Interactions with ATR

The image exploitation task is becoming the bottleneck of information gathering in the reconnaissance domain. The reasons are manifold. Human interpreters have to cope with an enormous amount of image data, multisensor data, a huge variety and multitude of target signatures, collateral data amount and time pressure. It is evident that help for reduction, sorting and prescreening is imperative. Reduction, sorting and prescreening takes place in the domain of space and time as well as in the domain of target object classes. Reduction in space and time is aided by detection and the reduction in the object domain is performed by recognition. ATR (Automatic Target Recognition) algorithms are on the way to give humans effective support in image exploitation, actually with greater emphasis on target detection. Additionally "intelligent" databases will help interpreters to perform target recognition. Embedding ATR in an interactive recognition assistant promises to become a fruitful and compre hensive approach for image interpretation in the near future.