基于视觉的囊胚自动注射过程中的囊胚检测及结构识别

Blastocyst detection and structure identification in automated blastocyst micro injection process based on machine vision

为实现囊胚自动注射过程中的囊胚位姿信息获取,研究了一种基于视觉的囊胚检测及其内部结构识别的方法。首先用Hough变换和聚类方法检测圆,实现囊胚的位置定位;然后等扇区划分囊胚区域,统计各扇区灰度特征,判断发育成熟度;最后根据发育成熟囊胚中内细胞团区域平均灰度偏差最大的特点予以识别。实验结果显示,该方法对囊胚的检测和结构识别的正确率分别大于90%和80%,表明其在囊胚检测和内部特征结构识别中的有效性。

This paper presented researches on blastocyst detection and structure identification based on machine vision to extract position-orientation information in an automated blastocyst micro injection process.Firstly,Hough transform together with clustering method was carried out for blastocyst detection.Then the target region was separated into equal sectors,in which gray-scale features of the image were calculated to judge maturity.Full grown blastocysts were further analyzed for inner cell mass（ICM） identification based on the fact that a maximum deviation of average gray-scale appeared on that region.Experimental result shows the effectiveness and the success rate of detection and structure identification are over 90% and 80% respectively.