Predict the quality and safety of chicken sausage through computer vision technology

Abstract

The aim of this study was to test the ability of image technology to predict quality and safety of chicken sausage. Chicken sausages were chosen for image capture. Traits evaluated were color indexes (L*, a*, b*), pH, drip loss, cooking loss, dry matter, moisture, crude protein, ether extract, ash, thiobarbituric acid reactive substances (TBARS), peroxide value (POV), free fatty acid (FFA), total coliform count (TCC), total yeast and mold count (TYMC) and total viable count (TVC). Images were analyzed using the software Matlab (R2015a). Conventional analytical technology i.e., proximate, bio-chemical and microbiological analyses were followed for reference value. Calibration and prediction model were fitted using The Unscrambler X software. Results of this work show that image technology may be a useful tool for prediction of meat quality traits in the laboratory and meat processing industries. The L* value from imaging analysis had medium correlation with a* (r=0.28), b* (r=0.29), pH (r=0.31). A medium correlation found in CP (0.29) with „a*‟ value obtained from imaging analysis. In this experiment we found lower calibration and prediction accuracy in a*, crude protein and ether extract value. From this study it may be recapitulated that image technology has a potentiality to replace analytical technology for meat laboratory and processing units.