基于DT-SVM优化算法的人体姿态特征提取与识别研究
Esearch on Human Posture Feature Extraction and Recognition Based on DT-SVM Optimization Algorithm
丁晓慧 1周磊1
作者信息
- 1. 淮北理工学院电子与信息工程学院,安徽淮北 235000
- 折叠
摘要
为确定人体运动行为在空间环境中的表现情况,实现对姿态特征的准确定义,针对基于DT-SVM优化算法的人体姿态特征提取与识别方法展开研究.利用DT-SVM优化算法,推荐必要的姿态特征节点,确定人体运动行为所处空间平面.实施对姿态特征的梯度化处理,根据获取到的轮廓节点,计算夹角向量的具体数值,从而求解姿态特征提取与识别的数学表达式,完成基于DT-SVM优化算法的人体姿态特征提取与识别方法的设计.实验结果表明,上述方法的应用,可同时在X轴、Y轴、Z轴三个方向上,控制人体运动行为,使其偏向角数值均不超过12°,符合精准定义人体姿态特征的实际应用需求.
Abstract
In order to determine the performance of human motion behavior in the spatial envi-ronment and achieve accurate definition of posture features,research has been conducted on the extrac-tion and recognition methods of human posture features based on the DT-SVM optimization algo-rithm.Using the DT-SVM optimization algorithm,recommend necessary pose feature nodes to deter-mine the spatial plane where human motion behavior is located.Implement gradient processing of pos-ture features,calculate the specific value of the angle vector based on the obtained contour nodes,and solve the mathematical expression for posture feature extraction and recognition.Complete the design of a human posture feature extraction and recognition method based on DT-SVM optimization algo-rithm.The experimental results show that the application of the above method can simultaneously con-trol human motion behavior in the X axis,Y axis,and Z axis directions,so that the deviation angle val-ues do not exceed 12°,which meets the practical application requirements of accurately defining hu-man posture features.
关键词
DT-SVM优化算法/人体姿态/特征提取/特征识别/梯度化处理/轮廓节点/夹角向量/运动行为Key words
DT-SVM optimization algorithm/human posture/feature extraction/feature rec-ognition/gradient processing/contour nodes/angle vector/sports behavior引用本文复制引用
基金项目
安徽省高校自然科学研究重点项目(KJ2021A1246)
安徽省高校优秀人才支持项目基金项目(gxyq2022179)
出版年
2024
NETL
NSTL
万方数据