Content-Based Image Retrieval .

Abstract:

This project explores the concept of matching images by specific features such as color, texture, image section, or using deep networks. It also introduces different distance metrics to compare 2 images of different colors, textures, etc, such as histogram intersection. When looking at matching images by features it's important to understand which features are of interest to match to other images, thus combining and weighing different or similar features can be helpful to achieve the desired result.