ZOU Shuang-shuang, GUO Rong, HONG Huan-huan, et al. Automatic identification algorithm for embryo based on EDContours and EDCircles[J]. 2019, (4): 135-138.
ZOU Shuang-shuang, GUO Rong, HONG Huan-huan, et al. Automatic identification algorithm for embryo based on EDContours and EDCircles[J]. 2019, (4): 135-138. DOI: 10.13873/J.1000-9787(2019)04-0135-04.
基于EDContours和EDCircles的胚胎自动识别算法
摘要
针对EDCircles算法在胚胎边缘检测上存在部分丢失的问题
集成EDContours算法
获得精确的胚胎边缘;结合EDCircles中的EDLines算法获得线段图
依据直线向量间的夹角和方向约束
将线段连接成弧
应用最小二乘圆拟合算法和增强的直接最小二乘椭圆拟合算法
输出圆和椭圆检测图;设置圆半径与椭圆短轴的约束值
去除伪圆和伪椭圆
给出最终的检测结果。实验结果表明:该算法具有良好的准确性与通用性
能够实现胚胎卵裂球的快速检测和精确定位。
Abstract
Aims at the problem of partial loss of embryonic edge detection by the EDCircles algorithm
and integrate EDContours algorithm to obtain accurate embryonic edge. Then
combine EDLines algorithm to obtain line segments graph
the line segments are converted into arcs according to the angle between the straight line vector and the direction constraints. Besides
circles and ellipses detection map is outputted by the least squares fitting of circles and the enhanced direct least square fitting of ellipses algorithm. Finally
the constraint values of radius of circles and minor axis of ellipses are set to remove pseudo-circles and pseudo-ellipses
and the final detection results are given. Experimental results show that this method has good accuracy and versatility
and can realize rapid detection and precise localization of embryonic blastomeres.