K-means cluster algorithm based on color image enhancement for cell segmentation

• Published in: 2012 5th International Conference on Biomedical Engineering and Informatics pp. 295 - 299
• Main Authors: Yan, Man, Cai, Jianyong, Gao, Jiexing, Luo, Lili
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.10.2012
• Subjects: cell segmentation; Clustering algorithms; Color; color cell image; Filling; Filtering; Image color analysis; Image enhancement; Image segmentation; k-means cluster; Morphology; Noise reduction; Pattern recognition; Ycbcr color space
• DOI: 10.1109/BMEI.2012.6513157
• ISBN: 9781467311830; 1467311839

Color cell image recognition and segmentation are two important issues in the field of biomedical cell morphology. The conventional segmentation method of color cell images based on k-means cluster is unreliable, since the color information from every category is similar. This paper presents a new method about cell segmentation by k-means cluster based on color image enhancement. Firstly, the cumulative distributions of the R, G, and B component gray value are calculated to find the mean value in the distribution. Secondly, the enhanced images are divided into three categories as masking images by k-means clustering algorithm in Ycbcr color space. And then, the binary images are de-noised via the morphological processing. Finally, the leukocytes and erythrocytes are segmented. The experimental results based on image enhancement mechanism by k-means clustering showed that the algorithm has a good discriminating and segmenting effect and while maintaining critical information color image.