Colour Texture Features Based Approach for White Blood Cells Segmentation

Blood cell segmentation is an important research topic in Hematology and other related fields. In this article, a technique for microscopic images segmentation is proposed in order to extract the white blood cells (WBC) and its components (nucleus, cytoplasm) from the red blood cells and plasma. The image is represented in different color spaces, Haralick features extracted from the chromatic co-occurrence matrices (CCM) are used to characterize the textures present in these color images. A pre-treatment is carried out to extract the background (plasma) in order to reduce the execution time and noise. Segmentation has been done by supervised pixel-based classification using support vector machines (SVM). The proposed method was tested on twenty-seven real microscopic color images with promising results and nucleus recognition accuracy reaching 95%.