Infrared Ship Detection at Low Signal-to-Noise

摘要

A multi-resolution signal processing algorithm (MRSI) was devised to maximize the detection range of imaging infrared seekers against ship targets. Image-based simulations performed on an IBM PC AT verify that the new algorithm provides a signal-to-noise enhancement (relative to hot-spot detection) given by eta N to 1/2 power PIX, where NPIX is the number of pixels occupied by the target in the image, and eta approx. equal 0.81. The processor efficiency, eta is remarkably invariant with respect to sensor/ship range (i.e., scale), viewing aspect, and ship class. Probabilities of detection, determined via 2500 image-based Monte Carlo simulations, agree with well-known analogous results of psychophysical experiments performed with human subjects. A new method is presented for suppressing false alarms caused by cloud reflections from the sea surface. Regardless of whether final target classification is performed on-board the missile by a computer algorithm or off-board by a person inspecting imagery telemetered from the missile, MRSI substantially improves predicted system performance. (Author)