Research On Image Segmentation Of Human Body's Peripheral Leukocytes

White blood cell (WBC) automatic image recognizing technology is one of the most important topics in the research of medical imaging field, and it has drawn more and more attention in recent years because of the following reasons: It can summarize pathologists' di(?)gnosing experiences scientifically, make good use of computer's smartness and exactness, meanwhile, it can help human being with the time-consuming manual checking work, work much faster than manpower at the same time. The segmentation of WBCs image is the first and an extremely significant step in the automatic classification of blood cells, so its segmenting quality, say, its degree of coincidence with the corresponding cells will directly affect the correctness of extracting parameters and the rate of system's identifying.This paper dealt with the problem of how to realize the aim of White blood cells' automatic segmentation. The background knowledge of image segmentation was presented in the first place. In order to work out an efficient way to realize the goal of WBCs' automatic recognition, basic research was done in the second section, and a new method to realize the segmenting goal was introduced here. This new method was organized as follows: Firstly, change the color image from Red-Green-Blue (RGB) model to Hue-Saturation-Intensity (HIS or HSI) space in order to get the Hue and Saturation components. The locations of cells are determined in the second step, the procedures of locating were as follows: the global gray histogram of Hue component was calculated in the first place; then, segmentation was done according to this global histogram, rectangles which encircled their corresponding cells were defined at last, every cell had only one rectangle; Based on the location of cells, the second segmentation - local segmentation was done in this step to reach the goal of cells' identification; The last step, internal segmentation was implemented in the Saturation component on the foundation of previous processing in order to get the areas of nucleoli, it was performed inside the segmented cells area gotten during the second step.All the work was performed on Microsoft Windows XP operation system, and the algorithms were realized with C/C++ programming language, in the famous Visual C++ 6.0 developing platform. Experimental results have shown that the method brought forward here...