Identification of Tomato Maturity Based on HIS Color Space Using The K-Nearest Neighbour Method

Abstract

Post-harvest activities, namely sorting and grading, are usually done by the visual method by looking at the color or the size of the fruit. Technological advancements by using computer assistance make harvesting and detecting the ripening process of tomatoes easier. The purpose of this research is to apply the Hue, Intensity, Saturation (HIS) color transformation method to detect the ripeness of vegetable tomatoes and determine the accuracy of the HIS and Red, Green, Blue (RGB) methods. In this research, some samples of tomatoes' RGB values were taken through image processing by ripening level, between unripe, under-ripe, and ripe. After the RGB value is obtained, it will be converted into HIS using means, variance, and range color feature extraction. In classifying this research using the K-Nearest Neighbour (KNN) and Support Vector Machine (SVM) method. With the results, the accuracy of the KNN method is better than SVM with an average of 87.80% and 74.38%. The result of this research indicates that the more accurate method is HIS with an average of 89.65% compared to the RGB method with an average of 89.35% and the CIElab method with an average of 84.41%. This is because it is easier for HIS to identify objects with different hues by providing a threshold value in the range of hue values (spectral wavelengths) that surround the object.