眼动信号的提取与分类识别研究

发布时间：2018-09-18 10:13

【摘要】：人机交互技术是连接人与计算机或其他电子设备的桥梁。近几年来,随着科学技术的快速发展,人机交互技术也因此取得了更大的发展,不断的向更自然、更和谐、更便利的方向发展。人作为生物体自身包含了各种各样的生物电信号,例如:脑电信号、肌电信号、心电信号、眼动信号等。人们可以对这些信号进行采集并进行有效的破译,然后再赋予这些信号特定的含义,这样就可以为人类所用。利用这些丰富的生命体征信息资源开发出更加具有交互性的以“人类为主导”的人机交互系统,逐渐成为科研工作者和相关领域专家不断探索的问题。眼动信号是一种由眼部运动而引起眼部周围电势发生变化的生物电信号。随着眼球运动,眼动信号也会发生变化,眼动信号的波形特点与眼球运动的方式有直接的对应关系;并且,眼动信号具有幅值较高、波形便于检查、处理容易等优势。因此,将眼动信号作为开发基于眼动信号的人机交互技术是解决特殊人士人机交互问题的新方向。本文主要研究了眼动信号的采集和预处理、端点检测、有效眼动信号的提取、特征提取,并分别用BP算法、SVM算法、DTW算法对眼动信号进行模式识别研究。其中,DTW算法有效的解决眼动信号因人而异、长短不一的问题,并提高了眼动信号的识别率,取得了较好的效果,为基于眼动信号的人机交互系统设计奠定了一定的研究基础。文章主要完成了以下几项主要工作:1.眼动信号的采集和预处理:设计了眼动信号采集实验并且对多名受试者进行数据采集,获取了大量的原始眼动信号数据。并通过硬件电路对原始数据进行初步的滤波、降噪处理。2.有效信号的提取:通过对信号进行软件滤波、分帧、加窗、计算短时能量以及端点检测等处理后,提取可靠的有效信号。3.眼动信号的特征提取及分析:根据有效信号的波形特点,分析可用于模式识别的特征值,提取基于波形特点的眼动信号特征值。4.分别利用BP算法、SVM算法以及DTW算法对眼动信号进行分类识别研究,讨论三种模式识别方法的优劣性。
[Abstract]:Human-computer interaction technology is a bridge between human beings and computers or other electronic devices. In recent years, with the rapid development of science and technology, human-computer interaction technology has also made greater development, constantly to more natural, more harmonious, more convenient direction of development. As an organism, human beings contain a variety of bioelectrical signals, such as EEG, EMG, ECG, eye movement and so on. These signals can be collected and deciphered effectively, and then given a specific meaning, so that they can be used by human beings. Using these abundant vital sign information resources to develop a more interactive human-computer interaction system, which is gradually becoming a problem that researchers and experts in related fields continue to explore. Eye movement signal is a kind of bioelectric signal caused by eye movement. With the eye movement, eye movement signal will also change, the characteristics of eye movement signal waveform and eye movement mode have a direct corresponding relationship; moreover, eye movement signal has the advantages of high amplitude, easy to check the waveform, easy to process, and so on. Therefore, it is a new direction to develop the human-computer interaction technology based on eye movement signal to solve the problem of human-computer interaction by special people. In this paper, the acquisition and preprocessing of eye movement signal, endpoint detection, extraction of effective eye movement signal and feature extraction are mainly studied, and the pattern recognition of eye movement signal is studied using BP algorithm. The DTW algorithm can effectively solve the problem that eye movement signal varies from person to person, and improves the recognition rate of eye movement signal, and achieves good results, which lays a certain research foundation for the design of man-machine interaction system based on eye movement signal. The article mainly completed the following main work: 1. The acquisition and preprocessing of eye movement signal: the experiment of eye movement signal acquisition was designed and a large number of original eye movement signal data were acquired. And through the hardware circuit to the original data preliminary filtering, noise reduction processing. 2. Effective signal extraction: after processing such as software filtering, framing, windowing, calculating short-time energy and endpoint detection, the reliable effective signal .3is extracted. Feature extraction and analysis of eye movement signals: according to the waveform characteristics of effective signals, the eigenvalues that can be used in pattern recognition are analyzed, and the eigenvalues of eye movement signals based on waveform characteristics are extracted. BP algorithm and DTW algorithm are used to classify and recognize eye movement signals, and the advantages and disadvantages of the three pattern recognition methods are discussed.
【学位授予单位】：上海师范大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN911.7

【参考文献】