基于Kinect的人体运动姿态捕捉和识别技术研究
本论文探讨了如何使用Kinect来进行手势识别的问题。主要通过使用Kinect的三种摄像机所识别到的人体骨骼的空间数据来控制鼠标。使得用户得以通过手部的移动来控制鼠标，并通过手掌的拍击来实现鼠标的左击，右击。从而实现简单的体感式人机交互。这是一种先进的人机交互方式，不难想象有了Kinect之后的电脑操控方式。键盘或许还在，但鼠标将逐渐消失，因为只要在红外摄像头所能捕捉的有效范围内，手掌甚至手指的任意移动都能精确地进行位置确认。而本文正是基于此的一次实践。10402
关键词 Kinect 体感设备 人机交互 姿态识别 动作捕捉
毕业设计说明书（论文）外文摘要
Title The research on Kinect based human motion capture and recognition.
Abstract
Kinect is a motion sensing input device released by Microsoft at November, 2011. In fact it’s a 3D motion camera can be used for motion capture, voice input and many other functions. We can stand in front of a Kinect and make various moves to interact with the computer.
This paper focus on how to use Kinect to recognize gesture. Mainly through the use of three camera of Kinect to identify human skeleton position data to control the mouse. The user is able to use hand to move the mouse and use hands to achieve left-click and right-click. In order to achieve simple gesture recognition type human-machine interaction. This is an advanced way of human-computer interaction; it is not difficult to imagine a control mode by using Kinect take place of the normal computer control mode. The keyboard may also exists, but the mouse will gradually disappear, because as long as the IR camera can capture the effective range, it can confirm the position of hand or even the fingers accurately. This article is about a practice about this fact.
Keywords Kinect motion sensing device human-computer interaction gesture recognition motion capture
目 次
1 绪论 1
1.1 Kinect 实物图及其组件 1
1.2基于Kinect的体感动作识别 2
1.2.1光编码（Light Coding） 2
1.2.2骨架追踪系统 2
1.3 Kinect的开发工具 3
1.3.1 Kinect for Windows SDK 3
1.3.2 Open CV 4
2 运动姿态的识别方法 5
2.1 什么是手势 5
2.2自然交互界面(NUI) 7
2.3实现手势识别的方法 10
2.3.1基于算法的手势识别 10
2.3.2基于神经网络的手势识别 11
2.3.3基于样本的识别 12
2.4 特征提取方法 13
2.4.1 人体区域检测 13
2.4.2人体表征 14
3 使用Kinect的手势识别 15
3.1 研究的内容与要求 15
3.2 识别方法 15
4 实验结果及分析 17
4.1对人体骨骼及操作者的识别 17
4.1.1一名操作者的情况 17
4.1.2 识别出两名及两名以上人物的情况 18
4.2实现鼠标控制的算法 19
上一篇：C++电子邮件客户端的设计与实现