Techniques for Comparison of Actual Target Signatures to Rendered or Synthetically Generated Models

In the development of modern sensor systems it is necessary to supplement actual performance testing with both digital and hardware-in-the-loop simulations to speed development and offset field test costs. The evaluation of signal/image processing algorithms along with sensor performance is often the primary goal of these simulations. The value of these simulations is only realized when the fidelity of the synthetic target and clutter models used is sufficient to provide an acceptable assessment of performance. As sensors attain higher resolutions and weapon systems become more advanced the requirements on synthetic target and background models become more difficult to meet. Some statistical methods, which have been used in the past to assess the fidelity of synthetic models compared to actual field measurements, have not provided the necessary understanding of the quality of the synthetic models. The objective of this research is to improve synthetic model verification capabilities by developing techniques to quantitatively compare synthetic models with data collected during field tests.