摘要： 传统Ostu法在分割大鼠精子图像过程中存在实时性较差的问题。为此，提出一种新的图像分割算法，基于改进的Ostu法对大鼠精子图像进行分割，利用Newton迭代法缩短算法执行时间，根据形状因子去除杂质，应用形态学在每个精子目标上显示数字标记。实验结果表明，该算法可以实现精子图像的快速准确分割，并自动统计出图像中精子的个数。

关键词: Ostu方法, 图像分割, 形状因子, 数字标记

Abstract: In response to the poor real-time ability of traditional Ostu in segmenting of pat sperm images, it proposes a new method to segment such images based on improved Ostu, which reduces the execution time by Newton iteration, removes impurities according to the shape factor, and applies morphology to mark each sperm with digital tags. Experimental results show this method can achieve fast and accurate segmenting of sperm images, and automatically count the number of sperms in the images.

Key words: Ostu method, image segmentation, shape factor, digital tag

中图分类号: 

• TP311.52