基于改进型最大类间方差法的瞳孔定位方法

摘要

To improve the accuracy and rapidity of pupil location in human-computer interaction, a method based on the improved OTSU algorithm is proposed in this paper. The image is pre-processed using light compensating and median filtering to calculate the maximum between-class variance of the pixels which exist in the image, and obtain the optimum threshold of the binary image, the pupil center and radius are got by using ellipse fitting. Experimental results show that the method is quite accurate and fast, which lays a great foundation for the eye tracking.%为了提高人机交互中瞳孔定位的精确性与快速性，提出一种基于改进型最大类间方差法的瞳孔定位方法。对图像进行基于光线补偿和中值滤波的预处理操作，进而根据直方图统计出图像中未出现的像素灰度值，计算余下像素的最大类间方差，求取二值化图像的最佳阈值，采用椭圆拟合方法确定瞳孔中心及半径。实验结果证明，该方法准确快速，为视线跟踪奠定必要的基础。