摘要
针对已有手势识别系统准确度不高和响应速度慢的问题,设计了一款基于RBF网络的手势识别装置。该装置以STM32单片机为控制中心,使用电容传感器FDC2214进行容值数据采集;利用RBF网络对样本数据进行训练并抽取权值,将权值带入程序,输入乘权值得到输出,实现装置对手势的预测;按键实现装置判决模式与训练模式的切换;最终由TFTLCD显示容值数据和手势结果。实验测试表明:手势识别装置能够对任意测试人员进行多种手势的识别,并达到平均96.25%的识别准确率和平均0.6735s的系统响应速度,从而验证了本装置的有效性。
For the problem of the existing gesture recognition system has low accuracy and slow re-sponse,a gesture recognition device based on RBF network is designed.The device take STM32MCU as the core and use capacitive sensor FDC2214 to collect the capacitive value.The sampled data is trained by RBF network weights,which are extracted from the training,is applied in the program.The data of input multiplied by weights is the data of output and the result hit the forecast.The key module realize modes switching and selection.The capacitive value and the result of gestures are displayed by the screen of TFTLCD finally.The experimental results show that the device of gesture recognition can recognize many kinds of gestures for every tester.The average recognition accuracy is 96.25%and the system response speed is lower than 0.8s each time,which further verifies the validity of the device.
作者
马杰
江亚峰
强东鑫
徐志佑
袁明新
MA Jie;JIANG Ya-feng;QIANG Dong-xin;XU Zhi-you;YUAN Ming-xin(School of Mechanical&Power Engineering,Jiangsu University of Science and Technology,Zhangjiagang 215600,Jiangsu Province,China;Zhangjiagang Industrial Technology Research Institute,Zhangjiagang 215600,Jiangsu Province,China)
出处
《信息技术》
2019年第12期24-28,共5页
Information Technology
基金
国家自然科学基金项目(61105071)
江苏科技大学苏州理工学院创新竞赛项目(SZLGCJ201709)
江苏科技大学苏州理工学院青年教师科研启动专项基金(SZLGQN2018002)
