Automated Visible Range Imaging Scheme to Identify Toxic Substance from Common Starchy Food

Toxic substance like acrylamide is a carcinogenic compound which is generally formed in the starchy food item when heated or fried to high temperatures. In the proposed work, a computer vision technique is employed to ascertain the presence of the acrylamide in the fried potato chips. K-means clustering has been used to perform a colour based segmentation of chips pixels from background. Distinct features, like standard deviation and moment,extracted from multi-channels, are fed to a random forest classifier for proper discrimination between acrylamide content potato chips samples and normal potato chips samples.Performance of the developed algorithm is evaluated on the comprehensive database of 80 sample images of the fried potato chips. The accuracy to detect the acrylamide contained potato chips is 95.83% which encourage the use of the proposed algorithm in real time application.