基于动态手势识别算法的VSTi系统研究  

Research on VSTi system based on dynamic gesture recognition algorithm

作　　者：任姗 REN Shan(Xi’an Siyuan University,Xi’an 710038,China)

机构地区：[1]西安思源学院,西安710038

出　　处：《自动化与仪器仪表》2025年第1期214-217,222,共5页Automation & Instrumentation

基　　金：西安思源学院校级科研项目(一般项目)《教育数字化视域下民办高校公共艺术教育的发展现状及路径研究-以西安思源学院为例》(XASYYB-A2207)。

摘　　要：为了降低音乐教育中部分乐器教育的设备普及难度和提升学习趣味性,研究提出了一种基于动态手势识别算法的虚拟弹奏技术。过程中通过体感周边外设获取使用者的骨骼信息,经过设备内置软件进行实时处理后输出关节运动情况,在与标准动作样本进行对比后输出动作识别结果,再建立虚拟乐器模型,利用七参数模型完成设备之间的坐标系转换。实验结果表明,研究方法在动作识别准确率测试中,在复杂背景中动作长度增加到250 ms时动作识别准确率保持在98.4%以上;在操作响应时延分析中,研究方法响应时延超过100 ms的操作数量少于总数的1%。以上结果说明研究方法具有良好的运行效率和精确性。In order to reduce the difficulty of popularizing certain instruments in music education and enhance learning interest,a virtual playing technology based on dynamic gesture recognition algorithm is proposed.During the process,the user’s bone information is obtained through peripheral devices that sense the body.After real-time processing by the device’s built-in software,the joint motion situation is output.After comparing with standard action samples,the action recognition results are output.Then,a virtual instrument model is established,and the coordinate system conversion between devices is completed using a seven parameter model.The experimental results show that the research method maintains an action recognition accuracy of over 98.4%when the action length increases to 250ms in complex backgrounds during action recognition accuracy testing;In the analysis of operation response delay,the number of operations with response delay exceeding 100ms in the research method is less than 1%of the total number.The above results indicate that the research method has good operational efficiency and accuracy.

关 键 词：虚拟乐器 手势识别 KINECT 欧式距离 七参数模型 

分 类 号：G434[文化科学—教育学] TP39[文化科学—教育技术学]