The Research Of Separating Leukocytes Algorithm Based On Intensity Transformation

Color information is always one of the most intuitive and important information of color images. According to color information, the human eye can distinguish a lot of things, and color category identified is far more than computer’s. So there will appear a lot of trouble by using computer to process images with different colors features or similar colors.In this paper, cytological image stained with Wright’s stain is researched. Because of different dyeing time and dyeing dosage, it will easily result in different colors of cytological image after dyeing. It is greatly increased the difficulty of the subsequent leukocytes segmentation. Combined with the characteristics of cell image, and using normalization method, which is very good for separating hue between erythrocyte and leukocyte, it is very conducive to gain automatic threshold, and then obtain the final image of leukocyte segmentation.According to cell images that can’t be used normalized processing directly, in this paper, we can analyze the difference between this kind of cell images and the above’ based on the original normalization method, and proposed a new normalization processing method by adjusting parameters of intensity transformation, through adjusting the left and right tensile range of the G component to obtain precise segmentation between erythrocyte and leukocyte, likewise, a better segmentation result of leukocyte can be acquired.This algorithm not only can let cell images,which can’t use normalization method directly, gain a good result of leukocyte segmentation based on the normalization processing after adjusting parameters, but also can increase robustness of the normalization processing.