Performance Assessment of Tracking Systems

This paper presents a new formalized procedure to assess the performance of automatic target tracking systems. The procedure involves comparing data output by a tracking system under test with track data in a truth file. The comparison of the data sources uses a total of 15 tracker performance metrics. These cover the categories of track initiation, track maintenance, track error and false tracks. The performance for each metric is expressed as a probability. An overall performance measure comes from using a weighted combination of these probabilities. The weighting relates to the metric’s importance. This approach is embodied in a Tracker Assessment Tool (TAT) and the processing steps in the TAT are reviewed with an example. A method to simplify the creation of truth data from recorded data is also described.