Being determined by the construction of the human eye whereupon the region of accurate vision is limited to a few degrees of visual angle, the human viewer is forced to move his eyes to build up an adequate internal representation of the environment. This process is supported by attentional steering procedures which determine the sequence of locations to be fixated. Thereby it has to be distinguished between intentional (top down) and stimulus-driven (bottom up) processes. For stimulus-driven processes there exists a number of parameters
which specify the amount of attentional attraction. Essential aspects are contrast relations, luminance profiles and complexity relations which can be determined by algorithms of digital image processing. Together with viewing habits these factors were implemented in a simulation model. The simulation model calculates attention attracting areas within the visual field. This information then was used to steer a two-axis camera system. The system operates completely autonomous and was able to scan the environment on the basis of salient stimulus properties. Problems of implementation and attempts to validate the system are reported. It is planned to demonstrate the performance of the whole system.