Current Army projects, the Intelligent Vehicle Highway System,and the Unmanned Ground Vehicle project require automated signal processingand information processing techniques that distinguish various types ofvehicles in real-time. Tasks, such as distinguishing 3-axle trucks from2-axle trucks or trucks from cars, require image understanding, artificialintelligence, and high performance computing. We proposed to use theadvanced processing architecture of the Image Understanding Architecture(IUA) in an AI paradigm that combines these various forms or processing. Inparticular, we proposed to utilize the IUA for recognizing vehicle typesfrom multisensor data fusion, and then using sensor and domain knowledge tocontrol the extraction, grouping, and matching of image features againststored abstract models for the various vehicle types. In Phase I, we willuse static images of military targets to experiment with extracting variousimage features such as lines, arcs, and regions from both visible and IRimages. These features will be perceptually organized to form symbolicdescriptions of complex image events. Model matching algorithms will bedeveloped to classify the image events based on abstract models of knownvehicles. This approach will be tested by measuring the accuracy and speedof the model matching process on the IUA.
