【作者】 贡国栋；

【导师】 余雪丽；

【作者基本信息】 太原理工大学 ， 计算机系统结构， 2010， 硕士

【摘要】 计算机的发明给人类社会带来了巨大的影响。随着硬件方面计算速度的提高与存储容量的扩大,各种应用软件与管理系统不断地被研发出来,人与计算机的交互方式也发生着深层次的变革。图形用户界面在极大地方便普通用户使用计算机的同时,也带了诸多的限制,使计算机仍然难以真正融入人们的工作和生活。人们希望计算机能听、会看、可以说话,甚至比人做得更好,并且能够进行实时处理,一种全新的交互方式成为当前计算机技术发展的迫切需要。计算机系统的拟人化,计算机的微型化、随身化和嵌入化,将是计算机两个重要的应用趋势,而人机交互技术是其中的瓶颈,以人为中心、自然、高效将是发展新一代人机交互主要的努力方向。普适计算要实现的目标之一,就是让计算机变得不可见、不需要提供有意识的操作,同时能够向人们提供无所不在的计算和信息服务。在对基于计算机视觉、语音识别、手势输入、感觉反馈等新的交互技术的研究中,人脸识别因其在身份验证、档案管理、视频会议等方面的巨大应用前景而越来越成为当前人机交互领域的研究热点。而视线跟踪技术由于其可代替键盘、鼠标实现输入和移动的功能,对行动不便的人群和飞行员等有较大的吸引力,也吸引了心理学家、交互技术专家的广泛关注。本文作者查阅了近年来国内外大量关于人机交互和计算机视觉的学术论文及文献,对人机交互的基本设计方法和普适计算的框架进行了探讨,阐述了人脸检测与识别的一般过程和方法,并且对其具体实现作了较为深入的讨论,提出一种用于扩展第三方软件功能的框架协议,将基于计算机视觉的人机交互无缝嵌入到现有系统中。本文的研究工作主要包括以下几个方面:(1)概括了本课题的研究意义和应用前景,回顾了人机交互、计算机视觉的国内外研究现状,并总结了今后的发展趋势;(2)对普适计算的交互框架作了简要分析,讨论了交互过程的两种设计方法,并结合实例讨论了在实际项目中的运用,对比了两种方法的不同设计过程;(3)重点阐述了人脸检测与识别的一般处理方法,对当前较为流行的开源计算机视觉库的实现算法进行了详细讨论,并借此完成人脸识别系统的开发;(4)提出了将基于计算机视觉的人机交互实现为一个框架协议程序,为第三方应用软件提供功能上的扩展,只需定义交互行为的脚本描述文件,即可实现对由摄像头捕捉到的眼部动作的响应。更多还原

【Abstract】 The invention of the computer to the human society has a huge influence. With the increase in hardware computing speed and storage capacity expansion, a variety of application software and management system is constantly developed and the manner of human-computer interaction is also undergoing profound changes. Graphical user interface greatly facilitate the use of ordinary computer users, but it still has so many restrictions that the computer is difficult to really integrate into people’s work and life. People want the computer is able to listen, to see, to speak, or do these work even better than people, and can carryout real-time processing. A new interactive way becomes an urgent need for the development of computer technology. Personification of computer systems, computer miniaturization, portable technology and embedded technology will be two important trends for using computer, but human-computer interaction technology is one of the bottlenecks. Human-centered, natural manner will be the main direction for the new generation of efficient human-computer interaction.One goal of the pervasive computing is to let the computer become invisible, with not providing a sense of the operation, while able to provide people with ubiquitous computing and information services. In the research of computer vision, speech recognition, gesture input and sensory feedback of the new interactive technology, the face recognition because of it’s has great future in the area of authentication, file management and video conferencing, is increasingly becoming the focus in the field of research of human computer interaction. The gaze tracking can be replaced has attracted psychologists, interactive technical experts attention, because it can take the place of keyboard and mouse to help disabled people and pilots to achieve the function of moving and input.The author researched a lot of access to human-computer interaction and computer vision on academic papers and literature at home and abroad in recent years, discussed the basic design of the human-computer interaction methods and a framework for pervasive computing, introduced the face detection and recognition of the general process and methods, and made its concrete realization of a more in-depth discussion, then proposed a framework protocol used to extend the functionality of third-party software to embed human-computer interaction seamlessly into the existing system based on computer vision. The main work includes the following:(1) Summarizes the significance of this research topic and prospects, reviewed human-computer interaction, computer vision research status, and summarizes the future development trends;(2) Interactive framework on pervasive computing are briefly analyzed and discussed the interaction of the two design methods, then discussed with examples in practical projects use and compared the two different methods of design process;(3) Based on face detection and recognition of the most popular currently open source computer vision library, the algorithm of detection and recognition is discussed in detail, and completed the development of face recognition systems;(4) Proposed a framework protocol to achieve human-computer interaction based on computer vision, which can be used by the third-party applications to provide functional extensions. Then the system is able to response to eye movements that were captured by the camera only by defining interaction description in the script file additionally.更多还原