基于支持向量机的红细胞彩色图像分割算法

Segmentation of Blood Cells Image Based on Support Vector Machines

针对细胞图像分割中红细胞目标提取和重叠红细胞分割两个难点,提出一种基于支持向量机(SVM)的红细胞彩色图像分割算法,并通过实验对算法进行验证。该算法利用SVM对原始图像进行红细胞提取,把原始细胞分割成红细胞和背景两类目标区域,然后使用改进距离标记的分水岭算法对红细胞区域进行重叠分割。算法选择线性不可分的SVM模型和核函数RBF(C=1,iξ=0.2)时能够较好的分割红细胞彩色图像。

Aiming at the difficulties of cells image segmentation which involve the blood image target extraction and segmentation of clustering blood cells, a segmentation algorithm based on the Support Vector Machines （SVM） for blood cells image is presented. The algorithm extracted the blood cells from the original image by SVM. The image was divided into of blood cells area and background. Then the improved distance marked watershed method was used to separate the blood cells from the overlapping areas. The choice of non-linear dividable SVM types and kernel RBF function at （C= 1 ,ξ= 0. 2）results in satisfactory blood cells image segmentation.