Non-Cooperative Moving Target Recognition Using ISAR.

摘要

Ground vehicles move so irregularly that a radar image formed at an arbitrary time will generally not allow identification. Usable images can be formed only at those times, and with those durations, when the target motion happens to be benign. The key to selecting such intervals is the realization that vehicle motion during an interval is correlated with the quality of individual responses in the corresponding image. We automated an algorithm for selecting intervals of benign motion by measuring distortions of individual responses. Benign motion is necessary but not sufficient for identification; the vehicle's apparent rotation must also provide sufficient crossrange resolution. We automated an algorithm that selects intervals of sufficient rotation by measuring the orientations of the vehicle in a sequence of short-duration images.