Hyperspectral Identification for Collaborative Tracking

This proposed SBIR program will capitalize on advances in hyperspectral processing and feature-aided tracking to enhance the military’s ability to detect and track moving targets over long time periods using data from UAV-based spectral imaging sensors. Existing multi-target tracking approaches that rely solely on spatial and temporal information from conventional electro-optical sensors fall short of demanding needs in complex environments, particularly urban areas. New technology developments such as time-sequential spectral imagers coupled with high-precision registration processing, and “hyperspectral video” cameras capable of acquiring spectral image cubes at fast frame rates, are removing previous limitations on simultaneous acquisition and utilization of high-resolution spatial, spectral and temporal information from moving platforms. Space Computer Corporation proposes to develop and demonstrate signature-based methods in support of collaborative target tracking, with emphasis on spectral-temporal signature extraction, evolution and matching to enable continuous recognition of specific targets moving in a complex scene. In Phase I we will provide a proof-of-concept for the signature-based tracking approach using data from a VNIR-band spectral imager in a stationary configuration. The Phase II program will optimize the algorithms, characterize tracking performance in realistic scenarios, and demonstrate a real-time processor implementation using data from an airborne hyperspectral imager.