An Automated Segmentation Scheme For Colored Leukocyte Images

Leukocyte examination is of real significance in hemanalysis. The basic task is to check and calculate the main parameters of leukocytes, such as the quantity, the comparative ratio, the modality, and etc., for diagnosing the presence, type and severity of diseases. This automatic process could not only help improve the efficiency of hemanalysis work and lighten the labor burden but also enhance the precision of examinations.This paper emphasizes on the automatic recognition research of colored leukocyte images, especially its most crucial technique - the automatic segmentation. The substantial problem discussed in the paper is how to extract the interested target, the leukocyte, from a complicated background. Hereby, we present an approach by three steps for the detection and segmentation of the leukocyte.Firstly, we analyze the color feature of leukocyte images, and then apply an algorithm of segmentation for nucleus extraction based on the HSI color space, which establishes a foundation for the detection of leukocytes. For the localization and the counting of leukocytes, we use an efficient algorithm for labeling connected region to obtain several separate single-cell windows.Secondly, separating the cytoplasm roughly by using a combined algorithm of the K-means clustering and the watershed, we disconnect the accompanying red blood cells from our region of interests. After this step, leukocytes can be extracted approximately in complete.Finally, in order to separate the leukocytes more accurately, we apply an active contour model. In our method, we utilize the features of leukocyte images, improve the vector gradient flow (GVF), include the region force from holistic information, and use the color messages, all of which help achieve precise contour and realize the aim of accurate segmentation.