Abstract: Perimeter need get the pupil position of patients rapidly and accurately in the checking process. Existing pupil positioning approaches have a good position result when the pupil’s boundary looks clear and the contrast is strong. However when the patients are wearing corrective lenses, the infrared ray passing through the corrective lenses, which is produced by the camera, will be blocked to some extent and makes a greater reflection, resulting in making the image acquired by the camera blurred and low contrast. In addition, the frame of astigmatism lens also makes strong interference, so the accuracy of existing approaches significantly decreases. In order to improve the accuracy of pupil positioning with a low contrast fuzzy environment, this paper proposes a new pupil positioning approach. Firstly, it uses image preprocessing and region growing to extract main area of the pupil. Then it uses arc length increment method based on a voting system to calculate the center and radius of the pupil. Experimental results show that this method is not only applicable to pupil position under low contrast fuzzy environment, but also has a good effect in normal condition.

Key words: pupil positioning, sliding arc length increment, voting

摘要： 在视野计检查过程中，需要准确快速地获取患者瞳孔的位置信息。现有定位方法在瞳孔边界清晰且与周围区域对比度较强时，有着较好的定位效果，但当患者带有矫正镜片时，相机发出的红外线在穿过矫正镜片时会受到一定程度阻挡，产生较大反射，造成相机获取到的图像模糊，对比度下降，再加上矫正镜框等噪声的干扰，现有瞳孔定位方法的准确性明显下降。为了提高在低对比度模糊环境下瞳孔定位的准确性，提出了一种新的瞳孔定位方法。该方法通过图像预处理和区域生长提取出瞳孔区域，采用基于投票的滑动递增弧长方法确定瞳孔中心及半径。实验结果表明，该方法不仅适用于低对比度模糊环境下的瞳孔定位，在正常环境下同样有着很好的效果。

关键词: 瞳孔定位, 滑动递增弧长, 投票