Multi-person tracking method based on millimeter-wave radar

Abstract

The invention relates to the technical field of millimeter-wave radar signal processing, and discloses a multi-person tracking method based on millimeter-wave radar, which is used for improving the multi-target tracking precision and tracking speed of the millimeter-wave radar in practical application. The method comprises the following steps: firstly, generating point cloud data from a millimeter-wave radar data cube, and then classifying tracking difficulties of tracks and condensation points by adopting a track clustering method; adopting a multi-mode tracking method to track tracks with different tracking difficulty degrees by adopting different tracking methods; fusing results of different tracking methods, and carrying out multi-modal trajectory starting and ending according to tracking results. The method is suitable for multi-person tracking.

Description

technical field [0001] The invention relates to the technical field of millimeter wave radar signal processing, in particular to a method for tracking multiple people based on millimeter wave radar. Background technique [0002] Multiple Object Tracking (MOT) technology is a very important basic problem in the fields of computer vision, image processing, radar signal processing, and multi-source data fusion. Multi-person tracking is an important special case of multi-target tracking. Multi-person tracking can calculate the individual movement trajectory and group number of people in the scene. It is an important technology to realize scene perception. It has important applications in human-computer interaction and other fields. [0003] At present, multi-person tracking can be realized through sensors such as optical cameras, infrared cameras, laser radars, and millimeter-wave radars. The multi-person target tracking technology based on optical camera mostly uses deep lear...

AI Technical Summary

Problems solved by technology

Due to the high complexity of MHT and RFS, there are few applications in low-cost miniaturized mmWave radars

[0005] In the existing multi-person target tracking and counting based on millimeter-wave radar, there are the following problems: using a simple global nearest neighbor data association algorithm (Global Nearest Neighbor, GNN), when the targets are relatively dense and there are occlusions between them When using a single tracking algorithm for all trajectories and measurements, it cannot effectively distinguish between trajectories and measurements that are easy to handle and those that are difficult to handle, and cannot solve the conflict between algorithm effectiveness and real-time performance Problem; Intuitive trajectory initiation and termination management strategies cannot handle the emergence, merging and disappearance of trajectories in complex situations

Images