Maximum Likelihood Narrowband Radar Data Segmentation and Centroid Processing

摘要

Electronically scanned narrowband radar systems detect non-extended targets in one or two range cells depending on whether the object straddles the range cell boundary. For two detections, the range estimate may be refined using a fusion process. However, for scenarios with multiple closely spaced objects ambiguity exists in how many objects are present and how the range cells should be paired to produce the refined estimates. In this paper, we present a new algorithm that first segments the primitive radar measurements, and second fuses paired measurements to produce object reports used by a tracking system. The segmentation algorithm is developed by forming a hypothesis partition model for a set of consecutive range cells with detections, and then evaluating the joint likelihood function for each feasible partition of the cells into pairs or singletons. Simulation results that demonstrate the utility of the algorithm are provided using a modern missile tracking simulation environment.