May 4, 2010
Ph.D. student Andres Rodriguez earned Best Student Paper honors at the SPIE Defense and Security Symposium's Automatic Target Recognition Conference last month. In his work with Professor Kumar Bhagavatula, Rodriguez studies correlation filters (CFs), which can detect and recognize multiple targets in one scene - making them well-suited for automatic target recognition (ATR) applications. His paper, "Automatic Multi-Target Recognition From Two Classes Using Quadratic Correlation Filters," builds on the work of recent Carnegie Mellon Ph.D. graduate Ryan Kerekes. The paper presents a method to increase a correlation filter's capabilities to detect and recognize two classes of targets and reject clutter, and combines the CF outputs from sequential video frames using a Bayesian framework to improve detection. The award, sponsored by Lockheed Martin, included a cash prize.