Light Detection and Ranging (LIDAR) Surface Feature Extraction Tool

NAVAIR requires accurate and timely 3-D geospecific features, terrain, and imagery data to support Modeling and Simulation (M&S) software applications used in training and pre-mission rehearsal simulators. LIDAR data provides an excellent source for spatially accurate digital terrain and 3-D feature information; however, current visual database production techniques require hundreds of man-hours to extract and attribute features. Automated processes are necessary to achieve the Navy’s goal of good M&S data that is efficiently processed and correctly formatted. Bottlenecks in producing good data are a result of manual processes used in (1) registering old features and imagery to newly acquired imagery and LIDAR, (2) extracting and editing complex 3-D geometries, and (3) updating and attributing features. The proposed Phase II workplan promises to significantly alleviate these bottlenecks by developing a system that exploits the information content of LIDAR and multiband imagery using machine learning-based feature extraction technology. The envisioned system will have the capability of automating the extraction of M&S features from LIDAR, registering features to images, allowing dynamic interaction with analysts, interrogating analysts when it needs additional information, and embedding the ability to create a continuously updated repository of feature extraction models.