一种应用于视频眼震分析的瞳孔中心定位方法

A Pupil Location Method Used for Video-Nystagmograph

为利用眼震红外视频图像进行眼震分析,介绍一种瞳孔中心定位方法。此方法基于形态学和区域连接标记算法,提出了左上邻域方法进行连通图像标记,并对图像进行连通区域分割,再进行连通区域合理校验和连通区域合并,最终得到的瞳孔区域,通过计算获取瞳孔中心位置坐标,实时地把每帧图像得到瞳孔中心坐标通过曲线拟合出来,得到瞳孔的运动轨迹,从而获得临床所期望的诊断信息。眼动实验表明,此瞳孔中心定位算法简单,只需一次扫描图像标记,且定位速度比Hough变换法快一倍,能够满足红外视频眼震分析系统对瞳孔定位的实时性;基于同样的数据源,此方法的定位结果,与一种基于Hough的无现场校准的眼方向检测算法定位结果的相关度为0.975,同样可以获取良好的瞳孔运动轨迹,为眼震分析提供信息依据。

In order to analyze infrared video-nystagmograph,a pupil location method was presented.The method was based on morphology and regional connectivity sign algorithm,which included following steps：（1） using left-up neighborhood method for image segmentation,（2） judging the regions and merged the regions and eventually got the pupil area,（3） calculating the coordinates of the pupil of each frame images to track the pupil movement to obtain clinical diagnostic information.Experimental results of eye movement showed that the pupil location algorithm was simple and less time-consuming,only one time of image scanning and labeling was required,and as fast as twice of Hough transform.The method could meet the accuracy and real-time requirements of the pupil location of the infrared video-nystagmograph system.For the same data,the relevancy between the location results of the proposed method and the results by a Hough based eye direction detection algorithm without on-site calibration was 0.975.The pupil tracking by proposed method well reflected the eye movement,and provided useful information for nystgamus analysis.