The number of different human leukocytes is one of the important basis of disease diagnosis. Accurate segmentation of leukocytes is an important prerequisite step before classification and counting of leukocytes, and it determines the accuracy of counting. For different blood smear images obtained using various staining methods, according to color calibration, adjusting R, G and B components of the RGB images will change the color information of the images. Then based on H component in HSI color space and combined with Otsu automatic threshold method and area threshold method, a complete and single leukocyte images will be located and segmented, which will be used in subsequent classification and counting. The experimental results show that this method has high segmentation accuracy, strong practicability, simple operation and high efficiency.