). A registered qualifier evaluated the degree of quality and marbling in AMSA (American Meat Science Association) units directly from the beef and once again from the obtained image displayed on the computer screen. Photographic images of marbling standards from the United States Department of Agriculture were analyzed with the CVS Lacombe to calibrate the CVS in order to make determinations of the degree of quality. The estimated percentage of intramuscular fat determined by the CVS for Canadian and US quality qualifiers is shown in Table 4.

Table 4 Estimates of the USDA marbling standards by CVS Lacombe Quality Grade Canada Quality Grade USA% intramuscular fat estimated by CVS Lacopfce Light AA 0.9% Short AAA 2.0% Superior Canadian Slightly abundant 9.7% The rule for assigning the grade of quality is illustrated in Table 5.

CVS Lacombe Rules for Assigning Quality Grades in Canada Grade Rules Bl 0% Marbling CVS A 0% < % marbled CVS < 0.9% AA 0.9%% marbled CVS < 2.0% AAA 2.0% < % marbled CVS < 9.7% Canadian Superior 9.7% < % marbled CVS Table 6 shows CVS marbling percentage determinations, the scores determined by the qualifier registered directly from the res and determined with the CVS computer screen.

Table 6 Marbling percentage by quality grade A AA AAA Number of cattle 19 19 22% marbling 2.73 ± 1.20 4.90 ± 1.12 6.33 ± 1.70 computer AMSA 427.50 ± 53.23 521.25 ± 33.64 563.33 ± 71.39 The correlation of the percentage of marbling estimated by CVS with that estimated by the AMSA qualifier in real time was 0.81 (Table 7). The correlation improved to 0.92 when the qualifier evaluated the cattle once again using the computer screen.

Table 7 Correlations between marble percentage estimated by CVS and the qualifier score with AMSA AMSA 0.81 AMSA with screen 0.92 Based on an = 65 EXAMPLE 3 Lacombe CVS was analyzed to determine if regular qualification determinations could be made on rib eye of the same res under different lighting conditions. 44 beef steaks with various marbling levels were purchased in supermarkets. Each churrasco was analyzed 5 times in the Lacombe Meat Research Center, Lacombe, Alberta, Canada, in the cutting room, in a meat cooler, and in a display. These areas represent the lighting conditions of bright fluorescent light, dim fluorescent light and incandescent light, respectively. As illustrated in Table 8, repeated calculations for the rib eye area and marbling percentage ranged from 0.86 to 0.96, indicating that the disposition for taking rib eye images can provide consistent results in different lighting conditions. . The repeated estimates for the degree of fat were lower since the churrascos were kept at room temperature and therefore were very soft and changed shape when moved from one place to another.

Table 8 Repeated form of CVS Lacombe measurements in various lighting conditions Refrigerator Room Measurements Location CVS Meat Cutting Showcase Grade of fat 0.66 0.84 0.44 Area of the rib eye 0.96 0.96 0.86% of marbling 0.96 0.96 0.96 EXAMPLE 4 A second study was carried out to evaluate the repeated form of the measurements taken by CVS Lacombe. On the first day of the study, a human qualifier took standard measurements of the degree of fat, rib eye area and marbling of the rib eyes of 14 cattle using a measurement grid. Then, the cattle were measured twice within a minute by the CVS. The cattle were pushed by hand one at a time to measure them with the rib eye camera. On the second day of the study, essentially the same procedures were carried out on 21 more cattle.

Table 9 illustrates the correlation between the two successive measurements taken by the Lacombe CVS in each res. In general, it can be seen that the average correlation is very high, close to 97%.

Table 9 Correlation of two repeated CVS measurements Correlation between Study 1 (n = 14) Study 2 (n = 21) First and second 0. 98 0. 97 measurements of fat grade by CVS First and second 0. 98 0. 98 measurements of rib eye area by CVS First and second 0. 97 0. 94 marbling measurements by CVS E JEMP LO 5 Another study was carried out repeatedly in which the degree of fat, the area of the rib eye, and the% of marbling for the rib eyes of 166 cattle with the CVS Lacombe were measured. The measurements were repeated between 1 and 11 times, with an average of 4 repeated measurements. The cattle were also qualified by human qualifiers. A total of 166 heads of cattle were sampled, representing cattle of both sexes by 3 groups of weight for three fat levels. The data collection was completed in a 3-day cycle. The first day of the data cycle, images of the intact cattle were obtained, before cooling. Low quality images were noted so those cattle were not selected to be butchered the next day. On the second day, the cattle were selected according to a pre-established sampling scheme. Rib eye images of chilled cattle were obtained and evaluations of human graders were recorded. On the third day, the selected cattle were dismembered (sectioned) by a team of 7 cutters who sectioned 20 cattle from the left half to a yield for the sale of 1/4"cut in 3 hours. weights of the various cuts.

The cut data of the cattle were collected and edited. The yield of meat for sale was defined as the sum of the weights of the cuts plus the weights of the cuts (50, 75 and 85% lean) expressed in terms of the weight percentage of the face. The analysis data of the images was merged.

Table 10 Repeated form of CVS measurements CVS measurements Repeatedly Grade of fat 0.99 Area of the rib eye 0.95% of Marbling 0.95 EXAMPLE 6 The correlations between CVS Lacombe measurements and human qualifiers for fat thickness and rib eye area were determined for the data collected in examples 2 and 3. As illustrated in table 11, the correlations between measurements of CVS Lacombe and the qualifiers ranged between 0.89 and 0.99 in 3 separate studies.

Table 11 Correlation between measurements of CVS Lacombe and qualifiers Between CVS and Study 1 (n = 14) Study 2 (n = 21) Study 3 (n = 166) qualifier for Degree of fat 0.89 0.96 0.95 Area of eye 0.91 0.99 0.95 rib EXAMPLE 7 The degree of correlation between the marbling determinations of the human qualifiers (grade of quality) and those of CVS Lacombe was determined. The estimation of the percentage of marbling by CVS was calibrated with the supervisor of regional rating. The Food and Agricultural Protection and Inspection Branch of Canada. The threshold for deciding whether a pixel is muscle or fat was tentatively adjusted in an effort to mimic the marbling levels seen by the qualifier. Once the adjustment was completed, a test was carried out to determine the correspondence between the marble grade assigned by CVS and by the qualification supervisor. Out of a total of 62 animals analyzed, mostly AA and AAA, only 4 did not coincide. This suggests that the Lacombe CVS can be calibrated to match an experienced qualifier at a reasonable level of accuracy.

EXAMPLE 8 Estimated yield equations for sale were determined using the data obtained in Example 5. The main cuts of the rump were analyzed., loin, skirt, ribs, needle, paddle, matambre and chest in the images of the cattle as described in step III (B) (4) (c) of the Detailed Description of the Preferred Embodiment. As described in step 111 (4) (d) of the Detailed Description of the Preferred Embodiment and illustrated in Figure 17, the images of the cattle were divided into six linear regions described as linear regions of the hindquarters. , lower quadril, upper quadra, lower dorsal part, middle dorsal part and scapula. Each linear region was divided into 10 equal divisions, demarcated by lines drawn in a direction transverse to the longitudinal axis of the image of the res. In each linear region, the reference lines 1-9 were assigned to the transverse lines, starting from the rear end of the image of the res. The distance was determined from the points at which each transversal line intercepted the ventral and dorsal regions of the image contour of the res to the midline (line b3-b9 of Figure 17). Angular and bending measurements were also determined as described in steps III (B) (e) (e) and (f). In total, more than 400 measurements of cattle and rib eye were made and analyzed for the correlation with the yield for the sale of the cattle. The stepped regression was applied to the data to arrive at the best models to predict the yield for the sale of the measurements of the intact res, the measurements of the rib eye and the combination of these two series of measurements. The R2 (multiple correlation squared), which indicates the extent to which the data correspond to the model, and the residual standard deviation (square root of the square of the residual mean), which indicates the error of estimating the cutoff value of the CVS measurements , are summarized in Table 12. The accuracy of CVS Lacombe (RSD = 1.03%) to estimate the yield for sale was higher than the Australian system evaluated by Jones et al. (1993, RSD = 1.27%) and Richmond et al. (1995). , RSD = 1.65%). The most accurate results were obtained when measurements of intact animal images, as well as rib eye images were included as independent variables in the predictive equation. The use of 20 measurements of the res image only proved to be the next most accurate approach, followed by 5 CVS measurements of only the rib eye. It should be noted that all the equations of performance for sale determined with the use of CVS measurements proved to be more accurate than equations based on measurements made by a human qualifier.

Table 12 Accuracy of the use of computer observation measurements to estimate the% yield for sale Measurements included in the model 'R2 RSD * (%) " Based on the measurements of the qualifier 0.57 1.55 at the place of qualification Based on the measurements of 20 cattle 0.67 1.46 only Based on measurements of 5 eyes 0.61 1.49 rib at the place of qualification Based on a combination of 20 cattle and 0.84 1.03 rib eyes * Multiple squared correlation ** Standard deviation RSD residual It was determined that the equation of estimation of the yield for sale based on measurements of the rib eye is only: % of yield for sale = 76.238706 + 0.010197 TASK -. 0.349852 TPGRASA - 0.427767 AGRASA + 0.25422 MGRASA - 0.193396 GGRASA where TASK = total rib eye area; TPS ASA = total percentage of intramuscular fat in the rib eye; A- = average measurements of subcutaneous fat thickness of the rib eye at the upper, middle and lower points as illustrated in Figure 35, reference numbers 322, 324 and 326; MGRASA = thickness of subcutaneous fat at midpoint as illustrated in Figure 35, in reference number 324; and GGRASA = thickness of subcutaneous fat at the point of measurement of standard fat grade, which is the minimum thickness of subcutaneous fat in the median right end quadrant as illustrated in Figure 35. It was determined that the estimated yield equation for the sale based on the measurements of the res only are: % of return for sale = 106.085803 176.062150 Pcost + 262.044441 Pfal - 340.168645 Pmat -270.060083 A4 + 295.14395 A6 + 32.222714 After3 + 24.016820 After22 + 7.035919 Lcuad34 - 51.034738 Lcuad35 + 14.94428 M-, or33 - 27.284773 M - Dorsl2 - 89.700540 Pal2 + 73.786202 pal6 - 15.522698 pal23 + 35.014305 AR? t where Peo: = ratio of the area of the main cut of the rib to area of the total image of the res; P fal. = ratio of the area of the main cut of the skirt to the area of the total image of the res; Pma t, = ratio of the area of the main cut of the matambre to area of the total image of the beef; A4 = area of the main cut of the skirt; A6 = area of the main cut of the matambre; After 3 = ventral distance to the midline for interval 1 of the linear region of the hind paw; After 22 = ventral distance to the midline for interval 6 of the linear region of the hind paw; L u = Width of the beef for interval 2 of the linear region of the lower quadril; L u a d - > • = dorsal distance for interval 9 in the linear region of the lower quadril; , u a 3 5 = Ventral distance for interval 9 in the linear region of the lower quadril; Mt ra s 3 = Ventral distance for interval 1 in the linear region of the middle dorsal part; Mt ra s l 2 = Width of the beef for interval 3 in the linear region of the middle dorsal part; Pal; Dorsal distance for interval 1 in the linear region of the pallet; Pal 23 Distance dorsal for interval 6 in the linear region of the pallet; AR lb Average width of the res in the linear region of the lower dorsal part.

It was determined that the estimative equation using the measurements of the whole res as well as the rib eye as independent variables was: % yield for sale = 79.448902 + 0.0018258 Task ~ 0.362784 Agra3a + 0.267664 Mgrasa -0.185617 Ggrasa - 20.087953 Pcost + 214.213295 Pfal -243.441006 Pmat - 224.112984 A4 + 171.424092 A6 + 13.781479 Tras3 + 14.152217 Tras22 + 2.862327 Lcuad8 + 20.933690 L uad34 - 25.216945 Lcuad35 + 2.567813 Mdors3 -1.173930 M - Dorsl-2 - 59.559750 Pal2 - 45.429554 pal6 - 11.739671 pal23 + 41.817415 ARib in which the independent variables are those previously exposed. It will be appreciated that in the predictive equations exposed for the yield for sale none of the measurements of oblique angle or curvature measurements proved to be closely associated with the yield for the sale of cattle in this case. In addition, it is somewhat surprising that the area of the main cut of the matambre and the ratio of the area of the main cut of the matambre to area of the total contour of the cattle proved to be important variables, since the main cut of the matambre is of less economic importance than Other main cuts of beef cattle. It will also be noted that the predictive equations of the yield for sale include at least two distances from each of the ventral and dorsal regions of the contour of the image of the beef to the midline, at least two widths of the beef, and the areas of the main cuts and the proportions of the areas of the main cuts to the total image area of the res. All publications mentioned in this specification are indicative of the level of training of those skilled in the art to which this invention relates. All publications are incorporated herein by reference to the same extent as if it had been indicated that each of them is specifically and individually incorporated herein by reference.

Although the foregoing invention has been described in some degree of detail by way of illustration and example in order to clarify and facilitate understanding, it is clear that certain changes and modifications may be practiced within the scope of the appended claims. It is noted that in relation to this date, the best method known to the applicant to carry out the aforementioned invention, is that which is clear from the present description of the invention.

Claims (27)

CLAIMS Having described the invention as above, property is claimed as contained in the following:

1. A process to determine a degree of quality of an animal res, characterized in that it comprises the steps of: (a) obtaining an image that includes the rib eye of the animal, image that is composed of a series of pixels that provide color data representative of the color information in the corresponding part of the image; (b) discriminating the pixels representing muscle tissue from the pixels representing adipose tissue on the basis of a pixel color characteristic threshold; (c) identify a conglomerate of muscle tissue pixels within the image representing the rib eye and trace the contour of the rib eye muscle to exclude the outer sections of the image representing the muscle tissue in contact, but not part of the rib eye; (d) determining the proportion of pixels within the contour of the rib eye that represents fat with respect to the total number of pixels within the rib eye contour to obtain a value of the intramuscular fat percentage of the rib eye; (e) repeating steps (b) to (d) for a plurality of reference images of rib eyes of cattle of a predetermined grade of quality to establish a relationship between the percentage of intramuscular fat in the rib eye and the degree quality of the beef; and (f) solving the ratio determined in step (e) for the value of the intramuscular fat percentage of the rib eye determined in step (d) to determine the quality grade of the beef.

2. The process according to claim 1, characterized in that it comprises the additional step of discriminating the pixels representing muscle tissue from the pixels that do not represent muscle tissue or adipose tissue on the basis of a color saturation level threshold.

3. The process according to claim 1, characterized in that in step (e), the reference images of the rib eye area are obtained from normal photographs showing the appearance of the rib eye eyes of cattle of known quality grades.

4. A process for determining the parameters of a rib eye of a beef, characterized in that it comprises the steps of: (a) obtaining an image that includes the rib eye of the beef, image that is composed of a series of pixels that provide data color representative of the color information in the corresponding part of the image; (b) discriminating the pixels representing muscle tissue from the pixels representing adipose tissue on the basis of a pixel color characteristic threshold; (c) identifying a selected conglomerate of contiguous pixels of muscle tissue within the image, selected conglomerate that includes the rib eye and tracking the contour of the conglomerate of muscle tissue pixels that includes the rib eye; (d) identify and separate the external sections of the image that represent muscle tissue that is in contact, but not part of the rib eye, including the step of identifying the external sections of the image: (i) identify external contour turns of the rib eye tracked in step (c); and (ii) rejecting an external rotation as a potential site to begin a cut to cleave the outer section of the image if a first line cutting the rotation is not substantially parallel to a second tangent line of a first generally elliptical figure surrounding the outline of the rib eye at a point of the collinear ellipse with the center of the elliptic figure and the vertex of the rotation.

5. The process according to claim 4, characterized in that the identification and splitting step (d) further comprises: (iii) plotting a plurality of search lines originating at the vertices of the external rotations that have not been rejected in step (ii) search lines that radiate inwardly through a range of at least about 20 ° on each side of a line that cuts off rotation; record the number of pixels representing muscle tissue along each search line until detecting a row of at least 4 consecutive pixels representing fat; and selecting the search line with the least number of pixels representing muscle tissue as a potential path of a cut to section an external section of the image.

6. - The process according to claim 5, characterized in that step (d) (iii) also comprises rejecting the search line as a potential path of a cut that would intercept a second generally elliptical figure completely contained within the contour of the eye muscle rib

7. The process according to claims 4, 5 or 6, characterized in that it comprises the additional step of: (e) identifying and splitting the rest of the external sections of the image that have not been identified and excised in step (d) of the following way: (i) identify the remaining external rotations in the contour of the rib eye image; (ii) plotting a plurality of search lines originating at the vertices of each of the remaining external rotations, search lines that radiate inwardly through a range of at least about 20 ° on each side of a line that intersects the rotation; (iii) recording the number of pixels along each search line until the length of the search line is determined until the search line is cut one more time with the contour of the rib eye image; (iv) select the search line with the shortest length as the potential path of a cut to section an external section of the image.

8. The process according to claim 7, characterized in that the step of identifying the remaining external rotation (e) (i) comprises tracing the path of a target moving along the contour of the rib eye image and measuring the radial movement of a line that has extreme points on the target and the center of the contour of the image of the rib eye, detecting a rotation towards the outside at the point of the contour occupied by the target when the movement of the line between the target and the center of the rib eye line reverses the direction of rotation.

9. The process according to claim 8, characterized in that the step of identifying and splitting (e) further comprises: (v) rejecting the potential path of a slice to split an external section of the image if the path length is greater than one predetermined ratio of the length of the longitudinal maximum axis of the external section of the image to be excised.

10. The process according to claim 9, characterized in that the step of identifying and splitting (e) further comprises: (vi) rejecting the potential path of a cut to excise an external section of the image if at least one of the following criteria: (I) the external section of the image to be excised by the cut has a surface to a predetermined area; (II) the cut results in the contour of the rib eye image having an area below the predetermined area; or (III) if less than a predetermined percentage of the area of the outer section of the image to be excised by the cut falls outside the first generally elliptical figure.

11. The process according to any of claims 4 to 10, characterized in that it comprises the following additional step: (f) once the possible external sections of the image have been excised, determine the proportion of pixels within the contour of the rib eye that represent fat , with respect to the total number of pixels within the rib eye contour to obtain a value of the percentage of intramuscular fat in the rib eye.

12. A process to determine the qualification parameters of a res, characterized in that it comprises the steps of: (a) obtaining an image of a view of the res, image that is composed of a series of pixels that provide representative data of the information in the corresponding part of the image; (b) tracing the profile of the image to produce an image contour of the res; (c) locating a plurality of first reference points on the contour of the beef image, first reference points representing anatomical features of the beef, anatomical features that are identified as prominences or depressions in the beef image contour; (d) locating at least one second reference point on or within the contour of the image of the res, these second reference points being located at predetermined positions with respect to the first reference points; (e) dividing the image of the beef into a plurality of sections, the boundaries of each section being determined as a function of the position of the first and second reference points, and determining the area of each section; (f) present a predictive equation of the qualification parameters in which the qualification parameter is included as a dependent variable, and at least one area of a given section in step (e) is included as an independent variable; and (g) solving the predictive equation of the rating parameter to produce a value for the res rating parameter.

13. The process according to claim 12, characterized in that it further comprises the step of: (h) determining the shortest distance between at least one reference point in a dorsal or ventral region of the contour of the beef image and a line mean of the image of the restrained substantially parallel to the longitudinal axis of the image of the res, middle line that divides the image of the res in ventral and dorsal portions; and in which at least one of the distances between the reference points, between the points of a dorsal or ventral region of the contour of the image of the animal and a middle line of the image of the animal which, according to the determination of the step (h) is included as an independent variable in the predictive equation of the qualification parameters provided in step (f).

14. The process according to claim 13, characterized in that it also comprises the step of: (i) determining the width of the contour of the image of the res in at least one reference point, perpendicular to the average line established in step ( h); and in which at least one of the widths of the image contour of the res determined in step (i) is included as an independent variable in the predictive equation of the rating parameters provided in step (f).

15. The process according to claim 14, characterized in that step (e) further comprises determining the proportion of the area of at least one of the sections of the image of the total area covered by the contour of the image of the res , and in which at least one of the proportions is included as an independent variable in the predictive equation of the qualification parameters provided in step (f).

16. The process according to claim 15, characterized in that in step (e), the sections include sections that have the limits of the main standard cuts used in a cattle grading system, main cuts that are selected from the group consisting of in main courts of cuadril, loin, skirt, rib, cut from the needle to the neck, paddle, matambre and chest.

17. The process according to claim 16, characterized in that it also comprises the step of: (j) measuring the value of at least one standard qualification criterion for the beef score parameter, standard rating criteria selected from the group that consists of the area of the rib eye, the percentage of intramuscular fat in the rib eye area, the thickness of a layer of subcutaneous fat over the rib eye in predetermined positions, the average thickness of the subcutaneous fat layer over the rib eye, and the width of the subcutaneous fat layer at a standard grade fat measurement site; and in which at least one of the standard qualification criteria for the rib eye of the res is included as an independent variable in the predictive equation of the rating parameter provided in step (f). •

18. The process according to claim 16, characterized in that the qualification parameter of the res to be determined is the yield for the sale of the beef, and in which: in step (e), the sections of the res include the main cuts of the rib, the 10 skirt and the matambre, and determine the area relationships of each of the main cuts of the rib, the skirt and the matambre to the total area encompassed by the contour of the image. the res; in step (h), distances are determined 15 from at least 3 reference points in each of the dorsal and ventral regions of the contour of the beef image to a midline of the contour of the beef image; in step (i), the width of the contour of the image of the res in at least two reference points is determined; and in step (f), the predictive equation of the qualification parameter is a predictive equation of the yield for sale in which the following independent variables are included: the areas of the main cuts of the skirt and the matambre; the proportions of the area of the main cut of the rib, the main cut of the skirt and the main cut of the matambre to the total area comprised by the contour of the image of the res; the distances between at least 3 reference points in each of the dorsal and ventral regions of the outline of the beef image and the midline of the contour of the beef image; and the width of the contour of the image of the res in at least two reference points.

19. An apparatus for determining a grade of quality of an animal res, characterized in that it comprises: (a) means of obtaining images to obtain an image that includes the rib eye of the animal; (b) computing and storage means for: (i) storing the image in the form of a series of pixels that provide data representative of the color information in the corresponding part of the image; (ii) discriminating the pixels representing muscle tissue from the pixels representing adipose tissue based on a threshold of pixel color characteristics; (iii) identify a conglomerate of muscle tissue pixels within the image representing the rib eye and trace the contour of the rib eye muscle to exclude the outer sections of the image representing the muscle tissue in contact, but not part of the rib eye; (iv) determining the proportion of pixels within the contour of the rib eye that represent fat with respect to the total number of pixels within the rib eye contour to obtain a value of the intramuscular fat percentage of the rib eye; (v) applying means (i) to (iv) to a plurality of reference images of rib eyes of cattle of a predetermined grade of quality to establish a ratio between the percentage of intramuscular fat in the rib eye and the degree quality of the beef; (vi) solving the relationship determined by the instrument (v) for the value of the percentage of intramuscular fat of the rib eye determined by the instrument (iv) to determine the grade of quality of the beef; and (c) means for producing a result of beef quality grade.

20. The apparatus according to claim 19, characterized in that it also comprises counting and storage instruments for: (viii) discriminating the pixels representing muscle tissue from the pixels that do not represent muscle tissue or adipose tissue based on the saturation level threshold of color of the pixels.

21. An apparatus for determining the parameters of the rib eye of a beef, characterized in that it comprises: (a) means for obtaining images to obtain an image that includes the rib eye of the beef; (b) computing and storage means for: (i) storing the image in the form of a series of pixels that provide data representative of the color information in the corresponding part of the image; (ii) discriminating the pixels representing muscle tissue from the pixels representing adipose tissue based on a threshold of pixel color characteristics; (iii) identify a conglomerate of contiguous pixels of muscle tissue within the image, selected conglomerate that includes the rib eye and trace the contour of the conglomerate of muscle tissue pixels that include the rib eye; (iv) identify and excise the external sections of the image that represent the muscle tissue that is in contact, but is not part of the rib eye, including the means to identify the external sections of the image: (a) means to identify rotations outward in the outline of the rib eye tracked by the middle (iii); and (b) means for rejecting an outward rotation as a potential site for starting a cut to cleave an external section of the image if a first line cutting the rotation is not substantially parallel to a second tangent line of a first figure generally elliptical that surrounds the outline of the rib eye at a point • of the collinear elliptical figure with the center of the elliptical figure and the vertex of the rotation; and (c) means for producing a result of the rib eye parameters.

22. The apparatus according to claim 21, further characterized in that .J.10 comprises computation and storage elements for: (iv) (c) tracing a plurality of search lines originating at the vertices of the external rotations that have not been rejected by the element (b), search lines that radiate internally to 15 through a range of at least about 20 ° on each side of a line that cuts the rotation; f record the number of pixels representing muscle tissue along each search line until a row of at least 4 pixels is detected 20 consecutive ones that represent fat; and selecting the search line with the least number of pixels representing muscle tissue as a potential path of a cut to section an external section of the image.

23. An apparatus for determining rating parameters of a beef, characterized in that it comprises: (a) means for obtaining images to obtain an image of a view of the beef; (b) computation and storage means for: (i) storing the image in the form of a series of pixels that provide data representative of the information in the corresponding part of the image; (ii) trace the outline of the image to produce a contour of the image of the res; (iii) locating a plurality of first reference points on the contour of the beef image, first reference points representing anatomical features of the beef, anatomical features that are identified as prominences or depressions in the beef image outline;; (iv) locating at least one second reference point on or within the outline of the beef image, these second reference points being located at predetermined positions with respect to the first reference points; (v) dividing the image of the beef into a plurality of sections, the boundaries of each section being determined as a function of the position of the first and second reference points, and determining the area of each section; (vi) produce a predictive equation of the qualification parameters in which the qualification parameter is included as a dependent variable, and at least one area of a section determined in step (v) is included as an independent variable; and (vii) solving the predictive equation of the rating parameter to produce a value for the res rating parameter; and (c) means for producing a result of the rating parameter of the res.

24. A process to determine the qualification parameters of a res, characterized in that it comprises the steps of: (a) obtaining an image that includes the rib eye of the res, image that is composed of a series of pixels that provide representative color data of the color information in the corresponding part of the image; (b) discriminating the pixels representing muscle tissue from the pixels representing adipose tissue on the basis of a pixel color characteristic threshold; (c) identify a conglomerate of muscle tissue pixels within the image representing the rib eye and trace the contour of the rib eye muscle to exclude the outer sections of the image representing the muscle tissue in contact, but not part of the rib eye; (d) determining the proportion of pixels within the contour of the rib eye that represents fat with respect to the total number of pixels within the rib eye contour to obtain a value of the intramuscular fat percentage of the rib eye; (e) measuring the value of at least one standard rating criterion for the rib eye image, standard rating criteria that are selected from the group consisting of the rib eye area, the percentage of intramuscular fat in the area of the rib eye, the thickness of a layer of subcutaneous fat over the rib eye at predetermined positions, the average thickness of the subcutaneous fat layer over the rib eye, and the width of the subcutaneous fat layer at its point narrower; (f) producing a predictive equation of the qualification parameters in which the qualification parameter is included as a dependent variable, and at least one area of a given section in step (e) is included as an independent variable; and (g) solving the predictive equation of the rating parameter to produce a value for the res rating parameter.

25. The process according to claim 24, characterized in that in step (c), excluding the outer sections of the image, includes: (i) identifying the outward rotations of the contour of the rib eye tracked in the passage (c) ); and (ii) reject an outward rotation as a potential site to begin a cut to cleave an external section of the image if a first line cutting the rotation is not substantially parallel to a second tangent line of a first generally elliptical surrounding figure. the outline of the rib eye at a point of the collinear ellipse with the center of the elliptic figure and the vertex of the rotation.

26. An apparatus for determining rating parameters of a res, characterized in that it comprises: (a) means of obtaining images to obtain an image that includes the rib eye of -0 the res; (b) computation and storage means for: (i) storing the image in the form of a series of pixels that provide data representative of the color information in the corresponding part of the image; (ii) discriminating the pixels representing muscle tissue from the pixels representing adipose tissue based on a threshold of pixel color characteristics; 0 (iii) identify a conglomerate of muscle tissue pixels within the image representing the rib eye and trace the contour of the rib eye muscle to exclude the outer sections of the image representing the muscle tissue in contact, but that it is not part of the rib eye; (iv) determining the proportion of pixels within the contour of the rib eye that represent fat with respect to the total number of pixels within the rib eye contour to obtain a value of the intramuscular fat percentage of the rib eye; (v) measure the value of at least one standard grading criterion for the beef score parameter, standard grading criteria selected from the group consisting of the rib eye area, the percentage of intramuscular fat in the area of the rib eye, the thickness of a layer of subcutaneous fat over the rib eye in predetermined positions, the average thickness of the subcutaneous fat layer over the rib eye, and the width of the subcutaneous fat layer at its most narrow; (vi) produce a predictive equation of the qualification parameters in which the qualification parameter is included as a dependent variable, and at least one qualification criterion of the rib eye is included as an independent variable; (vii) solve the predictive equation of the qualification parameter to produce a value for the res qualification parameter; and (c) means for producing a result of the rating parameter of the res.

27. The apparatus according to claim 26, characterized in that the tracking means (iii) includes: (a) means for identifying outward rotations in the contour of the rib eye tracked in step (c); and (b) means for rejecting an outward rotation as a potential site to begin a cut to cleave an external section of the image if a first line cutting the rotation is not substantially parallel to a second tangent line of a first generally elliptical figure. which surrounds the outline of the rib eye at a point of the elliptical figure which is collinear with the center of the elliptical figure and the vertex of the rotation. SUMMARY OF THE INVENTION In a process and apparatus for determining the qualification parameters of a beef, the contour of a beef image is traced and reference points are identified that represent the anatomical features of the beef. Then, second reference points located at predetermined positions are identified with respect to the first reference points. The image of the beef is divided into a plurality of sections, the limits of each section are determined according to the position of the first and second reference points, and the area of each section is determined. A predictive equation of the qualification parameter is determined in which the qualification parameter is included as a dependent variable, and at least one sectional area of the res image is included as an independent variable. Solving the predictive equation, we obtain a value for the qualification parameter of the res. Other measurements that can be obtained from the image of the beef and used as independent variables in the predictive equations include the distances between the dorsal and ventral regions of the contour of the beef image and a median line of the beef, the widths of the res, the angular measurements between reference points, and the curvature measurements of the contour of the res image. Improved rib eye tracking techniques allow accurate measurement of rib eye parameters. Measured rib eye parameters can be used to determine a grade of beef quality or as independent variables in the predictive equation of the grading parameter, alone or in conjunction with measurements taken from the beef image.