BOF Image/Video Retrieval Model with Global Feature

Local interest points serve as an elementary building block in many video retrieval algorithms, and most of them exploit the local volume features using a Bag of Features (BOF) representation. Such representation, however, ignores potentially valuable information about the global distribution of interest points. In this paper, we first present an R feature to capture the detailed global geometrical distribution of interest points. Then, we propose a fusion strategy to combine the BOF representation with the global R feature for further improving recognition accuracy. Convincing experimental results on several publicly available datasets demonstrate that the proposed approach outperforms the state-of-the-art approaches in video retrieval.