Guiding Attention by Cooperative Cues

A common assumption in visual attention is based on therationale of ``limited capacity of information processing''. From thisview point there is little consideration of how different informationchannels or modules are cooperating because cells in processing stagesare forced to compete for the limited resource. To examine themechanism behind the cooperative behavior of information channels, acomputational model of selective attention is implemented based on twohypotheses. Unlike the traditional view of visual attention, thecooperative behavior is assumed to be a dynamic integration processbetween the bottom-up and top-down information. Furthermore, top-downinformation is assumed to provide a contextual cue during selectionprocess and to guide the attentional allocation among many bottom-upcandidates. The result from a series of simulation with still and videoimages showed some interesting properties that could not be explainedby the competitive aspect of selective attention alone.