Adaptive thresholding and windowing in real-time multispectral correlation

Abstract

Both thresholding and windowing are important tools in the reduction of false alarm rates in high sensitivity point source detection and correlation sensor signal processors. In a variable background environment, adaptive thresholding is required to maintain a constant false alarm rate, and in a variable target environment, adaptive windowing is helpful in controlling false alarm rate, while maintaining correlation performance. Several issues arise when applying adaptive detection thresholding and windowing in a scanning infrared sensor processor. Choice of control loops for both thresholding and windowing should be such that instabilities caused by interactions between them are minimized. Instabilities in operation can also occur due to the various states of the system itself, such as scanning state changes or gain mode changes. The current presentation will describe the application of adaptive thresholding and windowing in the sensor subsystem design of the Airborne Surveillance Testbed. Control loop closure choices will be outlined for both thresholding and windowing, and their effect on false alarm rate and correlation will be covered. The correction of scanning state and gain mode effects on thresholding will also be addressed, as well as the impact on correlation performance.