White blood nucleus extraction using K-Mean clustering and mathematical morphing

The White blood cell detection is most important in detection of various kinds of diseases in human body as it provide valuable information to doctors for diagnosis of diseases. This paper focuses on automatic extraction of leukocytes using image processing techniques such as color segmentation, automatic thresholding and mathematical morphing. We have used the K-Mean clustering for the color based segmentation to detect white blood nucleus on a set of 480*640*3 images. Since manual segmentation is very tedious, tardy and sometime prone to error, besides that the medical equipments which are used for white blood cells detection are very costly and may not be exist in all the hospitals and clinics, so, the automatic system is preferred. In this paper we firstly apply color based segmentation on images for segmentation of white blood cell and platelets which is most important for localization of nucleus, then converting the segmented image to gray scale. We then analyze the nucleus features of white blood cells by mathematical morphing for removing the platelets from the segmented image and this result in final extraction of white blood nucleus. In this research results obtained give a good accuracy rate as compared to others.