摘要
为了实现对工业机械手的手势控制,并针对单个Leap Motion传感器在使用时因手指或者手掌遮挡而导致识别率降低的缺点,提出了基于多Leap Motion传感器,用于人手数据采集的新型控制系统。该系统采用基于主成分分析法的数据融合技术对该多传感器数据进行数据融合以得到人手姿态完整的信息,再将该融合数据送入SVM多分类器进行手势姿态识别。实验结果表明该系统能准确地检测人手完整信息,避免了使用单个Leap Motion传感器因视觉干扰或者不可避免的遮挡导致手势姿态数据不准确的问题,提高了手势识别率,可以准确、有效地实现对机械手进行控制。
In order to achieve the gesture control of industrial manipulator, considering the decreased recognition rate caused by fingers and palm blocks, the paper proposed a new contrnl system based on multiple leap motion sensors for detecting hand data. The system adopted the data fusion technology based on principal component analysis to obtain complete hand gesture informarion, and then the data was put into SVM classifier to achieve gesture recognition. Experimental results show that the system can accurately detect manpower complete information, which avoids the problem of inaccurate gesture data caused by visual disturbances or inevitable occlusion by using a single leap motion sensor, improving gesture recognition rate, and it can achieve the control of manipulator accurately and efficiently.
作者
唐春晓
王志红
TANG Chun-xiao WANG Zhi-hong(College of Electronic and Information Engineering, Tianjin Polytechnic University, Tianjin 300387, China Laboratory of Intelligent Measurement and Control, Tianjin Polytechnic University, Tianjin 300387, China)
出处
《仪表技术与传感器》
CSCD
北大核心
2017年第7期62-66,共5页
Instrument Technique and Sensor
基金
国家自然科学基金项目(31271871)
天津市高等学校科技发展基金计划项目(20120609)
作者简介
唐春晓（1982-），讲师，博士，研究领域为光电检测技术。E—mail：tangyifei82@hotmail．com
王志红（1990-），硕士研究生，研究领域为信号检测处理。E-mail：zhwang2016@sina．com
