Shape-Based Image Classification using Hu and Zernike Moments with Bee Colony Optimization for Tuning SVM Parameters

Abstract

In this paper, we propose a hybrid approach for image classification that combines invariant shape descriptors with metaheuristic hyperparameter tuning. Specifically, Hu and Zernike moments are used to represent geometric and structural properties of objects within images, while a Support Vector Machine (SVM) is employed as the base classifier. To optimize the performance of the SVM, we integrate the Bee Colony Optimization (BCO) algorithm, which efficiently searches for the optimal combination of hyperparameters CCC and γ. Experimental results on publicly available Fruit and Cotton image datasets from Kaggle show that the proposed system significantly improves the classification performance, particularly in class-imbalanced scenarios. The application of BCO resulted in a stable solution with a weighted F1-score of 0.7933, demonstrating both convergence efficiency and robustness. This framework is adaptable to a wide range of image recognition tasks where shape plays a dominant role.