Tomato Ripeness Detection Using Linear Discriminant Analysis Algorithm with CIELAB and HSV Color Spaces

Abstract

Tomatoes have a relatively short ripening period, making it essential to identify their ripeness level before distribution. The ripeness level of tomatoes can be detected based on their color. Therefore, the color of tomatoes serves as a crucial indicator in determining whether they are ripe and of good quality. However, classifying tomato ripeness levels manually has several drawbacks, namely requiring a long process, a low level of accuracy, and being inconsistent. The research aimed at developing a detection model for the ripeness level of tomatoes using the LDA algorithm based on color feature extraction, namely CIELAB (L*a*b) and HSV. The L*a*b and HSV color spaces are applied to obtain information about the color of the object being detected. Furthermore, the information obtained from feature extraction is then grouped by class using the LDA algorithm, which separates information for each class and limits the spread between classes through linear projection searches to maximize the covariance matrix between classes so that members within the class can be identified. This research produces a model that can detect the level of ripeness of tomatoes with an accuracy of 88.194%.