System and method for analyzing properties of meat using multispectral imaging
A system and method for obtaining multispectral images of fresh meat at predetermined wavelength bands at a first time, subjecting the images to analysis in an image analysis system comprising a computer programmed to perform such analysis, and outputting a forecast of meat tenderness at a later point in time. Predetermined key wavelength bands are precorrelated with a high degree of prediction of meat tenderness and/or other properties of meat and are used in the multispectral system and method. A system and method for determining the key wavelengths is also disclosed. The multispectral imaging system and method is suitable for use in an industrial setting, such as a meat processing plant. The system and method is useful in a method for determining quality and yield grades at or near the time of imaging in lieu of visual inspection with the unaided human eye, increasing efficiency and objectivity.
1. A system and method for analyzing one or more properties of meat, comprising an image acquisition system and an image analysis system, which image acquisition system employs multispectral imaging of said meat at predetermined key wavelength bands and communicates said images to said image analysis system which extracts data from said image, processes it according to programs residing therein, and produces an output which comprises a conclusion about said one or more of said properties, wherein said image is acquired at a time (t 0 ) and said output is a forecast of the tenderness properties of said meat at a time (t 1 ) which is later in time than (t 0 ).
2. The system and method of claim 1 , wherein said time (t 1 ) is the time when said meat will be available for preparation by a consumer.
3. The system and method of claim 1 , wherein said image is acquired at a time (t 0 ) and said output is also used to make a prediction of the properties of said meat at said time (t 0 ).
4. The system and method of claim 1 , wherein said number of predetermined wavelengths is from one to about twenty-five wavelengths.
5. The system and method of claim 4 , wherein said number of predetermined wavelengths is from two to twelve wavelengths.
6. The system and method of claim 5 , wherein said number of said predetermined wavelengths is four to eight wavelengths.
7. The system and method of claim 1 , wherein said predetermined wavelengths are in the visible/near infrared (VNIR) region (400-1000 nm).
8. The system and method of claim 7 , wherein said meat is beef and said wavelengths are selected from the group consisting of 455, 480, 520, 545, 568, 590, 604, 636, 722, 750, 814, and 850 nm and combinations thereof.
9. The system and method of claim 8 , wherein each of said wavelengths can vary by ±20 nm.
10. The system and method of claim 1 wherein said predetermined wavelengths are in the NIR region (900-2500 nm).
11. The system and method of claim 10 wherein said meat is beef and, wherein said wavelengths are selected from the group consisting of 1057, 1074, 1091, 1115, 1142, 1176, 1219, 1280, 1365, 1395, 1408, and 1462 nm and combinations thereof.
12. The system and method of claim 10 wherein each of said wavelengths can vary by ±20 nm.
13. The system and method of claim 1 , wherein at least one of said predetermined wavelengths is associated with fat properties and at least one wavelength is associated with protein properties of said meat.
14. The system and method of claim 1 wherein said meat is beef.
15. The system and method of claim 1 wherein said image analysis comprises correlating one or more image-textural features with tenderness of said raw meat, wherein said correlating comprises applying a pattern-recognition algorithm to said textural features.
16. The system and method of claim 15 , wherein said image-textural features are correlated by performing at least one computation of statistical values based on said image.
17. The system and method of claim 16 , wherein said computations are selected from the group consisting of a co-occurrence matrix analysis, a wavelet analysis, a covariogram analysis and an analysis utilizing Gabor filters.
18. The system and method of claim 15 , wherein said image-textual features comprise one or more of a mean, a variance, a homogeneity, a contrast, a dissimilarity, an entropy, a second moment, a correlation, autocorrelation, skewness and kurtosis.
19. The system and method of claim 1 wherein said forecast of tenderness is associated with slice shear force values corresponding to cooked cuts of beef.
20. The system and method of claim 1 , wherein said forecast of tenderness is associated with values selected from the group consisting of Warner-Bratzler shear force values, slice shear force values, sensory analysis and combinations thereof.
21. The system and method of claim 1 , wherein said meat is beef, and wherein a multispectral image of the cross section of a ribeye muscle exposed from a carcass side is captured.
22. The system and method of claim 1 , further comprising determining one or more additional properties of said beef from the same image simultaneously.
23. The system and method of claim 1 , further comprising an equipment and sample stabilization means so that shaking of camera is minimized allowing good quality images to be acquired.
24. The system and method of claim 1 , wherein an image acquisition means is comprises a camera, a wavelength dispersion unit, a lens, a light source and a housing.
25. The system and method of claim 24 , wherein said wavelength dispersion unit is selected from spectrograph LCTF- and AOTF-based systems.
26. The system and method of claim 24 , wherein said wavelength dispersion unit is a set of band-pass filters in a wheel, whereupon when the filter wheel is rotated a spectral band is collected.
27. A system and method for analyzing one or more properties of meat, comprising an image acquisition system and an image analysis system, which image acquisition system employs multispectral imaging of said meat at predetermined key wavelength bands and communicates said images to said image analysis system which extracts data from said image, processes it according to programs residing therein, and produces an output which comprises a conclusion about said one or more of said properties, wherein said image is acquired at a time (t 0 ) and said output is a forecast of one or more biological properties of said meat at a time (t 1 ) which is later in time than (t 0 ).
28. The system and method of claim 27 , wherein said time (t 1 ) is the time when said meat will be available for preparation by a consumer.
29. The system and method of claim 27 , wherein said image is acquired at a time (t 0 ) and said output is also a prediction of the properties of said meat at said time (t 0 ).
30. The system and method of claim 29 , wherein said property is tenderness.
31. The system and method of claim 27 , wherein said number of predetermined wavelengths is from one to about twenty-five wavelengths.
32. The system and method of claim 31 , wherein said number of predetermined wavelengths is from two to twelve wavelengths.
33. The system and method of claim 32 , wherein said number of said predetermined wavelengths is four to eight wavelengths.
34. The system and method of claim 27 , wherein said predetermined wavelengths are in the visible/near infrared (VNIR) region (400-1000 nm).
35. The system and method of claim 34 , wherein said meat is beef and said wavelengths are selected from the group consisting of 455, 480, 520, 545, 568, 590, 604, 636, 722, 750, 814, and 850 nm and combinations thereof.
36. The system and method of claim 35 , wherein each of said wavelengths can vary by ±20 nm.
37. The system and method of claim 27 , wherein said predetermined wavelengths are in the NIR region (900-2500 nm).
38. The system and method of claim 37 , wherein said meat is beef and, wherein said wavelengths are selected from the group consisting of 1057, 1074, 1091, 1115, 1142, 1176, 1219, 1280, 1365, 1395, 1408, and 1462 nm and combinations thereof.
39. The system and method of claim 37 , wherein each of said wavelengths can vary by ±20 nm.
40. The system and method of claim 27 , wherein at least one of said predetermined wavelengths is associated with fat properties and at least one wavelength is associated with protein properties of said meat.
41. The system and method of claim 40 , wherein said meat is beef.
42. The system and method of claim 27 , wherein said image analysis comprises correlating one or more image-textural features with tenderness of said raw meat, wherein said correlating comprises applying a pattern-recognition algorithm to said textural features.
43. The system and method of claim 42 , wherein said image-textural features are correlated by performing at least one computation of statistical values based on said image.
44. The system and method of claim 43 , wherein said computations are selected from the group consisting of a co-occurrence matrix analysis, a wavelet analysis, a covariogram analysis and an analysis utilizing Gabor filters.
45. The system and method of claim 42 , wherein said image-textual features comprise one or more of a mean, a variance, a homogeneity, a contrast, a dissimilarity, an entropy, a second moment, a correlation, autocorrelation, skewness and kurtosis.
46. The system and method of claim 27 , wherein said meat is beef, and wherein a multispectral image of the cross section of a ribeye muscle exposed from a carcass side is captured.
47. The system and method of claim 27 , further comprising determining one or more additional properties of said beef from the same image simultaneously.
48. The system and method of claim 27 , further comprising an equipment and sample stabilization means so that shaking of camera is minimized allowing good quality images to be acquired.
49. The system and method of claim 27 , wherein an image acquisition means is comprises a camera, a wavelength dispersion unit, a lens, a light source and a housing.
50. The system and method of claim 49 , wherein said wavelength dispersion unit is selected from spectrograph LCTF- and AOTF-based systems.
51. The system and method of claim 49 , wherein said wavelength dispersion unit is a set of band-pass filters in a wheel, whereupon when the filter wheel is rotated a spectral band is collected.