Pengembangan Model Probabilistic Neural Networks untuk Penentuan Kematangan Belimbing Manis

Abstract

The objectives of this research were to develop PNN model in determining starfruit ripeness. Experiments done by involving 354 image data of starfruits that classified into 5 classes. Feature extraction from those hundreds images was yield an image representation among others: amount of red pixels (R), amount of green pixels (G), substraction between R and G (R-G), and average value of Hue. Those features then became PNN inputs. After enjoining the last two classes, the best result could reach accuracy untill 90.86%.