Watermelon classification using k-nearest neighbours based on first order statistics extraction

Abstract

Watermelon (Citrullus Lanatus Tunb / Citrullus Vulgaris Schrad) is a vine that grows rapidly in Indonesia. Watermelon has many benefits that make watermelon one of the fruits sought by the community. The maturity of watermelon can be distinguished based on watermelon skin texture. The similarity in watermelon skin texture causes people difficulty in identifying the level of maturity of mature and immature watermelon. Based on these problems, research was carried out for the classification of watermelon maturity based on watermelon skin texture. The first-order statistical feature extraction method is used as a method to recognize watermelon maturity in terms of fruit skin texture in the classification process. The first order statistical characteristic parameters used are mean, variance, skewness, and kurtosis. Based on the extraction value of these statistical features, then used as a reference classification process using the k-nearest neighbour method. The data in this study are 100 images, of which 70 are for training data and 30 for testing data. From the results of testing the maturity classification of watermelon using k-nearest neighbour based on fruit skin texture based on extraction of firstorder features, 86.66% accuracy was obtained.