Evaluation of Laser Vibrometry ATR (ELVA)

"Laser Vibrometry is an exciting new sensor technology that promises to improve automatic target recognition (ATR) performance. Sensors such as AFRL's new Micro-Laser Interferometer Doppler (MLID) measure minute vibrations on the surface of an object from alarge standoff distance. As we demonstrated in Phase I, a baseline ATR designed for vibrometry data can reliably recognize military vehicles when trained and tested on similar operating conditions. However, since no fielded vibrometry ATR systems exist,performance on extended operating conditions (EOCs, conditions unlike those used for training) is unknown. Charles River Analytics proposes a Phase II effort to provide the foundation for MLID-based ATR with consistent high performance across a range ofEOCs. To accomplish this goal, we will specify operating conditions which maximize the expected ATR performance improvement when vibrometry data collected for those operating conditions is added to the ATR's training set. We will use such data (collectedby AFRL) to improve our baseline ATR design and optimize its performance for the widest possible set of EOCs. For future research in this highly promising field, we will also develop a software toolset with an integrated graphical user interface andback-end database support for vibrometry ATR design and evaluation."