人体上肢运动姿态多点视觉自动跟踪仿真
Multi-Point Visual Automatic Tracking and Imitation of Human Upper Limb Movement Posture
刘硕 1荆瑞俊2
作者信息
- 1. 山西大学,山西 太原 030006
- 2. 山西农业大学 软件学院,山西 太原 030031
- 折叠
摘要
由于视觉传感器本身的噪声以及运动模糊等原因,使得采集的运动上肢姿态数据存在噪声和误差,导致多点视觉自动跟踪精度较低.为此,提出人体运动上肢姿态多点视觉自动跟踪方法仿真.布设多维传感器对姿态多点采集,建立传感器加速度误差模型,获取误差修正目标函数,利用蚁群算法(Ant Colony Optimization,ACO)寻优,修正传感器采集过程中的误差;采用卡尔曼滤波算法(Kalman Filtering Algorithm,Kalman),实现人体运动上肢姿态的多点视觉自动跟踪.实验结果表明,所提方法能够在保障跟踪稳定性的用时,提高姿态跟踪的精度和效率,跟踪耗时仅为 10ms左右.
Abstract
The noise inherent in visual sensors and motion blur results in noisy and erroneous data when captu-ring the upper limb movement poses,as well as lower accuracy in multi-point visual automatic tracking.Therefore,a multi-point visual automatic tracking method for human upper limb movement postures was proposed.Firstly,multi-dimensional sensors were deployed to collect the postures from multiple points.Then,a sensor acceleration error model was built to obtain an objective function of error correction.Moreover,the Ant Colony Optimization(ACO)al-gorithm was utilized for optimization,thus correcting the errors in the acquisition process of sensors.Finally,the Kal-man Filtering Algorithm(Kalman)was employed to achieve the multi-point visual automatic tracking of human upper limb movement poses.Experimental results show that the proposed method can improve the accuracy and efficiency of pose tracking while ensuring tracking stability.The tracking time is only around 10ms.
关键词
多维传感器/蚁群算法/误差修正/卡尔曼滤波/姿态跟踪Key words
Multi-dimensional sensor/Ant colony algorithm/Error correction/Kalman filtering/Pose tracking引用本文复制引用
基金项目
山西省科技厅重点项目(202204031401003)
出版年
2024
NETL
NSTL
万方数据