Image analysis – Image sensing – Hand-held
Reexamination Certificate
2001-10-15
2004-06-15
Johns, Andrew W. (Department: 2621)
Image analysis
Image sensing
Hand-held
C382S110000
Reexamination Certificate
active
06751364
ABSTRACT:
BACKGROUND OF INVENTION
(1) Field of Invention
The invention relates to automated grading of meat and predicting meat yield and quality of meat from an animal carcass and, more particularly, to capturing images of meat portions of an animal carcass and processing information in the image for grading of meat and predicting meat yield.
(2) Background Art
Grading of animal carcasses for the purpose of predicting meat yield and quality is an important aspect of the meat processing industry. Meat grading has historically been performed by a human grader. To perform the meat grading process the human grader will typically examine key physical aspects of the carcass. The type of grading being performed determines what physical aspects of the carcass need to be examined by the grader. The two main types of meat grading for a carcass are quality grade and yield grade. The quality grade describes the meat's palatability or tenderness. Whereas, yield grade describes the proportion of lean boneless meat that a given carcass will yield.
In the meat industry, it is common for the human grader to examine various physical aspects of a cross section of the Longissimus dorsi (commonly referred to in beef as the ‘ribeye’ and in pork as the ‘loin eye’) for both yield grading and quality grading. In a typical beef processing facility after the animal has been slaughtered, head removed and skinned, the carcass is further disassembled by splitting the carcass in half along the midline. The carcass is then ‘ribbed’ or severed between the twelfth 12
th
and 13
th
ribs thereby exposing for examination and grading a cross section of meat or a cutting surface of meat, which specifically includes the ribeye and associated subcutaneous fat. For quality grade, the human grader will typically examine the ‘marbling’ (intramuscular fat). For yield grade, the human grader will typically examine the area of the ribeye cross section and the thickness of subcutaneous fat adjacent to the ribeye at various points around its area and adjust the fat thickness based on fatness of the entire carcass. For yield grade, the human grader also utilizes parameters such as hot carcass weight and percentage kidney, pelvic and heart fat for determining yield grade.
Grading by a human grader is typically based upon the human grader's perception of the appearance of the ribeye. Photographs can be utilized as standards for determining grade. Photographs are used for training for quality grade, but are not typically used for grading. This process clearly introduces a substantial amount of subjectivity into the meat processing industry. The Human Grader's subjectivity is problematic because this grading determines the valuation of animal carcasses and therefore clearly effects the financial bottom line.
Yield grade is typically denoted by a numerical value from 1 to 5 based upon the yield from the carcass of boneless, closely trimmed (approximately 0.25 in.), retail cuts from the round, loin, rib and chuck. An accurate yield grade for these four wholesale cuts of meat is extremely important to an accurate valuation of the carcass, thus reducing the amount of subjectivity is desirable. These four wholesale cuts make up approximately 75% of the weight, and about 90% of the carcass value. Regression equations for carcass grading have been developed from actual carcass data using factors such as fat thickness at the twelfth ribeye cross section, ribeye area and carcass weight. However, the regression equations are not practical for a human grader to utilize during actual everyday grading of a carcass in a production facility. Therefore, working formulas have been developed which make certain adjustments to the yield grade based on the same type of factors utilized by the regression equations. However, this process still results in a substantial amount of subjectivity.
In order to reduce operator subjectivity, automated instrumental grading systems have been developed. For example, various type of image analysis grading systems have been developed, which capture and analyze digital images of portions of a carcass. The image analysis systems typically examine parameters similar to or identical to the type of parameters examined by a human grader. Typically the image analysis systems try to determine and distinguish portions that are lean and portions that are fat and their respective areas. To distinguish meat portions (i.e. lean or fat; ribeye muscle or non-ribeye muscle surrounding ribeye) the image analysis system will typically utilize parameters such as color and contrast.
It is typical for the image analysis to be performed on the 12
th
ribeye cross section. However, regardless of the section of meat that is being analyzed, there are various problems in utilizing image analysis to characterize the features of the meat. For example, the muscle or the lean area of interest can be surrounded by other lean areas with minimal fat separation, which is typically true of a ribeye cross section. Therefore, it is often difficult for the image analysis system to distinguish between the muscle of interest and the adjacent muscle because the dimensions and shape of a given muscle type may vary considerably from carcass to carcass. Another example is that a muscle of a given carcass may have large areas of intramuscular fat, whereas that same muscle type for another carcass may not have the large area of intramuscular fat. This is problematic because it is difficult due to the intramuscular fat to determine where the desired muscle ends and the adjacent muscle begins. Dense marbling can also make it difficult to determine the border or the cross section area of the muscle of interest. Yet another example is distinguishing color transitions from fat to lean. Color distinction is critical particularly with dense marbling and large areas of intramuscular fat because digital analysis algorithms often look for continuous adjacent pixels of the same color to determine if a red or lean region of the image is within the area of the desired muscle. Due to the above problems many image analysis systems have difficulty identifying the correct area of the desired muscle and then appropriately analyzing the image.
Image analysis of the ribeye poses unique problems particularly in a production meat processing environment where the ribbed carcass halves are graded for quality and yield. In a typical production meat processing facility, particularly beef processing, the halved and ribbed beef carcass travels through the grading area suspended from a conveyor hook by the achilles tendon. The ribbed section of the carcass partially exposes the 12
th
rib cross section. The cross section is not fully exposed for ease of viewing because the ribbing incision is minimized such that the carcass stays intact. If the ribbing incision is too deep the head portion of the carcass will separate from the hind portion due to weight and gravity. Therefore, due to the minimized incision, it is sometimes difficult even for the human grader to get a clear view of the cross section for grading purposes without physically manipulating the carcass to obtain a better view. It is even more difficult to insert a camera in the incision to capture a good image consistently that has adequate lighting, minimal shading, and with minimal angular distortions of the image. Obtaining a good and consistent image must be achieved prior to even addressing the problems of image analysis identified above. However, obtaining a high quality image is difficult and most systems are inadequate, particularly with the inconsistent and non-uniform lighting found in most facilities.
BRIEF SUMMARY OF INVENTION
The invention is an image analysis system and method for grading of meat, predicting quality of meat and/or predicting meat yield of an animal carcass. One embodiment of the invention is particularly designed to capture an image of the 12
th
rib cross section of the carcass side and perform image analysis of the ribeye for grading purposes. The image capturing camera portion of the syst
Eger Horst
Haagensen Annett
Haagensen Peter
Koohmaraie Mohammed
Shackelford Steven D.
Blackwell Sanders Peper Martin LLP
Haagensen Annett
Johns Andrew W.
Stallion Mark E.
Tyson Fresh Meats, Inc.
LandOfFree
Image analysis systems for grading of meat, predicting... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Image analysis systems for grading of meat, predicting..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image analysis systems for grading of meat, predicting... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3342316