A novel CBCD scheme based on local features category

Global and local features have been applied extensively to improve the performance of Content-Based Copy Detection (CBCD) systems. A novel CBCD scheme which uses global features to category local features is proposed in this paper, containing frame prerecession, database establishment and copy detection. The scheme aims to reach a fast speed and a high accuracy. A proper combination of global and local features in our scheme breeds higher efficiency and accuracy than taking use of any single feature alone. Experimental results indicate that the scheme achieves a rapid scene index and an accuracy frame-frame match.