Enhancing target detection using a hybrid human-computer system

Our goal is to develop an assistive technology for image analysts, which helps them to extract information (e.g., detect targets) efficiently. The images are displayed using the rapid serial visual presentation (RSVP) paradigm, in which a large image is broken into numerous smaller chips which are then displayed serially in the center of the monitor at a high rate. Detection performance is affected by several factors, one of which is the target eccentricity. We show how computer vision (CV) can be used to reduce the mean eccentricity in a manner that is expected to improve detection performance, while maintaining overall throughput.