摘要
作为人机交互的一种重要形式,手势识别在医疗康复领域已尤显重要。针对手势识别技术存在的不足,提出粒子群优化支持向量机(PSO-SVM)的多手势精确识别方法。首先,利用表面肌电信号采集仪采集16种手势所对应的表面肌电信号(SEMG);其次,分别从时域、频域和时频域提取所需要的SEMG特征;然后,采用主成分分析法(PCA)对数据特征进行降维;最后,使用PSO-SVM对降维后的数据特征进行分类识别。经过与传统支持向量机(SVM)分类以及遗传算法优化支持向量机分类(GA-SVM)相对比,本方法识别精度高、速度快,研究结果可为手势识别提供新的思路,为人体上肢动作判断和上肢康复机器人的研究提供参考。
As an important form of human-computer interaction,gesture recognition has become the focus of research in the field of medical rehabilitation.Aiming at the shortcomings of gesture recognition technology,a multi-gesture accurate recognition method based on particle swarm optimization support vector machine(PSO-SVM)is proposed.Firstly,surface electromyography(SEMG)signals corresponding to 16 kinds of human gestures are collected by surface electromyography signal acquisition instrument.Secondly,SEMG features are extracted from time domain,frequency domain and time-frequency domain respectively.Then,principal component analysis(PCA)is used to reduce the dimension of data features.Finally,according to the data characteristics,PSO-SVM is used for classification and recognition.Compared with traditional support vector machine(SVM)classification and genetic algorithm optimized support vector machine classification(GA-SVM),this method has high recognition accuracy and speed.The research results can provide a new idea for gesture recognition,and provide the reference for human upper limb motion judgment and the research of upper limb rehabilitation robot.
作者
王博
闫娟
杨慧斌
徐春波
吴晗
WANG Bo;YAN Juan;YANG Huibin;XU Chunbo;WU Han(School of Mechanical and Automotive Engineering,Shanghai University of Engineering Science,Shanghai 201620,China)
出处
《智能计算机与应用》
2023年第7期173-178,共6页
Intelligent Computer and Applications
关键词
手势识别
表面肌电信号
主成分分析
粒子群优化
支持向量机
gesture recognition
surface electromyography signal
principal component analysis
particle swarm optimization
support vector machine
