Content-Based Image Retrieval System

This document introduces a content-based image retrieval system implemented in MATLAB. The system employs a modular architecture that allows users to easily modify and integrate new feature extraction algorithms through configurable function modules. Key technical advantages include the implementation of scalable feature vectors and similarity measurement algorithms (such as Euclidean distance or cosine similarity), which significantly reduces customization efforts. Users can personalize the system according to specific requirements by adapting the feature extraction pipeline - for instance, adding histogram-based features via color_histogram.m or texture features using gray_level_cooccurrence.m. This flexibility enables efficient image retrieval across various application domains, making the system a practical tool for research and industrial applications. The object-oriented design pattern ensures maintainability while providing hooks for integrating deep learning features through pre-trained CNN models.