用于低功耗人体运动姿态估计系统的数据压缩感知方法  

Data Compressed Sensing Method for Low-Power Human Motion Pose Estimation Systems

作　　者：王功 赵柏山[1] WANG Gong;ZHAO Baishan(School of Information Science and Engineering,Shenyang University of Technology,Shenyang 110870,China)

机构地区：[1]沈阳工业大学信息科学与工程学院,沈阳110870

出　　处：《微处理机》2025年第2期52-55,共4页Microprocessors

摘　　要：本研究旨在降低惯性传感器式人体运动姿态估计系统中信息采集节点的功耗。针对传统方法功耗高的问题,提出一种低功耗运动信息数据处理方法,基于优化的二值托普利兹测量矩阵对加速度数据进行低维投影,并结合近似消息传递的广义稀疏贝叶斯估计重构模型,设计非稀疏加速度数据重构算法,实现高效低功耗的数据获取。实验结果表明,该方法在保证信号重构精度的同时,显著缩短了重构时间,优于传统方法,并有助于提高数据处理效率,简化硬件设计,有效降低采集节点的功耗,为低功耗运动姿态估计系统提供了技术支持。This study aims to reduce the power consumption of data acquisition nodes in inertial sensor-based human motion pose estimation systems.To address the high power consumption of traditional methods,a low-power motion data processing method is proposed.The method employs an optimized binary Toeplitz measurement matrix for low-dimensional projection of acceleration data and integrates a generalized sparse Bayesian estimation reconstruction model based on approximate message passing to design a non-sparse acceleration data reconstruction algorithm,enabling efficient and low-power data acquisition.Experimental results demonstrate that the method significantly reduces reconstruction time while maintaining signal reconstruction accuracy,outperforming traditional approaches.It also contributes to improved data processing efficiency,simplified hardware design,and effective reduction in the power consumption of acquisition nodes,providing technical support for low-power motion pose estimation systems.

关 键 词：压缩感知 低功耗 运动信息采集 加速度数据 人体运动姿态估计 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]