红外眼震视频瞳孔寻迹

Infrared Video-nystagmograph Tracing

针对眼震红外视频图像灰度分布不均匀造成瞳孔边缘检测精度不高的状况,介绍了一种基于形态学和Canny算法实现瞳孔中心定位方法,运用形态学去除一些无意义的区域,使目标瞳孔平滑,分离和提取最大连通区域瞳孔,再根据所设计的Canny算法实现瞳孔边缘提取,通过计算获取瞳孔中心位置坐标,实时地把每帧图像得到的瞳孔中心坐标通过曲线拟合出来,得到瞳孔的运动轨迹,从而获得眼震临床所期望的诊断信息。实验结果表明,该方法能够很好地适应不同实验对象灰度值差别,进行准确的边缘提取,拟合的瞳孔运动轨迹良好地反映眼睛运动情况,为国内视频眼震瞳孔中心定位的研究提供一种可借鉴的实用方法。

As the infrared image gray distribution of nystagmus video is uneven,the accuracy of pupil edge detection could not be high enough.A pupil location method based on morphology and Canny algorithm was presented in this paper.Some meaningless regions were removed by morphology,and target was smoothed.The connected regions were separated,and the largest connected region-pupil was found.Then the pupil edges were extracted by the designed Canny edge detection algorithm.The coordinates of pupil for each frame images were calculated and the pupil movement tracking was fitted.Finally,desired clinical diagnostic information would be obtained from the tracking.Experimental results showed that the method could be adapted to different gray values for different objects,and could accurately detect the edge of the pupil.As the pupil was tracked well,the experimental results could reflect the eye movement.The method provides a practical way for the pupil location of existing domestic researches in video-nystagmograph.