基于活动轮廓模型的彩色白细胞图像自动分割方法研究

为解决血液白细胞显微图像自动识别中的图像分割问题，提出了一种基于活动轮廓的彩色白细胞图像自动分割方法，首先在Hue，Saturation，Intensitv（HSI）彩色空间中运用聚类分割得到细胞核，从而得到细胞所在的位置，然后用流域算法得到细胞大致的轮廓，最后将此轮廓作为初始轮廓，用梯度矢量流（GVF）外力及来自全局信息的区域力驱动，结合彩色信息，使得轮廓收敛于真实的细胞边界。实验结果表明，此方法能精确、有效地分割出单个以及部分重叠白细胞区域。

Automated Segmentation of Colored Leukocyte Images Based on the Active Contour Model

To solve the problem of image segmentation for automatic recognition of colored leukocyte images,a novel cell segmentation method employing active contour model for leukocyte cell images is presented.Firstly,K-means clustering algorithm based on Hue,Saturation,Intersive(HSI) color space is applied for nucleus localization.Then the leukocytes roughly by using watershed algorithm is segmented.Finally,the initial snake contour placed inside the leukocyte is driven by the gradient flow vector force and region force used in the active contour model,combined withcolor and region messages,to fit the leukocyte and attain the precise contour at last.The presented process has been tested on sets of cell images.It shows that the proposed means performs well on well-separated cells and some overlapped ones.