摘要随着如今计算机中的计算机视觉领域部分的快速发展和用户需求的不断提高，关于人体目标的检测与跟踪技术的研究显的越发的重要，而人体目标检测与识别这方面的研究在人体目标检测与跟踪技术研究中占据着十分重要的地位。众所周知，随着Kinect的出现，给人体目标检测识别技术带来了新的希望，由于Kinect在传统摄像机的基础上还可以轻松准确的获得深度图形，并且在识别过程中不受周围光线与被检测人体的服装等影响，使得人体识别与检查的精确度得到了有效的改善，本篇论文主要是通过使用Kinect摄像机来对人体进行深度图像数据，彩色图像数据以及骨骼数据的提取，之后在对每一种数据流进行相应的处理，在提取彩色数据流，深度数据流和骨骼数据流的时候，我在事件方法和“拉”方法这两种方法中选择了事件方法，为了是使得代码简化和减少异常。在处理骨骼数据流的时候，选择对数据进行从骨骼坐标系到深度坐标系的转换。最终可以实现当人体处在距离Kinect1.2米到3,5米范围内（若人体在有效范围外可能无法识别）的场景中时，可以被Kinect检测识别出来，并在界面上显示出来。28576
关键词 Kinect人体检测 图像数据获取 骨骼采集 事件方法 坐标转换
毕业论文设计说明书外文摘要
Title Design and Implementation of Human Detection System Based on Kinect
Abstract
Recently,with the rapid development of the field of computer vision and rising user demand,the research on detection and tracking of human targets become more and more important,the technology of detection and recognition play an important role in it.As we all know, with the emergence of Kinect,the technology of detecting human targets has brought new hope.Because of the Kinect can easily obtain accurate depth graphics besides the information of traditional cameras, and the light of context and human’s clothes do not have any influence in the identification process,it make the accurate identification and examination of the body has been effectively improved. In this paper,we extract the depth of image data, the color image data as well as dates of human bones by using Kinect camera,and then have some treatments on the every data stream.At last, when the body is at a distance of Kinect1.2 meters to 3,5 meters (the body may not be recognized outside the valid range) scene, the body can be detected and displayed on the screen.
Among the process of Extracting the data flow of depth and the data flow of bones,we have chosen one method from the method of event and “pull” in order to simplify the code and reduce abnormal.
While dealing with the skeletal data stream, our data have been converted from the bone coordinate system to the coordinate system of depth.
Keywords Kinect image acquisition skeleton collection method of enent
coordinate transformation
目 次
1 绪论 1
1.1研究背景及意义 1
1.2 国内外研究现状以及Kinect的实际应用 2
1.3技术难点 3
1.3.1 获取数据方法的选择与使用 3
1.3.2 如何获取关节进行跟踪，并且在界面以骨架模式显示出来 3
1.4 论文结构安排 3
2 Kinect传感器 4
2.1自然用户的接口(NUI) 4
2.2 Kinect传感器的硬件组成 4
2.3 Kinect SDK介绍 5 基于Kinect的人体检测系统设计与实现:http://www.youerw.com/jisuanji/lunwen_23468.html