Development of Probabilistic Neural Networks Model for Starfruits Sortation Based On RGB Images

Abstract

Starfruits (Averrhoa Carambola L.) is one of agroproducts which is liked by most society. Typically it is seen like a star when sliced athwartly. The characteristics of starfruits type which pre-eminent are size big, colour draw, smooth fibre, water containment, and fresh sweet. One of the vital process in post-harvest treatment is sortation. Fruits are grouped according to size measure and ripe phase. Starfruits which marketed have to fulfill the standard which accepted by consumer widely, either in domestic market and also global market. This matter in order to give best quality and also improve product competitiveness by farmer. Determination of starftruits sweetness is generally based on fruit colour visually. But, change of colour gradually complicated determination the taste of fruit. This needs more intelligent sortation methods and tools that overcome the sortcomings of manual process. Probabilistic Neural Network (PNN) is one of Artificial Neural Network (ANN) variants that can be used to develop a computer-based sortation engine for starfruits. This research aim is to develop PNN model for starfruits sortation based on RGB image value. Conclusion at this research indicates that PNN can be used to classify starfruits with high accuration reach 92%.