基于瓶颈点探测与椭圆拟合的细胞图像分割方法探究

Cell image segmentation based on bottleneck point detection and ellipse fitting

针对显微图像中常出现的细胞重叠现象,提出了一种基于瓶颈点探测与椭圆拟合的细胞图像自动分割方法。首先,对重叠细胞图像进行闽值分割并得到边缘特征点,然后通过计算特征点间瓶颈率选出候选瓶颈点对,并使用椭圆拟合法对瓶颈点对进行判断,最后使用正确瓶颈点对分割重叠细胞。实验结果表明本文算法能够有效地抑制过分割和欠分割现象,且分割准确率高。

To segment the overlapping cells in microscopy images, an automatic cell image segmentation method based on bottleneck point detection and ellipse fitting is proposed. First, the cell image is segmented by the thresholding method, and get the feature points of the cell edges. Then, the candidate pairs of bottleneck point are obtained by detection method for pairs of bottleneck point, and are further judged by ellipse fitting. Finally, the overlapping cells are split by the correct pairs of bottleneck point. The experimental results show that the proposed method can prevent both over and under segmentation effectively and can achieve higher accuracies.