Track correlation method and device

Abstract

The invention provides a track correlation method and apparatus, and belongs to the sensor data processing technical field. The method comprises the following steps of for any two sensors, acquiring a track point set corresponding to each sensor in the same time period separately; establishing a t distribution mixture model corresponding to a sample point set based on the sample point set and a floating point set; determining an optimal displacement vector of nonrigid registration between the floating point set and sample point set based on the t distribution mixture model; and determining displacement results of the floating point set after moving according to an optimal displacement vector, and serving the displacement results as track correlation results between the floating point set and the sample point set. Since a heavy-tailed t distribution mixture model which is more robust to abnormal points is established, accurate correlation rate of the track correlation, soundness of the track correlation and robustness can be improved when the sensor has system errors, false alarm, missing alarm and multiple sensors have overlapping monitoring areas.

Description

technical field [0001] The present invention relates to the field of sensor data processing, and more specifically, to a track association method and device. Background technique [0002] In the distributed multi-sensor fusion system, track correlation is the key link of information fusion, and it is the process of judging whether two tracks from different sensors originate from the same target. Wherein, the track refers to the trajectory of the monitoring target, and the monitoring target may be a ship or an aircraft. The sensor can be a radar or an infrared sensor, etc., and the distributed sensor is a multi-sensor data processing method established with a distributed computer as a reference. The track association of distributed multi-sensor information location layer aims to make a judgment on whether tracks from multiple local nodes correspond to the same target. For example, for two sensors, if each sensor monitors a moving target at a series of time points, and each ...

AI Technical Summary

Problems solved by technology

When the sensors have false alarms, missing alarms, or partially overlapping monitoring areas between multiple sensors, the observation targets between the sensors are not completely consistent, resulting in the inability to perform one-to-one matching and alignment of the target tracks, and the difficulty of track correlation increases

in addition, systematic errors can skew target measurement positions, making track correlation more difficult

Images