Gaussian Process Based Probabilistic Data Association Filtering Extended Target Tracking Method

Abstract

The invention proposes a Gaussian process-based probabilistic data association filtering extended target tracking method. This method firstly proposes a joint tracking gate based on Gaussian process to select valid measurements in each moment of measurement, summarize the situation of each valid measurement source, and obtain relevant events about the measurement sources. Second, based on the Kalman filter, the target state estimation corresponding to the relevant event is obtained under the condition of the relevant event, the effective measurement at the current moment and the approximate sufficient statistics. Third, based on the Bayesian probability theory, the weight of relevant events is obtained on the condition of valid measurements at all times. Finally, combined with the overall probability formula, the conditional estimates and corresponding weights of all relevant events are summarized to obtain fused state estimates and covariance estimates.

Description

technical field [0001] The invention belongs to the technical field of target detection and tracking, and relates to a Gaussian Process-Probability Data Association (GP_PDA_ETT) extended target tracking method based on Gaussian process-probability data association filtering. Background technique [0002] Extended Target Tracking (ETT) technology means that with the rapid development of sensor technology, high-resolution sensors can provide multiple measurements for multiple observation points on moving targets. At this time, the target is no longer a point target, but is It is called an extended target. Through multiple measurements received by high-resolution sensors, the shape and motion state of the extended target can be tracked and estimated simultaneously. Compared with the traditional point target tracking, the extended target tracking can not only estimate the position, velocity and heading angle of the target, but also can estimate the shape of the extended target, ...

AI Technical Summary

Problems solved by technology

The multi-ellipse random matrix method combines multiple ellipses to estimate irregular shape targets, thus producing more accurate shape estimation results, but this method cannot estimate the shape of extended targets when the measurement source is uncertain; the probability hypothesis density method (PHD) considers In order to solve the situation where the measurement source is uncertain, the target shape can be estimated in the unknown clutter environment, but this method cannot accurately estimate the target shape

[0004] The method proposed in the above literature is highly complex and time-consuming, and it is difficult to accurately estimate the shape of the target in the clutter environment

Images