Analysis of Color Feature-Based Segmentation and Classification of Fruits Images Using Linear Discriminant Analysis Method

This study discusses the analysis of fruit image segmentation and classification based on color features, with bananas and strawberries as the research objects. This study addresses the problem of separating fruit objects from the background and automatically classifying fruit types using simple, efficient, and interpretable color features. The objectives of this study are to analyze the effectiveness of color-based segmentation and evaluate the performance of the Linear Discriminant Analysis (LDA) method in classifying fruit images and analyzing separability between classes. The proposed method includes image pre-processing, segmentation based on a combination of Lab* and HSV color spaces, color feature extraction in the form of average Hue and Saturation values, and classification using Linear Discriminant Analysis. The dataset used consists of images of bananas and strawberries, which are divided into training data and test data. The results show that color-based segmentation effectively separates fruit objects from the background and produces a clear feature distribution between classes. The classification process using LDA produces an accuracy rate of 100% on the test data, indicating that the color features used have high discriminatory power. Based on these results, it can be concluded that the combination of color-based segmentation and the LDA method is effective for fruit image classification and is capable of providing good class separability analysis with low computational complexity.