Abstract: Color feature plays an important part in bone marrow cell segmentation as well as classification. Firstly, a color feature extraction and analysis method was discussed based on the relationship among r, g, b values of the sampling data from known classes. Then,fcm segmentation and multi-level thresholding were used for cell segmentation considering the color features. With respect to the defect of loss edges and holes inside, a weighted range filter was proposed by introducing one coefficients called in-neighbor consistency. Experiments were done and results show that this approach is simple and effective.

Key words: Bone marrow cell segmentation, Range filter, Edge enhancement, Feature extraction