Perception-Based Image Interpretation

This research project is directed at developing an operator perception model based on insight from empirical perception data combined with knowledge of how the human vision system processes information. As a first step in this direction, we are proposing to collect perceptual data and use this data to develop, implement, and evaluate concept approaches for augmenting human operator performance in interpreting images. We believe that the resulting system will significantly improve operator performance, and that in addition to solving an immediate problem will benefit other applications that require operator decisions from image data. Central to this approach is the application of a tool that Cybernet has developed for ARPA to collect and analyze visual perception and operator performance on such image interpretation problems. The proposed effort will produce technology that has significant commercial potential in both the government and commercial sectors. Potential government applications include image enhancement tools for the intelligence community and visual decision systems for strategic defense. Potential industrial/commercial applications include enhancing medical and dental images, industrial security, and computer art.