# US 8,280,144 B2 · System and method for analyzing material properties using hyperspectral imaging

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 8,280,144 B2
- Document ID: US8280144B2
- Status: Granted
- Kind: B2
- Application number: 12/035,080
- Filed: 2008-02-21
- Published: 2008-08-21
- Granted: 2012-10-02
- Earliest claimed: 2007-02-21
- Assignees: Goldfinch Solutions, LLC
- Inventors: Jeyamkondan Subbiah; Chris Richard Calkins; Ashok Samal
- Web page: https://patentgrep.com/patent/US8280144B2

## CPC

- G01N 21/31: Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry {(G01N21/72 takes precedence)}
- G01N 33/12: Meat; Fish
- G01N 21/49: within a body or fluid
- G01N 2201/129: Using chemometrical methods
- G01N: INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES (measuring or testing apparatus or processes other than immunoassay, involving enzymes or microorganisms C12M, C12Q)

## Abstract

Systems and methods are provided for analyzing material properties of an object using hyperspectral imaging. An exemplary method includes obtaining a hyperspectral image of an object; analyzing the hyperspectral image according to an algorithm; and correlating data obtained from the analysis with material properties of the object.

## Claims

### Claim 1

A method for determining material properties of an animal tissue, said method comprising:
a. using an imaging component to obtain a hyperspectral image of at least a portion of said animal tissue, said hyperspectral image comprising data along the spectral axis consisting of hundreds of narrow spectral bands, said imaging component comprising a hyperspectral camera and a lighting system, said lighting system selected from the group consisting of a concentrated lighting system and a diffuse lighting system; wherein said hyperspectral camera is configured to perform a technique of imaging selected from the group consisting of reflectance, transmission, scattering, fluorescence, and x-ray transmission, and wherein said hyperspectral camera is configurable to collect images in the wavelength range of 200 nm to 12,000 nm, whereby said hyperspectral camera is actuated and a hyperspectral image of said portion of said animal tissue is obtained; and
b. using an analysis component comprising a system having computer programming to analyze said hundreds of spectral bands of said hyperspectral image to generate material properties data that describes material properties of said animal tissue, said system employing a reduction component to remove redundant information in said data contained in said hyperspectral image, whereby a simplified image is created, said system further employing an extraction component to extract image-textural features from said simplified image, thereby creating extracted image-textural feature data, said system further employing a recognition component for using pattern recognition algorithms to relate said extracted image-textural feature data to said material properties associated with said animal tissue, and said system further comprising an output component for outputting said materials property data,
wherein said animal tissue is a meat product, wherein said meat product is a fresh cut of beef, and wherein said material property is a prediction of cooked-beef tenderness, and wherein said beef is classifiable into a category based on said material property prediction of cooked-beef tenderness.

### Claim 2 (depends on 1)

The method of claim 1, wherein said removing of redundant information comprises performing at least one analysis selected from the group consisting of principal component analysis and partial least squares regression analysis to reduce the dimension of the data along the spectral axis.

### Claim 3 (depends on 1)

The method of claim 1, wherein said extracting of image-textural features is performed by a technique selected from a co-occurrence matrix analysis, a wavelet analysis, an analysis that utilizes Gabor filters, and combinations thereof.

### Claim 4 (depends on 1)

The method of claim 1, wherein a region-of-interest from the hyperspectral image is selected to correspond to an approximate location on said cut of beef at which a shearing procedure was performed.

### Claim 5 (depends on 4)

The method of claim 4, wherein a region-of-interest is selected from a lean region of said meat product and a fat region of said meat product.

### Claim 6 (depends on 5)

The method of claim 5, wherein a fat mean spectra of all fat pixels and a lean mean spectra of all lean pixels is determined by said analysis component from said hyperspectral image, and thereafter said analysis component performs partial least squares regression analysis on each of said fat and lean mean spectra.

### Claim 7 (depends on 6)

The method of claim 6, wherein said analysis component determines eigenvectors to explain variance in spectra.

### Claim 8 (depends on 6)

The method of claim 6, wherein said analysis component determines loading vectors for lean pixels and loading vectors for fat pixels.

### Claim 9 (depends on 8)

The method of claim 8, wherein the loading vectors give the coefficients for linear combinations of reflections at various wavelengths that may be used to determine tenderness scores.

### Claim 10 (depends on 9)

The method of claim 9, wherein said tenderness scores at each pixel are integrated or aggregated to determine a single tenderness category per image or sample.

### Claim 11 (depends on 9)

The method of claim 9, wherein said tenderness scores are analyzed by said analysis component by implementing programming selected from the group consisting of: an averaging scheme with thresholds for different categories, a nonlinear function defined to determine a tenderness category for the entire sample, and discriminant analysis used to classify each pixel in an image according to material properties.

### Claim 12 (depends on 6)

The method of claim 6, wherein said analysis component implements a regression approach to determine tenderness scores at each pixel using partial least squares regression analysis.

### Claim 13 (depends on 1)

The method of claim 1, wherein said analysis component processes said hyperspectral image to obtain a spectral profile from a single pixel of said hyperspectral image.

### Claim 14 (depends on 13)

The method of claim 13, wherein said spectral profile of a pixel corresponds to a lean portion of a cut of beef.

### Claim 15 (depends on 13)

The method of claim 13, wherein said spectral profile of a pixel corresponds to a fat portion of a cut of beef.

### Claim 16 (depends on 1)

The method of claim 1, wherein said pattern recognition algorithm is selected from the group consisting of regression, discriminant analysis, neural networks, and fuzzy modeling, whereby said textural features are related to properties associated with said animal tissue.
