Modeling of interaction between different visual attentions

Visual attention plays an important role in human visual system, helping people to attend to things even before recognition. By its functionality, it can be treated as three kinds: SBA (spatial-based attention), OBA (object-based attention), and FBA (feature-based attention). Early works that model the mechanism of visual saliency mainly concern only one aspect of saliency. Recently, it has been shown that these three kinds of saliency can interact simultaneously, and work together to bias visual attention. In this paper, a unique mechanism is proposed to deal with this issue with both certain bio-plausibility and accuracy. The competition and cooperation between different visual scale, which can emphasis visual characteristics while minimize information, plays a center role of the model. This comprehensive saliency of different kinds and different spatial scales maybe an important forepart of object recognition.