On Evaluation of Keypoint Matching Pattern for Image Near-Duplicate Detection

Wan-Lei Zhao,Chong-Wah Ngo
2009-01-01
Abstract:Near-duplicate (ND) detection appears as a timely issue re- cently, being regarded as a powerful tool for various emerg- ing applications. In the Web 2.0 environment particularly, the identiflcation of near-duplicates enables the tasks such as copyright enforcement, news topic tracking, image and video search. In this paper, we describe a practical framework for rapid ND detection. It mainly consists three compo- nents: bag-of-words representation, local keypoint matching and matching pattern evaluation. For the matching pattern evaluation, an algorithm, namely Scale-Rotation invariant Pattern Entropy (SR-PE), is given. It generally aims to measure the spatial regularity of matching patterns formed by local keypoints. Its efiectiveness is illustrated with a large-scale image set.
What problem does this paper attempt to address?