Using More Information to Determine Trajectory Association: An Improved Method Based on Weighted Cascade Hausdorff Distance

Abstract

The problem of ship trajectory association is well recognized in the overlapping area cross multiple sensors. In this paper, an improved method based on weighted cascade Hausdorff distance, with a variable time sliding window, is proposed. This method can effectively solve the problem of ship target fusion in the field of border and coastal defense, thereby providing a basis for the next step of ship monitoring and management. The similarities of tracks from different radars are determined by the proposed method, which combines the information of both ship position and motion characteristics. Several experiments are designed, with both real and simulated track data as input, to evaluate the effectiveness of the proposed method. The results showed that the method, due to the configurable parameters of time sliding window, has good universality to meet the needs of different radar frequency. The best performance among different configuration settings has the better advantage of 10%-20% improvement compared with the traditional Hausdorff distance model with fixed time window.