色彩空间变换和基于距离变换的分水岭算法在白细胞图像分割中的应用

Application of Color Space Transformation and Watershed Algorithm Based on Distance Transform in White Blood Cell Image Segmentation

在血液细胞显微图像中,由于红细胞、血小板与白细胞颜色相近,并且存在粘连白细胞,提取白细胞用于分类计数往往达不到较好的效果。本文提出一种色彩空间变换的方法滤除红细胞和血小板,即将RGB色彩空间上急性淋巴细胞白血病患者的血液细胞显微图像转换至HSV色彩空间,检测并提取白细胞。对于提取结果中的粘连白细胞,采用基于距离变换的分水岭分割算法标记粘连边界。同时实验证明,本文所提出的方法在分割精度上有着良好的表现。

In the microscopic images of blood cells,since red blood cells,platelets and white blood cells are similar in color, and there are adhesion white blood cells,the extraction of white blood cells for classification and counting often cannot reach better results.In this paper,a color space transformation method was proposed to filter out red blood cells and platelets,that was,to transform the color space from RGB to HSV of blood cell microscopic images for acute lymphoblastic leukemia patients, which aimed to detect and extract white blood cells.For the adhesion white blood cells in the extraction results,the watershed segmentation algorithm based on distance transformation was used to mark the adhesion boundary.At the same time,the experiment indicates that the proposed method has a good performance in segmentation accuracy.