A comparative study of feature extraction methods in defect classification of mangoes using neural network

The “king of fruits” Mango (Mangifera indica L.) is the most sought after fruit for both direct and indirect consumption across the globe. Since it has very high export value, there is a need to develop a technique that is capable of classifying the defects of mangoes objectively. Any classifier performance is dependent on the features extracted from the region of interest of the sample. In this paper, a comparative study of feature extraction methods is made to classify the visible defects of Mangoes. “Alphonso” mango cultivar was chosen for the experimentation. 1766 color images with different quality classes were acquired, pre-processed and textural features were extracted considering one feature at a time and also in combination for each color image. Hence, we obtained 9 different cases of different textural features combination. Furthermore, most relevant features were selected from each case using sequential forward selection algorithm. The textural features like statistical, LBP and filter banks were found to be effective in designing neural network (NN) using generalized linear model classifier with cross validated performance accuracy of 90.09%, 90.26% and 90.26% for linear, logistic and softmax activation functions respectively.