A Study On Preprocessing Technology Of Leukocytes Segmentation Based On Color Space

With the rapid development of medical image technology in recent years, leukocytes automatic identification technology becomes an important research subject in this application field. It’s a key step to extract leukocytes from blood images in leukocytes automatic identification technology, so it’s necessary to preprocessing the original blood images before extracting leukocytes. Its purpose is to get better visual and useful blood images for leukocytes segmentation, thus can lay the foundation for segmenting leukocytes accurately.Through analyzing the current development situation and application prospect of leukocytes segmentation method, a leukocytes segmentation normalization method based on color space has been proposed. At first, it normalizes the original images in RGB space, and then transform the normalized images from RGB space to HSL space in order to extract H and L component images. In L component images, use threshold segmentation method to remove backgrounds. Finally, segmenting leukocytes and erythrocytes in H component images. Considering color cast blood images can lead to normalization method invalid, hence do color correction on color cast blood images.The experimental results show, the normalization algorithm is not only remove erythrocytes and backgrounds and keep leukocytes well, but also we can transplant the algorithm to ARM development platform as network terminal products, which can apply in biomedicine field to increase the segmentation accuracy of leukocytes、erythrocytes and backgrounds in images, therefore improve some misdiagnosis and misjudgment problem about blood diseases.