A multitarget tracking algorithm based on radar and infrared sensor data fusion

In this article we proposed a sequential GM-PHD Filter (Gaussian Mixture Probability Hypothesis Density Filter). We extended the existing GM-PHD, which is fit for single-sensor multi-target tracking, to a multi-sensor multi-target case, and applied it in radar and infrared sensor multi-target tracking problem. Considering the infrared sensor is more accurate in angle information measuring than radar but can not get the distance information, we first estimate the observations collected by radar with the GM-PHD filter, we set the estimate results of the radar as the predicted values of the infrared sensor, then the multi-target states are updated at the fusion center with the infrared sensor observations. Simulation shows this algorithm can get more accurate estimate results than single radar multi-target tracking with GM-PHD filter, and it is robust under the clutter and omission.