Physics-Based Maritime Target Classification and False Alarms Mitigation Navy SBIR FY2016.1 Sol No.: Navy SBIR FY2016.1 Topic No.: N161-018 Topic Title: Physics-Based Maritime Target Classification and False Alarms Mitigation Proposal No.: N161-018-0633 Firm: Electromagnetic Systems, Inc. 108 Standard St. El Segundo, California 90245 Contact: Stephen Hershkowitz Phone: (310) 524-9103 Web Site: http://www.emagsys.com Abstract: Current Navy radar-based vessel classification relies on the generation of plan- and profile-view images of each target, the extraction of response positions in those images, and the comparison of the positions to a database that contains the locations of prominent vessel structures. However, reliance on just positional information limits the capability of the classifier to distinguish between similar vessels. EMSI proposes to demonstrate the feasibility of significantly improving vessel classification on legacy maritime radar systems, by applying Deep Learning networks to extract physical features from complex SAR, ISAR, and HRR data. To this end, we will extend our Deep Learning CNN target classification technology and apply it to simulated and real data. Benefits: The proposed technology applies directly to any wideband maritime radar data, whether for defense or homeland security applications. Initial potential transition platforms include the P-8 Poseidon, MH-60 Seahawk, and the MQ-8 Fire Scout aircraft. The proposed extraction of physical features applies more widely, to any surveillance radar. Return