Image matching and retrieval system using object shape and structure

Abstract

This project works on the use of shape features for content-based image retrieval. Edges, lines, corners, and straight ribbons have been extracted from a query image. These features and their attributes, along with various relations are than used to match representations of other images in a database. Our project is mainly based upon the work done by Naveed Sarfraz Khan Khattak and is the continuity of his area of research. He developed his algorithms in MATLAB. We continued upon his work by improving the edge detection and line detection techniques. The better line detection has led to better corner and ribbon detection thus improving our work tremendously. As the project has been implemented in Visual C++ so we have the added advantage of fast processing.