Automatic Analysis And Its Algorithms Of Colorful Leukocyte Images

Leukocyte classification is very important in clinical blood examination. In this thesis, the structure of a leukocyte classification system is analyzed and an applied system structure using a PC, a microscope and a color CCD camera is proposed. Based on the HSI color space of a blood cell image, an adaptive method using S channel thresholding is developed to. locate the leukocyte and eliminate the background. Then, a region-growing algorithm relied on the local color similarity and the convexity of a leukocyte profile is described to segment cytoplasm and proved to be quite effective. The feature description of leukocyte is discussed and several features about morphology, color and texture are extracted. Finally, the problems of feature selection and the design of leukocyte classifier are discussed.