Feature-Based Classification of Medicinal Plant: Pigweed (Uray) and Birch Flower (Himbabao) Using K-Nearest Neighbors

Resumo

This study focuses on the feature extraction of an individual image samples for both Pigweed and Birch Flower. The features are categorized with three Main types: Morphology, Color and Texture features. The extracted features will then be trained using the MATLAB Machine Learning using the Classification Learner App. Morphological features describes the shape and structural properties of the plant region in the image. This includes the area, perimeter, eccentricity, solidity, extent, major and minor axis length. The color features are also extracted from both RGB (Red, Green, Blue) and HSV (Hue, Saturation, Value) color spaces to capture the visual appearance of the plant region wherein the algorithm calculated the mean and standard deviation which describes the average and the variability of the color. The texture features characterized the spatial variation in pixel intensity, which is important for distinguishing surface patterns. This includes contrast, correlation, energy, and homogeneity. The result shows that the validation and testing accuracy level using the k-Nearest Neighbor ranges from 88.57-99.29% and 83.33-98.33% respectively. The study’s outcome demonstrates the potential of feature-based image analysis in the accurate identification of medicinal and botanical species, encouraging future developments in automated plant recognition systems.