White blood cell classification and counting of a blood smear is critical during patient diagnosis, however it is time consuming and fallible to do it manually. People has long been expecting systems to acquire clear blood smear images, segment and classify these images automatically, in which accurate white blood cell segmentation is a fundamental and essential process. This paper proposes a robust and high-performance white blood cell segmentation algorithm, which segments nucleus area from scaled saturation image first, then segments the plasma area in snipped sub image from green image and blue image delta normalized by red image. The algorithm is tested with 1350 images captured from various patients in diverse hospitals, which shows the algorithm results are 98% in accordance with the manual inspection results.