Durian cultivar recognition using discriminant function

Abstract

The distinction of the Durian Cultivar is its physical characteristics such as smell, thorn's color, the resonant sound when knocking the husk. This research study which characteristics can classify the two popular Durian Cultivar; “Chanee” and “Monthong”. The array of thorns in vertical, horizontal and diagonal and the geometric lines at the thorn's bases; rectangular, pentagon, hexagon and heptagon were the features used in this study. The process starts with Durian image edge detection to obtain the outline for identifying the position of thorn's peaks and the geometric outlines. The attributes are analyzed by the Linear Discriminant Analysis Method. The experimental results show that Durian Cultivar can be classified according to the thorn's array in vertical and horizontal. The results provide the efficient performance of classifier. The accuracy of the discriminative model is 94.44%.