Online feature selection based on generalized feature contrast model

To really bridge the gap between high-level semantics and low-level features in content-based image retrieval (CBIR), a problem that must be solved is: which features are suitable for explaining the current query concept. In this paper, we propose a novel feature selection criterion based on a psychological similarity measurement - generalized feature contrast model, and implement an online feature selection algorithm in a boosting manner to select the most representative features and do classification during each feedback round. The advantage of the proposed method is: it doesn't require Gaussian assumption for "relevant" images as other online FS methods; it accounts for the intrinsic asymmetry between "relevant" and "irrelevant" image sets in CBIR online learning; it is very fast. Extensive experiments have shown our algorithm's effectiveness.