Preliminary estimation of fat depth in the lamb short loin using a hyperspectral camera
S. Rahman A , P. Quin A D , T. Walsh A , T. Vidal-Calleja A , M. J. McPhee B , E. Toohey C and A. Alempijevic AA Center for Autonomous Systems, University of Technology Sydney, 15 Broadway, Ultimo, NSW 2007, Australia.
B NSW Department of Primary Industries, Livestock Industries Centre, Armidale, NSW 2351, Australia.
C NSW Department of Primary Industries, PO Box 865, Dubbo, NSW 2830, Australia.
D Corresponding author. Email: phillip.quin@uts.edu.au
Animal Production Science 58(8) 1488-1496 https://doi.org/10.1071/AN17795
Submitted: 10 November 2017 Accepted: 9 April 2018 Published: 7 May 2018
Abstract
The objectives of the present study were to describe the approach used for classifying surface tissue, and for estimating fat depth in lamb short loins and validating the approach. Fat versus non-fat pixels were classified and then used to estimate the fat depth for each pixel in the hyperspectral image. Estimated reflectance, instead of image intensity or radiance, was used as the input feature for classification. The relationship between reflectance and the fat/non-fat classification label was learnt using support vector machines. Gaussian processes were used to learn regression for fat depth as a function of reflectance. Data to train and test the machine learning algorithms was collected by scanning 16 short loins. The near-infrared hyperspectral camera captured lines of data of the side of the short loin (i.e. with the subcutaneous fat facing the camera). Advanced single-lens reflex camera took photos of the same cuts from above, such that a ground truth of fat depth could be semi-automatically extracted and associated with the hyperspectral data. A subset of the data was used to train the machine learning model, and to test it. The results of classifying pixels as either fat or non-fat achieved a 96% accuracy. Fat depths of up to 12 mm were estimated, with an R2 of 0.59, a mean absolute bias of 1.72 mm and root mean square error of 2.34 mm. The techniques developed and validated in the present study will be used to estimate fat coverage to predict total fat, and, subsequently, lean meat yield in the carcass.
Additional keywords: hyperspectral imaging, lamb processing, meat composition.
References
Anon. (2005) ‘Handbook of Australian Meat.’ 7th edn. (International Red Meat Manual)Boyd S, Vandenberghe L (2004) ‘Convex optimization.’ (Cambridge University Press: UK)
Burges CJ (1998) A tutorial on support vector machines for pattern recognition. Data Mining and Knowledge Discovery 2, 121–167.
| A tutorial on support vector machines for pattern recognition.Crossref | GoogleScholarGoogle Scholar |
Cannell RC, Tatum JD, Belk KE, Wise JW, Clayton RP, Smith GC (1999) Dual-component video image analysis system (VIASCAN) as a predictor of beef carcass red meat yield percentage and for augmenting application of USDA yield grades. Journal of Animal Science 77, 2942–2950.
| Dual-component video image analysis system (VIASCAN) as a predictor of beef carcass red meat yield percentage and for augmenting application of USDA yield grades.Crossref | GoogleScholarGoogle Scholar | 1:CAS:528:DyaK1MXntlOksLo%3D&md5=f84769947b68a47db6aa1bb915fe6541CAS |
Chen T, Morris J, Martin EL (2007) Gaussian process regression for multivariate spectroscopic calibration. Chemometrics and Intelligent Laboratory Systems 87, 59–71.
| Gaussian process regression for multivariate spectroscopic calibration.Crossref | GoogleScholarGoogle Scholar | 1:CAS:528:DC%2BD2sXkslyqtr0%3D&md5=65ea15e9d54e0abd4dc3f7281abb8d51CAS |
Crichton S, Kirchner S, Porley V, Retz S, Gersdorff G, Hensel O, Weygandt M, Sturm B (2017a) Classification of organic beef freshness using VNIR hyperspectral imaging. Meat Science 129, 20–27.
| Classification of organic beef freshness using VNIR hyperspectral imaging.Crossref | GoogleScholarGoogle Scholar |
Crichton S, Kirchner S, Porley V, Retz S, Gersdorff G, Hensel O, Sturm B (2017b) High pH thresholding of beef with VNIR hyperspectral imaging. Meat Science 134, 14–17.
| High pH thresholding of beef with VNIR hyperspectral imaging.Crossref | GoogleScholarGoogle Scholar | 1:CAS:528:DC%2BC2sXht1Wns7vM&md5=19f3e4168873da8e3b65ef510e811ab8CAS |
Garcia-Allende PB, Anabitarte Garcia F, Conde Portilla OM, Mirapeix Serrano JM, Madruga Saavedra FJ, Lo’pez Higuera JM (2008) ‘Support vector machines in hyperspectral imaging spectroscopy with application to material identification.’ (SPIE Society of Photo-Optical Instrumentation Engineers)
Gardner G, Glendenning R, Brumby O, Starling S, William A (2015) The development and calibration of a dual X-ray absorptiometer for estimating carcass composition at abattoir chain-speed. In ‘Fourth annual conference on body and carcass evaluation, meat quality, software and traceability’, Edinburgh, Scotland.
Hopkins D, Safari E, Thompson J, Smith C (2004) Video image analysis in the Australian meat industry: precision and accuracy of predicting lean meat yield in lamb carcasses. Meat Science 67, 269–274.
| Video image analysis in the Australian meat industry: precision and accuracy of predicting lean meat yield in lamb carcasses.Crossref | GoogleScholarGoogle Scholar | 1:STN:280:DC%2BC3MbnsVShsQ%3D%3D&md5=8f13dc872b0d3c53b2c930e176520c8eCAS |
Huang H, Liu L, Ngadi M (2014) Recent developments in hyperspectral imaging for assessment of food quality and safety. Sensors 14, 7248–7276.
| Recent developments in hyperspectral imaging for assessment of food quality and safety.Crossref | GoogleScholarGoogle Scholar | 1:CAS:528:DC%2BC2MXjt1Wrurw%3D&md5=a6b92a54385735a35b7d408e5402f0f9CAS |
Huynh CP, Robles-Kelly A (2010) A solution of the dichromatic model for multispectral photometric invariance. International Journal of Computer Vision 90, 1–27.
| A solution of the dichromatic model for multispectral photometric invariance.Crossref | GoogleScholarGoogle Scholar |
Kempster AJ (1981) Fat partition and distribution in the carcasses of cattle, sheep and pigs: a review. Meat Science 5, 83–98.
| Fat partition and distribution in the carcasses of cattle, sheep and pigs: a review.Crossref | GoogleScholarGoogle Scholar | 1:STN:280:DC%2BC3MbmvFequg%3D%3D&md5=0d658d349b72bf4275f861eec610a899CAS |
Kongsro J, Røe M, Kvaal K, Aastveit AH, Egelandsdal B (2009) Prediction of fat, muscle and value in Norwegian lamb carcasses using EUROP classification, carcass shape and length measurements, visible light reflectance and computer tomography (CT). Meat Science 81, 102–107.
| Prediction of fat, muscle and value in Norwegian lamb carcasses using EUROP classification, carcass shape and length measurements, visible light reflectance and computer tomography (CT).Crossref | GoogleScholarGoogle Scholar | 1:STN:280:DC%2BC3Mbns1ajuw%3D%3D&md5=c7991a96d8b1b904a09e84a993f6fd73CAS |
Lambe NR, Navajas EA, Bünger L, Fisher AV, Roehe R, Simm G (2009) Prediction of lamb carcass composition and meat quality using combinations of post-mortem measurements. Meat Science 81, 711–719.
| Prediction of lamb carcass composition and meat quality using combinations of post-mortem measurements.Crossref | GoogleScholarGoogle Scholar | 1:STN:280:DC%2BC38zisFemug%3D%3D&md5=9017bc8f3a91d514bb184af4950ca7e1CAS |
Naganathan GK, Grimes LM, Subbiah J, Calkins CR, Samal A, Meyer GE (2008) Visible/near-infrared hyperspectral imaging for beef tenderness prediction. Computers and Electronics in Agriculture 64, 225–233.
| Visible/near-infrared hyperspectral imaging for beef tenderness prediction.Crossref | GoogleScholarGoogle Scholar |
Pu H, Sun D, Ma J, Chen J (2015) Classification of fresh and frozen-thawed pork muscles using visible and near infrared hyperspectral imaging and textural analysis. Meat Science 99, 81–88.
| Classification of fresh and frozen-thawed pork muscles using visible and near infrared hyperspectral imaging and textural analysis.Crossref | GoogleScholarGoogle Scholar |
Rahman S, Robles-Kelly A (2013) An optimisation approach to the recovery of reflection parameters from a single hyperspectral image. Computer Vision and Image Understanding 117, 1672–1688.
| An optimisation approach to the recovery of reflection parameters from a single hyperspectral image.Crossref | GoogleScholarGoogle Scholar |
Rasmussen CE, Williams C (2006) ‘Gaussian processes for machine learning.’ (MIT Press: Cambridge, MA)
Saadatian F, Liu L, Ngadi MO (2015) Hyperspectral imaging for beef tenderness assessment. International Journal of Food Processing Technology 2, 18–25.
| Hyperspectral imaging for beef tenderness assessment.Crossref | GoogleScholarGoogle Scholar |
Shafer SA (1985) Using color to separate reflection components. Color Research and Application 10, 210–218.
| Using color to separate reflection components.Crossref | GoogleScholarGoogle Scholar |
Siddell J, McLeod BM, Toohey ES, van de Ven R, Hopkins DL (2012) The prediction of meat yield in lamb carcasses using primal cut weights, carcass measures and the Hennessy grading probe. Animal Production Science 52, 584–590.
Vert JP, Tsuda K, Scho¨lkopf B (2004) A primer on kernel methods. In ‘Kernel methods in computational biology’. (Eds B Schölkopf, K Tsuda, JP Vert) pp. 35–70. (MIT Press: Cambridge, MA)
Williams A, Anderson F, Siddell J, Pethick DW, Hocking Edwards JE, Gardner GE (2017) Predicting lamb carcase composition from carcase weight and gr tissue depth. In ‘63rd International congress of meat science and technology’, Cork, Ireland. (Eds D Troy, C McDonnell, L Hinds, J Kerry) pp. 729–732. (Wageningen Academic Publishers)
Zhu F, Zhang H, Shao Y, He Y, Ngadi MO (2014) Mapping of fat and moisture distribution in atlantic salmon using near-infrared hyperspectral imaging. Food and Bioprocess Technology 7, 1208–1214.
| Mapping of fat and moisture distribution in atlantic salmon using near-infrared hyperspectral imaging.Crossref | GoogleScholarGoogle Scholar | 1:CAS:528:DC%2BC2cXjs1Cmtb0%3D&md5=df5d609227f036bd7fba9ba3e961ede0CAS |
