Knowledge-Based Plot Agglomeration for Multiple Extended Targets in Partial Occlusion Conditions

Extended target models are common in modern high-resolution moving target detection(MTD) radars. However, they present a challenge for many tracking algorithms. Plot agglomeration is a virtual step before multiple extended target tracking. A knowledge-based plot agglomeration is proposed for multiple extended marine targets in mutual occlusion conditions. The radial velocity is used as an important feature to distinguish closely targets. The velocity is extracted from the CFAR detector of the MTD procedure. A weighted distance between detected pixels from the MTD procedure is designed to incorporate the radial velocity information. Then a connection-rule-based two-dimensional clustering method is used to efficiently perform the plot agglomeration task. Experiment results of inner river radar data processing validate the new method.