The Classification of Papaya using Digital Image Processing and Artificial Neural Networks

Abstract

The quality identification process using a visual human eyes has its limitations, it requires more power to sort out. Furthermore, it has different levels of human perception in identifying the quality. This research develops digital image processing programs and artificial neural network for the quality classification of papaya Calina IPB-9 into three quality classes: Super, A, and B classes. Shape feature is extracted consistsing of compactness and roundness. Texture feature which is extracted consisting of the value of energy, entropy, contras, homogeneity, inverse difference moment, variance, and dissimilarity obtained by GLCM, another texture feature sought a feature LBP. Color feature is extracted consisting of the mean red, greeen, blue, hue, saturation, and value. These features serve as input to the training backpropagation neural network. The recognition process results shows the features of energy and entropy can distinguish quality classes of the papaya with the best accuracy rate that is equal to 87%.

Development of a Data-Driven Model to Predict Landslide Sensitive Areas

“…Artificial neural networks are one of the computational methods inspired by the neural system of the human brain. One of the remarkable characteristics of this type of network is their ability to learn and the ability to generalize this learning, because of this feature, they make it possible to learn to understand patterns (El_Jerjawi and Abu-Naser, 2018). The most important advantage of artificial neural networks over regression methods for modeling a pattern is that there is no need for an initial model in linking input and output data (Lee et al, 2018).…”