Estimation of the Colour Properties of Apples Varieties Using Neural Network

The consumer acceptance and the quality standard of agricultural products such as apple are determined mostly by their colour. Colour is measured with a colorimeter and quantified using the C.I.E. L-*, a(*), b(*) colour space system. It is used commonly by researchers for the classification and identification of apple fruit. To the best of our knowledge, the present study is the first study investigating the prediction of some colour properties of six apple varieties through artificial neural networks (ANN). The apple varieties are 'Amasya', 'Starking', 'Granny Smith', 'Pink Lady', 'Golden Delicious', 'ArapkA +/- zA +/-' and the colour properties are L-* (lightness), a(*) (redness), b(*) (yellowness), C-* (chroma), h(*) (hue angle), CI (chroma index). General Regression Neural Networks (GRNN) and Adaptive Neuro Fuzzy Interface System (ANFIS) structures were employed to predict the colour properties. According to the experimental and simulation results, the proposed ANFIS predictor had a superior performance in prediction of these colour parameters.