血红细胞的计算机自动识别与快速计数方法

A Computer Procedure for Recognizing and Counting of Blood Cells

本文介绍了一种用计算机数字图象技术对显微视场中血红细胞进行自动识别与快速计算方法。该方法是依次对图象作辐条滤波处理及快速标号,达到识别与计数的目的。检验计算证明,对一帧64×64象素的图象作处理,能正确识别各个血球细胞,在IBM/AT机上完成整个处理过程不到1分钟。

The data of blood cells were obtained using an optical system (including a microscope) and a TV camera. A computer procedure consisted of a spoke filter, an afterprocessing algorithm and a fast counting algorithm was applied. The method was based on the presumptions that red blood cells, white blood cells and platelets are different in sizes and other features. They all have round shape and their size are known. The spoke filter made use of the direction feature of the edges. It can detect a large variety of shapes within a specified range of sizes. Once blood cells were detected, their numbers were counted with fast labelling algorithm effectively. As an example, red blood cells were discriminated from other blood cells. A 64×64 pixels digital image of blood cells was used in computer simulation experiment. The whole process took less than 1 min on IBM PC/AT.