Advanced Signature-Matched Hyperspectral Change Detection

NovaSol is pleased to present a program to improve the utility and performance of existing and future hyperspectral reconnaissance sensors. The development of approach to signature-matched hyperspectral change detection will address the Air Force’s specific interest in lowering false alarm rates while maintaining or increasing probability of detection. NovaSol’s novel approach improves the existing state of the art in hyperspectral matched-filter change detection by directly targeting shortcomings in current covariance-based methods that have been identified by the military and research community. Use of simple data conditioning steps and a spectral distance based matched filter detection metric make our algorithm computationally efficient and amenable to implementation in low power and weight parallel processing hardware. In addition, by employing object-based change detection, the proposed algorithm will offer greater tolerance to image misregistration than currently deployed pixel-level subtraction schemes.