Geometric Invariant Moments for Extracting Image Geometric Features

Resource Overview

This MATLAB-based code extracts 7 Hu moments as feature vectors to characterize image geometric properties. It can be applied to content-based image retrieval systems for robust feature representation.

Detailed Documentation

This code provides a MATLAB implementation for extracting geometric features from images using invariant moments. It calculates 7 Hu moments as distinctive feature vectors, which are rotation, scale, and translation invariant descriptors suitable for content-based image retrieval applications. The implementation involves reading input images, preprocessing them (converting to grayscale if necessary), and computing central moments followed by normalized Hu moments through mathematical transformations. These features capture essential shape characteristics while remaining insensitive to geometric transformations. The code serves as a foundational template that can be extended with additional preprocessing steps, feature selection mechanisms, or similarity measurement algorithms to enhance retrieval accuracy for diverse application scenarios.