基于BEMD-MTS算法的肢体动作轮廓智能捕捉方法  

Intelligent Capturing Method of Limb Motion Contour Based on BEMD-MTS Algorithm

作　　者：马璿 张会庆 MA Xuan;ZHANG Hui-qing(Department of Information Engineering,Tibet University for Nationalities,Xianyang Shaanxi 712000,China)

机构地区：[1]西藏民族大学信息工程学院,陕西咸阳712000

出　　处：《计算机仿真》2023年第10期224-227,232,共5页Computer Simulation

基　　金：西藏民族大学“青年学人”资助项目(21MDX02);西藏自治区自然科学基金项目(XZ202001ZR0060G);西藏民族大学2021年改革发展专项一流课程(XZ2021);国家民委高等教育教学改革研究项目(21077)。

摘　　要：人体肢体动作多变,具有一定复杂性,且肢体细节较多,动作轮廓捕捉难度较高。提出基于BEMD-MTS算法的肢体动作轮廓智能捕捉方法。建立人体姿态模型,利用BEMD-MTS算法提取人体姿态特征,确定人体姿态行为类型;构建肢体动作轮廓捕捉模型,并将置信度检测方法嵌入模型中,结合轮廓形态学重建,实现基于姿态图像特征提取的肢体动作轮廓捕捉。实验结果表明,研究方法对200幅图像轮廓捕捉时容错率可达5%,平均吻合度指标可稳定在98%以上,且主观测试结果证明了上述方法能够精准捕捉肢体动作轮廓,且无遗漏特征和细节。In this paper,an intelligent method of capturing body movement contour based on the BEMD-MTS algorithm was proposed.Firstly,a model of human posture was built,and then some features of human posture were extracted by the BEMD-MTS algorithm so that the types of human posture behaviors could be determined.Moreover,a model for capturing body movement contour was constructed.Meanwhile,the method of confidence detection was embedded in the model.Combined with the morphological reconstruction of contour,the capture based on pose image feature extraction was achieved.Experimental results show that the fault-tolerance rate of the proposed method can reach 5%after capturing the contours of 200 images,and the average index of the goodness of fit can be stable at more than 98%.In the meanwhile,subjective test results prove that this method can accurately capture the contour of body action without missing features and details.

关 键 词：姿态图像 特征提取 肢体动作轮廓 捕捉方法 形态学重建 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]