Shape Extraction and Equivalent Key Recognition (SEEKER)

ABSTRACT: The proposed Shape Extraction and Equivalent KEy Recognition (SEEKER) program will perform Automatic Target Recognition (ATR) of small targets (e.g. vehicles) from airborne LIDAR point clouds using a new method of recognition based on topological techniques. Recent years have seen rapid progress in the data collection abilities of airborne surveillance plat- forms. Key challenges remain in recognizing medium and small sized objects, such as vehicles, from both overhead or non-overhead (oblique angle) LIDAR data. SEEKER applies mathematical techniques used for processing n-dimensional data in big data and natural language applications to object recognition. The approach is noise resistant, invariant to rotation, reflection and scaling is- sues, and supports multi-scale classification. By exploiting features in the mathematical structures and real-world physical constraints, the algorithms support operation in a SWaP constrained envi- ronment. The technology transfer plan includes both commercial and U.S. Government customer targets.; BENEFIT: SEEKER is applicable to two broad transition paths, commercial and government. Com- mercial applications include infrastructure, agriculture, and public safety. Government applications include vehicle based ATR as well as terrestrial applications in intelligence analysis and visualiza- tion.