基于视场边界最大最小聚类分析的快速扫描方法  

A fast leukocyte scanning method based on field border max-min distance cluster means

作　　者：杨坤[1] 曹益平[1] 

机构地区：[1]四川大学电子信息学院,四川成都610065

出　　处：《光电子．激光》2014年第4期823-828,共6页Journal of Optoelectronics·Laser

基　　金：国家'863'计划(2007AA01Z333);国家重大专项(2009ZX02204-008);四川省学术和技术带头人培养基金(07GRC-01)资助项目

摘　　要：针对用高倍镜头按一定规律路径会扫描到很多无法提取人体外周血白细胞的图像，为此提出一种基于低倍显微图像扫描定位细胞坐标、高倍显微图像提取细胞细节特征的外周血白细胞图像扫描新方法，有效提高了自细胞扫描速度，降低了细胞扫描重复率；提出了一种基于视场边界条件的改进型最大最小距离聚类算法，彻底避免了细胞重复扫描。实验表明，新方法扫描112个白细胞由传统方法的8～14min缩短到1．5～2．0min，减少了拍摄次数，提高了拍摄效率，避免了重复拍摄。Usually a certain number of leukocytes can be provided as a reference for diagnosis. By simple scanning blood specimen along a regulated path with the high power lens directly,only a few leukocytes can be extracted from lots of captured ones due to the lower proportion and the uneven distribution of leukocytes in the specimen,which will cause a very low scanning efficiency. A new fast scanning method is proposed,where peripheral blood leukocytes in low-magnification microscopic image are located, then the leukocyte detail features in high-magnification microscopic image are got. The new method improves scanning efficiency and reduces repetition rate based on field border max-mAn distance cluster means. The experimental results indicate that to scan 1,2 leukocytes,the conventional method needs 8-14 min, while the new method needs only 1.5-2.0 min. The new method reduces the capture times greatly,enhances scanning efficiency and saves computer resources.

关 键 词：白细胞扫描 视场边界 最大最小距离聚类 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]