室外人体脚步声事件及环境联合识别  

Outdoor Human Footsteps Event and Environment Joint Recognition

作　　者：徐峰 李平[1] XU Feng;LI Ping(College of Information Science and Engineering,Huaqiao University,Xiamen 361021,China)

机构地区：[1]华侨大学信息科学与工程学院,福建厦门361021

出　　处：《华侨大学学报（自然科学版）》2021年第5期676-683,共8页Journal of Huaqiao University(Natural Science)

基　　金：国家自然科学基金资助项目(61603144);福建省自然科学基金资助项目(2018J01095);福建省高校产学研合作科技重大项目(2013H6016);华侨大学中青年教师科技创新资助计划项目(ZQN-PY509)。

摘　　要：为了实现室外人体脚步声事件及环境联合识别,首先,设计一个复杂相似环境下的人体跑动和行走数据集,提出一种交叉双脚步声的分割方案,对连续脚步声信号进行交叉分割;然后,从事件和环境的角度分别提取特征,并从任务平衡的角度设计两种融合特征;最后,采用3种深度学习模型对任务进行精确地识别.结果表明:文中方法简化平衡了任务,使室外人体脚步声事件及环境联合识别的多任务设计不需要复杂模型就能实现精确识别.In order to realize the joint recognition of outdoor human footsteps events and environment,firstly,a human running and walking data set in a complex and similar environment was designed,and a cross double footsteps segmentation scheme was proposed to cross segment the continuous footsteps signals.Then,features were extracted from the perspectives of events and environment,and two fusion features were designed from the perspective of task balance.Finally,three deep learning models were used to identify the task accurately.The results showed that the proposed method simplified and balanced the tasks,and the multitask design of joint identification of outdoor human footsteps events and environment could realize accurate identification without complicated models.

关 键 词：交叉双脚步声 联合识别 多任务学习 融合特征 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]