A Model of Saliency-Based Visual Attention for Rapid Scene Analysis
IEEE Transactions on Pattern Analysis and Machine Intelligence
A Computational Perspective on Visual Attention
A Computational Perspective on Visual Attention
| Hi-index | 0.00 |
It is impossible to identify all objects in the visual world at the same time. Accordingly, we must direct attention to specific objects in order to fully recognize them. The deployment of attention is far from random. Attention is guided toward likely targets by a limited set of stimulus attributes such as color and size ("classic guidance"). Attention is also guided by a number of scene-based properties. Thus, if we were looking for sheep, we would expect them on surfaces that could support sheep, not in mid-air. We use information about the 3D layout of a space to determine which objects could plausibly be sheep-sized in that space. This paper briefly reviews the diverse set of guiding properties and the rules that govern their use.