摘要
隔墙人体姿态重建和行为识别在智能安防和虚拟现实等领域具有广泛应用前景。然而,现有隔墙人体感知方法通常忽视了对4D时空特征的建模以及墙体对信号的影响,针对这些问题,该文创新性地提出了一种基于4D成像雷达的隔墙人体感知新架构。首先,基于时空分离的分步策略,该文设计了ST~2W-AP时空融合网络,解决了由于主流深度学习库缺少4D卷积而无法充分利用多帧3D体素时空域信息的问题,实现了保留3D空域信息的同时利用长序时域信息,大幅提升姿态估计任务和行为识别任务的性能。此外,为抑制墙体对信号的干扰,该文利用深度学习强大的拟合性能和并行输出的特点设计了深度回波域补偿器,降低了传统墙体补偿方法的计算开销。大量的实验结果表明,相比于现有最佳方法,ST~2W-AP将平均关节位置误差降低了33.57%,并且将行为识别的F1分数提高了0.51%。
Through-wall human pose reconstruction and behavior recognition have enormous potential in fields like intelligent security and virtual reality.However,existing methods for through-wall human sensing often fail to adequately model four-Dimensional(4D)spatiotemporal features and overlook the influence of walls on signal quality.To address these issues,this study proposes an innovative architecture for through-wall human sensing using a 4D imaging radar.The core of this approach is the ST2W-AP fusion network,which is designed using a stepwise spatiotemporal separation strategy.This network overcomes the limitations of mainstream deep learning libraries that currently lack 4D convolution capabilities,which hinders the effective use of multiframe three-Dimensional(3D)voxel spatiotemporal domain information.By preserving 3D spatial information and using long-sequence temporal information,the proposed ST2W-AP network considerably enhances the pose estimation and behavior recognition performance.Additionally,to address the influence of walls on signal quality,this paper introduces a deep echo domain compensator that leverages the powerful fitting performance and parallel output characteristics of deep learning,thereby reducing the computational overhead of traditional wall compensation methods.Extensive experimental results demonstrate that compared with the best existing methods,the ST2W-AP network reduces the average joint position error by 33.57%and improves the F1 score for behavior recognition by 0.51%.
作者
张锐
龚汉钦
宋瑞源
李亚东
卢智
张东恒
胡洋
陈彦
ZHANG Rui;GONG Hanqin;SONG Ruiyuan;LI Yadong;LU Zhi;ZHANG Dongheng;HU Yang;CHEN Yan(School of Cyber Science and Technology,University of Science and Technology of China,Hefei 230026,China)
出处
《雷达学报(中英文)》
北大核心
2025年第1期44-61,共18页
Journal of Radars
基金
国家自然科学基金(62172381,62201542)。
关键词
穿墙
人体姿态估计
行为识别
射频感知
深度学习
Through-wall
Human pose estimation
Activity recognition
RF sensing
Deep learning
