基于椭圆拟合的改进Hough变换算法在眼控系统中的应用研究  

Application of Improved Hough Transform Algorithm Based on Ellipse Fitting in Eye-control System

作　　者：梁琼 赵慧民 葛任贤 熊建斌 杨盛雄 苏冠元 LIANG Qiong;ZHAO Hui-min;GE Ren-xian;XIONG Jian-bin;YANG Sheng-xiong;SU Guan-yuan(School of Computer,Guangdong Polytechnic Normal University,Guangzhou Guangdong 510665;School of Automation,Guangdong Polytechnic Normal University,Guangzhou Guangdong 510665)

机构地区：[1]广东技术师范大学计算机科学学院,广东广州510665 [2]广东技术师范大学自动化学院,广东广州510665

出　　处：《广东技术师范大学学报》2020年第3期9-15,共7页Journal of Guangdong Polytechnic Normal University

基　　金：国家自然科学基金委面上项目(61473331);广东省自然科学基金委面上项目(2019A1515010700)。

摘　　要：针对现有眼睛控制系统检测瞳孔位置准确度不高的问题,设计了一种基于椭圆拟合的改进Hough变换算法的眼控系统.首先,把PC机上的15个键盘的用户界面输出到本实验室自主研制的头戴式显示屏上;然后,通过摄像头拍测试者的眼球图像,把人眼图像传输到PC机中,在PC机上采用瞳孔角膜反射方法以及改进Hough变换算法对人眼图像进行处理并实现瞳孔定位;其次,依据测试者前期校正过程中所建立的人眼移动、眨眼等特征与头戴式显示屏所呈现内容之间的关系,采用眼球映射模型计算测试者的视线方向,判断出测试者所看的键盘上的按键;最后输出键盘按键所对应的信息到文字显示框或改变键盘界面状态.实验表明系统检测瞳孔位置的精度达到95%,能够识别眨眼动作且准确通过瞳孔跟踪实现眼睛控制键盘的功能,解决了现有眼睛控制系统瞳孔检测精度不高的问题,提高了瞳孔检测的精度.To solve the low accuracy problem of previous eye control systems in detecting the pupil position,an eye control system via improved Hough transform algorithm based on ellipse fitting has been designed in this study.Firstly,fifteen PC keyboards are outputted in user interface of the head-mounted display independently developed by our laboratory.Secondly,the tester’s eyeball image is taken through the camera,and the human eye image is transmitted to the PC.Then the pupil corneal reflection method and the improved Hough transform algorithm are used to process the human eye image and realize the pupil positioning.Thirdly,according to the relationship between human eye movement,blinking and other features of the tester established in the previous correction process,and the content displayed on the head-mounted display,the eyeball mapping model is used to calculate the direction of the tester’s line of sight so that the keys on the keyboard that the tester looking at is determined.Finally,the information corresponding to the keyboard keys is outputted.For example,the text display box or the state of the keyboard interface are changed.Experiments show that the accuracy of the system to detect the position of the pupil reaches is 95%.It can recognize the blinking movement and accurately realize the function of the eye control keyboard through pupil tracking,which solves the problem of the low accuracy of the pupil detection of the existing eye control system and improves the accuracy of the pupil detection.

关 键 词：眼睛控制系统 瞳孔位置 改进Hough变换算法 椭圆拟合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]