Detecting and tracking low-observable targets using IR

Abstract

Detecting and Tracking Low—Observable Targets Using IRM. F. Fernandez, A. Aridgides, and D. BrayGE Advanced Technology LaboratoriesElectronics Parkway, Syracuse, N.Y. 13221ABSTRACTSensitivity to low—observable targets—-given a sequence of N frames of pre—processed(e.g., spatial/time-filtered) but unthresholded data—-may be enhanced by selecting,from among all possible paths traversing these frames, those containing anyindication that a target may be present. However, since only upper bounds regardingtarget velocities are usually known, explicit formulation of all feasible paths (andaccompanying confidence factors) becomes a formidable task even for small values ofN. In this paper we address this problem by utilizing a Dynamic Programming(Viterbi) algorithm to efficiently generate and evaluate, in an unthresholdedfashion, all possible paths through the N frames. Trajectories are tracedrecursively by assigning accumulated trajectory scores to each entry in a givenframe of data so as to maximize that entry's updated score. This Viterbi Track-Before—Detect procedure differs from standard Multiple Hypothesis Testing (MHT)methods in two ways. First, while in the MHT method the number of plausible pathsgrows exponentially (hence the need for introducing thresholds), in the Viterbiapproach they remain constant, equal to the number of data entries in a frame.Second, whereas in the MHT method trajectories are updated by selecting for eachexisting trajectory the best current (thresholded) detection, in the Viterbiapproach each current data value is updated with the best trajectory up to thatpoint. Simulation results show that application of the Viterbi Track-Before-Detectalgorithm over ten frames of IR data yields roughly a 7 dB improvement in detectionsensitivity over conventional thresholding/peak-detection procedures.1. BACKGROUNDOur philosophy in addressing the low-observable target problem (regardlessof the number of sensors being used) is that the closer the data processingfunctions can be brought to the sensor, the better the attainable performance. Inother words, the detection (or threshold exceedence) decision should be avoidedor retarded because, quite simply, as soon as a thresholding function is invoked,information gets irretrievably lost. The low-observable detection/tracking problemthen becomes how to efficiently process the data so as to keep a manageable falsealarm rate and thus maintain a feasible computational load.For an infrared (IR) sensor, given a sequence of N frames of pre—processed (i.e.,clutter—filtered) but unthresholded data, sensitivity to low—observable targets maybe further enhancedThy selecting, from among all possible paths traversing theseframes, those containing any indiation that a target may be present. If thevelocity of the targets were known--and assuming that the filtered data has beensufficiently decorrelated between frames-—simple addition of the target strengthscould theoretically produce a Signal—to—Clutter ratio (SCR) gain of up to lOlogN dB.