基于Kinect手势识别的智能家居系统研究与设计

发布时间：2018-02-25 00:33

本文关键词： Kinect 手势识别加权动态时间规整算法智能家居　出处：《辽宁科技大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着社会的不断发展,生活水平的不断提升,智能家居行业也随之迅猛发展。但是,各式各样的设备功能给用户带来交互体验不佳和操作繁琐的问题也更加明显。相比于传统人机交互系统使用鼠标、键盘等操作方式,更加简单、自然且人性化的手势识别技术在人机交互中正扮演着重要的角色。为了解决当前智能家居系统中存在的操作繁琐,用户不能通过自然的方式与机器进行交互而导致较差的用户体验等问题,本文将基于Kinect的手势识别技术融入至设计的智能家居人机交互系统,其使用Kinect获取人体图像的深度信息与骨骼数据。研究了基于相对位置算法、隐马尔可夫模型算法、动态时间规整算法和加权动态时间规整的模板匹配手势识别算法,并采用基于加权动态时间规整的模板匹配手势识别算法提取预先定义的手势动作。实际的手势实验结果表明:使用该算法实现手势识别是可行且有效的,其最佳识别位置是在Kinect的正前方2到2.5米,识别准确率达到98%左右。此外,研究了智能家居系统的管理流程与系统硬件模块,设计了硬件电路与相应的模块软件,并采用Zig Bee技术实现了手势控制指令的传输。本文工作主要集中在研究了基于Kinect的四种手势识别算法,最终采用基于加权动态时间规整的模板匹配手势识别算法提取手势动作,使用C#.NET语言编程实现了基于Kinect手势识别的人机交互系统;并完成了智能家居系统硬件模块电路设计及相应软件功能实现;熟悉并应用RS485、I2C总线、CAN总线及Zig Bee协议并应用到控制系统中。
[Abstract]:With the development of society and the improvement of living standard, the smart home industry is also developing rapidly. All kinds of device functions bring users a more obvious problem of poor interaction experience and cumbersome operation. Compared with the traditional man-machine interaction system using mouse, keyboard and other operating methods, it is much simpler. The natural and humanized gesture recognition technology is playing an important role in human-computer interaction. The user can not interact with the machine in a natural way, which leads to poor user experience. In this paper, the gesture recognition technology based on Kinect is integrated into the intelligent home human-computer interaction system. Kinect is used to obtain depth information and bone data of human body image. Based on relative position algorithm, hidden Markov model algorithm, dynamic time warping algorithm and weighted dynamic time warping algorithm, template matching gesture recognition algorithm is studied. A template matching gesture recognition algorithm based on weighted dynamic time regularization is used to extract predefined gesture actions. The experimental results show that the algorithm is feasible and effective. The best recognition position is 2 to 2. 5 meters in front of Kinect, and the recognition accuracy is about 98%. In addition, the management flow and system hardware module of smart home system are studied, and the hardware circuit and corresponding module software are designed. This paper mainly focuses on the research of four gesture recognition algorithms based on Kinect, and finally uses the template matching gesture recognition algorithm based on weighted dynamic time regularization to extract gesture actions. The human-computer interaction system based on Kinect gesture recognition is realized by using C#.NET language, and the hardware module circuit design and corresponding software function realization of smart home system are completed. Familiar with can bus and Zig Bee protocol and apply to control system.
【学位授予单位】：辽宁科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;TU855

【参考文献】