Open AccessRange–Doppler‐based centralised framework for human target tracking in multistatic radar
Author(s) -
Zhang Jun,
Jin Tian,
He Yuan,
Qiu Lei,
Zhou Zhimin
Publication year - 2017
Publication title -
iet radar, sonar and navigation
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.489
H-Index - 82
eISSN - 1751-8792
pISSN - 1751-8784
DOI - 10.1049/iet-rsn.2016.0140
Subject(s) - computer science , clutter , artificial intelligence , tracking (education) , computer vision , range (aeronautics) , constant false alarm rate , radar , object detection , detector , doppler effect , tracking system , radar tracker , false alarm , statistical power , pattern recognition (psychology) , mathematics , statistics , engineering , kalman filter , physics , psychology , telecommunications , pedagogy , astronomy , aerospace engineering
Human target detection and tracking have great potential in surveillance, rescue and security applications. Traditional human detection and tracking are performed on the range profile. If targets are close or overlapped in range, it is difficult to distinguish these targets. As velocity provides another aspect to distinguish targets, a novel framework is proposed for human tracking using both range and velocity information. The tracking framework consists of six steps, including clutter reduction, range–Doppler (RD) calculation, target detection, measurement estimation, target localisation and target tracking. Primary attention is devoted to the middle four steps. The calculation process of the RD image is described in detail. The ordered statistics constant false alarm rate detector is extended to a two‐dimensional scenario for the RD target detection. An efficient approach is given for automatic measurement estimation. A minimum root‐mean‐square error pruning algorithm is proposed for multi‐target localisation. As the algorithm combines both range and velocity information for measurement association, it clearly shows a lower wrong association probability than the method using range information only. The effectiveness of the proposed tracking framework is evaluated by the experimental data in the foliage‐penetration environment.