Annual Technical Report: Research in Knowledge-Based Automatic Feature Extraction

R. Nevatia and A. Huertas

This report provides a descrption of USC's activities as part of the DARPA APGD (automated population of geospatial databases) project. Focus of this effort is on detection, delineation and description of 3-D buildings from aerial images. Buildings are objects of obvious importance in urban environments and accurate models for them are needed for a number of tasks such as mission planning, mission rehearsal, tactical training, damage assessment and change detection. Difficulties in modeling buildings come from problems of segmentation, 3-D inference and shape description. Development of a multi-image based system is described. This system operates on a hypothesize and verify paradigm. Hypotheses for rectilinear rooftops are generated by hierarchical grouping of lower level features. 3-D cures, such as from walls and shadows are used to verify and select from these hypotheses. 3-D cues from range sensors, such as IFSAR, can be used when available. Results on selected windows of Ft. Hood, TX and Ft. Benning, GA, sites are presented. An efficient interactive system that can be used to edit the results of the automatic system or to construct new building models is also described. The interactive system uses partial results of the automatic system to reduce the effort required from the user. Current system is limited to rectilinear shaped buildings and assumes that ground height, camera models and illumination parameters are known.