1、北京科技大学本科生毕业设计（论文） - I - 摘 要 近年来，随着用户对人机交互的友好性和自然性的需求不断增加，兴起了一种新的人机交互技术 多点触摸技术。 多点触摸技术超越了传统人机交互技术中使用的鼠标和键盘，实现了用户通过手指在传统应用平台上的动作直接与电脑进行对话。它的主要优点是，能够实现同一时间的多用户交互，多触点交互，并且能够对交互的对象进行识别。多点触摸技术较传统的单一触点交互技术更灵活，更友好，更具自然性。 本文首先根据 GUI Group 提供的多点触摸系统开源程序代码完成了多点触摸人机交互系统底层软件的剖析。软件功能主要分为四大部分：图 像预处理、触摸点识别、运动跟踪、触点信
2、息的传输。之后在图像预处理步骤中的“图像分割”部分，本文采用了一种基于大津法（ OSTU）的自适应阈值分割方法，使系统能够在一定程度上自动完成在变光照条件下由灰度图像到二值图像的转化。 其次，考虑到原有特征点跟踪方法算法“ KNN（ K 最近邻分类算法）”的跟踪滞后性这一不足，本文基于“最小距离优先算法”改进出了一种适用于多点触摸系统的更高效的特征点跟踪方法“最小距离触点跟踪算法”，此方法能够更快速的完成对特征点的识别和跟踪。 最后，手势识别一直是多点触摸技术研究的主要内容，本 文定义了几种基本的手势，并参考相关资料利用 ActionScript 3.0 编程实现了多点触摸技术的简单应用。 实
3、验证明，使用本论文设计的多点触摸底层软件能够正确的检测并跟踪到触点，完成手势的识别和多点触摸的简单应用；并且利用本文改进的“最小距离触点跟踪算法”可以很好的提高触点跟踪速率，特征点跟踪滞后时间缩短。 关键词：多点触摸；特征点跟踪；最小距离优先算法；手势定义 北京科技大学本科生毕业设计（论文） - II - Study on Methods of Multi-touch Human-computer Interaction Based on Optical Induction Abstract In recent years, since users demands on nature and
4、friend attributes of computers are increasing, a new human-computer interaction technology called multi-touch human-computer interaction technology has been brought up. Multi-touch human-computer interaction technology exceeds the mouse and the keyboard used in traditional ways, making it possible f
5、or users direct communication with computers through the movements and gestures of fingers on the traditional application interface. Its main advantages are that it supports multi-consumers and multi-points direct interaction at the same time. Whats more, multi-point touch is able to distinguish int
6、eractive object identification. Compared with traditional single-point touch, it is more flexible, friendlier and more natural for users to use. Firstly, this paper completes the analysis of the ground software of multi-point touch system based on the open source code of multi-point touch system pro
7、vided by GUI Group. The function of the software is mainly divided into four parts: image pre-processing, the recognition of blobs, movement tracking and the transmission of blob information. In the image segmentation step of image pre-processing, this paper uses a self-adaptive threshold method bas
8、ed on the universal OSTU algorithm and makes a contribution to improve the effect of the binary image transformed from the grey-scale image in different light condition. Secondly, taking into consideration the disadvantage of track delay caused by original blob tracking algorithm KNN, this paper wor
9、ks out a more efficient tracking method named Nearest Distance Blob Tracking algorithm based on the universal Nearest distance Foremost algorithm. This improved algorithm is capable to complete the blob tracking process more quickly and cut down the delay time. Finally, as the gesture recognition ha
10、s always been the main research content of multi-point touch technology, this paper gives definitions of several basic gesture models and 北京科技大学本科生毕业设计（论文） - III - accomplishes the simple application of multi-point touch using ActionScript 3.0 based on related material. The experiment proves that th
11、e ground software of the multi-point touch system designed by this paper can correctly detect and track points, realizing the identification of gestures and basic multi-point touch application. Meanwhile, the Nearest Distance Blob Tracking algorithm improved by this paper succeeds in enhancing the tracking efficiency and the delay time of tracking is obviously decreased. Key Words： Multi-touch； Blob tracking； “Nearest Distance Foremost” algorithm； Self-adaptive Threshold； Gesture models
