Visual Matching on User-Generated Photo Collections Logo Matching Algorithm Final Report

Abstract

Imagine you were assigned the task to maintain the map of a new city on the Internet, how would you keep on the pace of the fast developing city in order to make the map up-to-date? How would you know if there was a new restaurant springing up? Until now visual matching is still a challenging area due to variant factors such as affine distortion, change in 3D viewpoint, addition of noise and illumination change. In this paper, I focus on mining photos on a user-generated photo collections using logo matching algorithm. As this is a teamwork project, and my team mate Ximing Hou is concentrating on the logo mining application part, I will make experiments based on his application. Table of

Topics

    14 Figures and Tables

    Download Full PDF Version (Non-Commercial Use)