Vision based model for identification of adulterants in milk

Abstract

Milk adulteration is a social problem that exists in both developed and developing countries. This is due to lack of regulations or enforcement, proper refrigeration techniques, high yields with no market and hence the use of high levels of different adulterants to elongate the shelf life, prevent spoilage, increase thickness and whiteness. This research proposes the use of a mobile phone application, to determine the intensity and type of adulterant used in milk, specifically water adulterant, by use of back propagation artificial neural network (ANN). A scanned image of milk spiked with acid-base indicator (bromothymol blue) was taken, after it changed color. Using ANN, the image was classified in terms of color descriptors such as mean of red (R), green (G), blue (B), luminosity (L, which is the sum of R, G, and B). After classification, partial least squares regression (PLSR) and principal component regression analysis (PCR) model, was used to predict the adulteration intensity in milk using the intensity of adulteration as a dependent variable.