【作者】 邹虹；

【导师】 郝冰；

【作者基本信息】 哈尔滨工业大学 ， 广播电视艺术学， 2009， 硕士

【摘要】 形体交流及表达等非语音类交流与语音交流有着同样的重要地位,包括人的身体姿势、头部动作、手势、注视和表情等。而计算机技术的发展使人与计算机的交流越趋频繁和复杂,以键盘和鼠标作为人机界面交流的传统方式,已无法传达出人机交流中形体所表达的细腻信息,更谈不上释放人机交流的多种潜能,于是研究符合人机交流习惯的新型交互技术也变得活跃,包括基于计算机视觉的动作识别技术。基于计算机视觉的动作识别系统,涉及到计算机科学、人工智能、神经生物物理学、认知心理学、艺术学等多门学科,本文对于新型人机交互模式的探索以及对机器人智能技术的研究势必对上述每一个学科都有不同的重要意义和影响。首先采用归类分析法对基于计算机视觉的动作识别技术与理论进行了分析研究,这其中包括人眼的跟踪识别、人脸识别、头部动作跟踪识别、手势识别以及姿势识别等。之后比较现有动作识别方法的各种优缺点,归纳总结出目前常用人机界面在人机交流中所起的正负作用。在上述研究基础之上,通过实验设计法,设计制作了一个基于头部动作识别的虚拟画展系统,以便深入了解基于计算机视觉的动作识别系统的每一个技术环节的基本功能和可能性,力图在实验和测试新型的人机交互模式时能够有效地分析出在新技术支持下人机交互系统的优势与不足。最后通过总结归纳法得出了本文的最终结论,即基于计算机视觉的动作识别,能使计算机识别并理解用户的行为,在键盘鼠标等传统交流设备的基础之上,用户可以更自然地与计算机进行丰富与复杂的交流。随着技术的发展与推进,当计算机视觉具备了人眼的基本功能,计算机处理系统就可以,也能够模拟人类处理信息的方式而衍生出具有“自我进化”的能力,到那时,计算机的智能将可以达到理解人类的交流方式的地步,人与机器之间的界面也将逐渐消隐,达到“隐性人机交互”的理想状态。更多还原

【Abstract】 Such non-verbal communication as human body communication and expression, including gazing, facial expression, head motion, gesture and posture of equal importance with the verbal communication. HM(Human Machine) Communication are getting more frequent under the development of digital technology. However, traditional keyboard-mouse interaction would not be able to explore the potential of the human body.In the dissertation, research is done on the gesture recognition based on computer vision, including eye-gaze recognition, face recognition, head recognition, gesture recognition, and posture recognition. By rearranging development of the human machine interaction technique, it is concluded that ordinary human machine interface is double-sword in HM interaction.At the same time,“Head-gesture-recognition based Virtual Library”was designed in order to study the techniques in every part of the computer-based gesture recognition. The advantages and disadvantages were analyzed while showing some novel ways of HM interaction.By the ways of classifying, analyzing, experimenting ,designing and concluding, conclusion is drawn that the computer vision based recognition will help computers understand of the users’behavior, and free them out from restriction of mouse and keyboard, make interaction more naturally. What if the computer saw as the human did, will the artificial intelligence reach the level of understanding human. Then the HM interface will disappear ,taken place by the“Hidden HM Interaction”.The study on this system ,which covers multiple subjects such as compute science, artificial intelligence, neuron science, bio-physics, cognition science, and artology, have great significance in exploring new ways of HM interaction and new techniques of robot design.更多还原