基于卡尔曼滤波改进的精子图像序列分割方法

图像分割是精子图像识别的一项关键技术,在精子运动能力分析中起着至关重要的作用。本文对采集的连续精子图像序列进行灰度化、去噪等预处理后,采用Otsu算法对首幅动物精子图像二值化,对后续图像采用Kalman Filter确定二值化阈值范围,改进Otsu算法求出每一幅图像的适当阈值并进行二值化,缩短算法时间并能保证分割精度。应用形态学消除精子尾部和部分精子之间的粘连现象,通过计算和比较目标面积、形状因子,去除小颗粒杂质以及形状及灰度和精子相似的杂质,为精子运动能力检测提供高质量的分割图像。

An Improved Sperm Image Sequence Segmentation Method Based on Kalman Filtering

Image segmentation in the analysis of sperm’s motility plays an important role.Firstly,the collected sperm image sequence are carried out a pre-processing,such as graying and de-noising,then the first piece of pretreated sperm image’s binarization using the Otsu algorithm and the threshold range of binarization of the subsequent images can be determined by the Kalman filter,which improves the Otsu algorithm to calculate the images’ appropriate threshold and binarization under the predicated threshold range,and reduces time.In order to get high-quality segmentation images for the analysis of sperm’s motility,we apply morphology to eliminate sperm tail and part of the phenomenon of sperm adhesion,and by comparing with the area and shape factors to remove the tiny noisy particles and impurities in the sperm image.