Integrating Self-Organizing Map and Principal Component Analysis for Enhanced Herbal Leaf Image Classification

This study presents a classification model for herbal leaf images by combining Principal Component Analysis (PCA) for dimensionality reduction with Self-Organizing Map (SOM) as an unsupervised classification algorithm. The model was developed using a dataset of five herbal leaf types: Betel, Papaya, Moringa, Katuk, and Turmeric. Shape- based morphological features were extracted from segmented leaf images, including area, perimeter, eccentricity, major axis length, and minor axis length. PCA was applied to reduce the five-dimensional feature vectors into two principal components, enhancing data representation and reducing redundancy. SOM was then used to classify the PCA-transformed data. The proposed model achieved an accuracy of 94.44%, outperforming the SOM- only configuration, which attained 85.56%. The improvement demonstrates that PCA effectively enhances SOM performance by providing more informative inputs. These results confirm the potential of integrating dimensionality reduction with unsupervised learning for accurate and efficient herbal plant classification.