The widespread use and increasing sophistication of surveillance systems, both military and civilian, have generated a great deal of interest in computer algorithms capable of initiating and developing air-vehicle tracks in a set of measurements. Improved optical sensors and a need for frequency diversity in weak-target detection surveillance have added optical imagery to the type of data that can be used to detect the presence of a moving target. Unfortunately, the desire for a wide-area coverage capability in most surveillance applications often limits the amount of signal-to-noise ratio (SNR) gain most optical systems can offer. This reduces the class of targets that can be detected and tracked to those characterized by a strong optical signature.; Detecting and tracking low-contrast moving-targets in optical image sequences requires a type of processing algorithm that enhances target energy while simultaneously reducing background clutter and system noise. The main theme of this dissertation is to develop new spatio-temporal matched-filtering routines for this purpose. These techniques give potential SNR improvements far in excess to the processing gains one derives from classical spatial matched-filtering.; After discussing general target tracking by data association and track-before-detect procedures, a moving-target-indication (MTI) algorithm based on a modified form of Three-Dimensional Matched-Filtering is described. This technique is a Fourier domain-based time-delay-and-integrate matched-filter algorithm.; A multi-spectral MTI procedure is derived next. This technique adds additional SNR gain to that created by the MTI processing through a weighted-differencing of two correlated images with uncorrelated target signatures.; When multiple targets traversing an image sequence are extremely dim, the SNR gain from either a data association or MTI tracker may not be sufficient to indicate their individual presences. A new maximum log-likelihood ratio test is described that combines the various integrated matched-filter results from several MTI processors into one effective intensity peak before thresholding. Thus overall multiple target detectability is improved.
展开▼
