Partial least squares analysis of near-infrared hyperspectral images for beef tenderness prediction

Govindarajan Konda Naganathan, Lauren M. Grimes, Jeyamkondan Subbiah, Chris R. Calkins, Ashok K Samal, George E. Meyer

Research output: Contribution to journalArticle

75 Citations (Scopus)

Abstract

Tenderness is a primary determinant of consumer satisfaction of beef steaks. The objective of this study was to implement and test near-infrared (NIR) hyperspectral imaging to forecast 14-day aged, cooked beef tenderness from the hyperspectral images of fresh ribeye steaks (n = 319) acquired at 3-5 day post-mortem. A pushbroom hyperspectral imaging system (wavelength range: 900-1700 nm) with a diffuse-flood lighting system was developed. After imaging, steaks were vacuum-packaged and aged until 14 days postmortem. After aging, the samples were cooked and slice shear force (SSF) values were collected as a tenderness reference. After reflectance calibration, a Region-of-Interest (ROI) of 150 × 300 pixels at the center of longissimus muscle was selected. Partial least squares regression (PLSR) was carried out on each ROI image to reduce the dimension along the spectral axis. Gray-level textural co-occurrence matrix analysis with two quantization levels (64 and 256) was conducted on the PLSR bands to extract second-order statistical textural features. These features were then used in a canonical discriminant model to predict three beef tenderness categories, namely tender (SSF ≤ 205.80 N), intermediate (205.80 N < SSF < 254.80 N), and tough (SSF < 254.80 N). The model with a quantization level of 256 performed better than the one with a quantization level of 64. This model correctly classified 242 out of 314 samples with an overall accuracy of 77.0%. Fat, protein, and water absorption bands were identified between 900 and 1700 nm. Our results show that NIR hyperspectral imaging holds promise as an instrument for forecasting beef tenderness.

Original languageEnglish (US)
Pages (from-to)178-188
Number of pages11
JournalSensing and Instrumentation for Food Quality and Safety
Volume2
Issue number3
DOIs
StatePublished - Sep 25 2008

Fingerprint

Beef
Infrared radiation
Infrared imaging
Water absorption
Oils and fats
Imaging systems
Muscle
Absorption spectra
Aging of materials
Lighting
Pixels
Fats
Calibration
Vacuum
Proteins
Imaging techniques
Wavelength
Hyperspectral imaging

Keywords

  • Beef tenderness
  • Instrument grading
  • Near-infrared hyperspectral imaging

ASJC Scopus subject areas

  • Chemical Engineering(all)
  • Safety, Risk, Reliability and Quality
  • Industrial and Manufacturing Engineering

Cite this

Partial least squares analysis of near-infrared hyperspectral images for beef tenderness prediction. / Naganathan, Govindarajan Konda; Grimes, Lauren M.; Subbiah, Jeyamkondan; Calkins, Chris R.; Samal, Ashok K; Meyer, George E.

In: Sensing and Instrumentation for Food Quality and Safety, Vol. 2, No. 3, 25.09.2008, p. 178-188.

Research output: Contribution to journalArticle

Naganathan, Govindarajan Konda ; Grimes, Lauren M. ; Subbiah, Jeyamkondan ; Calkins, Chris R. ; Samal, Ashok K ; Meyer, George E. / Partial least squares analysis of near-infrared hyperspectral images for beef tenderness prediction. In: Sensing and Instrumentation for Food Quality and Safety. 2008 ; Vol. 2, No. 3. pp. 178-188.
@article{0940ac87b775474988b7dde11b00f5d6,
title = "Partial least squares analysis of near-infrared hyperspectral images for beef tenderness prediction",
abstract = "Tenderness is a primary determinant of consumer satisfaction of beef steaks. The objective of this study was to implement and test near-infrared (NIR) hyperspectral imaging to forecast 14-day aged, cooked beef tenderness from the hyperspectral images of fresh ribeye steaks (n = 319) acquired at 3-5 day post-mortem. A pushbroom hyperspectral imaging system (wavelength range: 900-1700 nm) with a diffuse-flood lighting system was developed. After imaging, steaks were vacuum-packaged and aged until 14 days postmortem. After aging, the samples were cooked and slice shear force (SSF) values were collected as a tenderness reference. After reflectance calibration, a Region-of-Interest (ROI) of 150 × 300 pixels at the center of longissimus muscle was selected. Partial least squares regression (PLSR) was carried out on each ROI image to reduce the dimension along the spectral axis. Gray-level textural co-occurrence matrix analysis with two quantization levels (64 and 256) was conducted on the PLSR bands to extract second-order statistical textural features. These features were then used in a canonical discriminant model to predict three beef tenderness categories, namely tender (SSF ≤ 205.80 N), intermediate (205.80 N < SSF < 254.80 N), and tough (SSF < 254.80 N). The model with a quantization level of 256 performed better than the one with a quantization level of 64. This model correctly classified 242 out of 314 samples with an overall accuracy of 77.0{\%}. Fat, protein, and water absorption bands were identified between 900 and 1700 nm. Our results show that NIR hyperspectral imaging holds promise as an instrument for forecasting beef tenderness.",
keywords = "Beef tenderness, Instrument grading, Near-infrared hyperspectral imaging",
author = "Naganathan, {Govindarajan Konda} and Grimes, {Lauren M.} and Jeyamkondan Subbiah and Calkins, {Chris R.} and Samal, {Ashok K} and Meyer, {George E.}",
year = "2008",
month = "9",
day = "25",
doi = "10.1007/s11694-008-9051-3",
language = "English (US)",
volume = "2",
pages = "178--188",
journal = "Sensing and Instrumentation for Food Quality and Safety",
issn = "1932-7587",
publisher = "Springer New York",
number = "3",

}

TY - JOUR

T1 - Partial least squares analysis of near-infrared hyperspectral images for beef tenderness prediction

AU - Naganathan, Govindarajan Konda

AU - Grimes, Lauren M.

AU - Subbiah, Jeyamkondan

AU - Calkins, Chris R.

AU - Samal, Ashok K

AU - Meyer, George E.

PY - 2008/9/25

Y1 - 2008/9/25

N2 - Tenderness is a primary determinant of consumer satisfaction of beef steaks. The objective of this study was to implement and test near-infrared (NIR) hyperspectral imaging to forecast 14-day aged, cooked beef tenderness from the hyperspectral images of fresh ribeye steaks (n = 319) acquired at 3-5 day post-mortem. A pushbroom hyperspectral imaging system (wavelength range: 900-1700 nm) with a diffuse-flood lighting system was developed. After imaging, steaks were vacuum-packaged and aged until 14 days postmortem. After aging, the samples were cooked and slice shear force (SSF) values were collected as a tenderness reference. After reflectance calibration, a Region-of-Interest (ROI) of 150 × 300 pixels at the center of longissimus muscle was selected. Partial least squares regression (PLSR) was carried out on each ROI image to reduce the dimension along the spectral axis. Gray-level textural co-occurrence matrix analysis with two quantization levels (64 and 256) was conducted on the PLSR bands to extract second-order statistical textural features. These features were then used in a canonical discriminant model to predict three beef tenderness categories, namely tender (SSF ≤ 205.80 N), intermediate (205.80 N < SSF < 254.80 N), and tough (SSF < 254.80 N). The model with a quantization level of 256 performed better than the one with a quantization level of 64. This model correctly classified 242 out of 314 samples with an overall accuracy of 77.0%. Fat, protein, and water absorption bands were identified between 900 and 1700 nm. Our results show that NIR hyperspectral imaging holds promise as an instrument for forecasting beef tenderness.

AB - Tenderness is a primary determinant of consumer satisfaction of beef steaks. The objective of this study was to implement and test near-infrared (NIR) hyperspectral imaging to forecast 14-day aged, cooked beef tenderness from the hyperspectral images of fresh ribeye steaks (n = 319) acquired at 3-5 day post-mortem. A pushbroom hyperspectral imaging system (wavelength range: 900-1700 nm) with a diffuse-flood lighting system was developed. After imaging, steaks were vacuum-packaged and aged until 14 days postmortem. After aging, the samples were cooked and slice shear force (SSF) values were collected as a tenderness reference. After reflectance calibration, a Region-of-Interest (ROI) of 150 × 300 pixels at the center of longissimus muscle was selected. Partial least squares regression (PLSR) was carried out on each ROI image to reduce the dimension along the spectral axis. Gray-level textural co-occurrence matrix analysis with two quantization levels (64 and 256) was conducted on the PLSR bands to extract second-order statistical textural features. These features were then used in a canonical discriminant model to predict three beef tenderness categories, namely tender (SSF ≤ 205.80 N), intermediate (205.80 N < SSF < 254.80 N), and tough (SSF < 254.80 N). The model with a quantization level of 256 performed better than the one with a quantization level of 64. This model correctly classified 242 out of 314 samples with an overall accuracy of 77.0%. Fat, protein, and water absorption bands were identified between 900 and 1700 nm. Our results show that NIR hyperspectral imaging holds promise as an instrument for forecasting beef tenderness.

KW - Beef tenderness

KW - Instrument grading

KW - Near-infrared hyperspectral imaging

UR - http://www.scopus.com/inward/record.url?scp=77957727090&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77957727090&partnerID=8YFLogxK

U2 - 10.1007/s11694-008-9051-3

DO - 10.1007/s11694-008-9051-3

M3 - Article

AN - SCOPUS:77957727090

VL - 2

SP - 178

EP - 188

JO - Sensing and Instrumentation for Food Quality and Safety

JF - Sensing and Instrumentation for Food Quality and Safety

SN - 1932-7587

IS - 3

ER -