悬液细胞图象识别的一种有效新算法

为进一步提高细胞识别的速度和识别率,本文研究一种细胞识别新算法。不同于其它的只对整个细胞形态进行一次图象处理的细胞识别方法,该算法在图象处理过程中,先后以细胞形态为重点和以细胞核形态为重点对原始图象进行两次图象处理,分别得到图象清晰度很高的整个细胞的二值图和细胞核的二值图。然后提取各项特征参数,对细胞图象进行识别。最后应用该算法对多幅细胞图象进行识别,结果表明该算法不但识别速度快,平均每幅图象的时间在3秒左右,而且识别率高达85%以上。

An effective new Algorithm of Image Recognition of suspended Cells

In order to enhance the speed of cell recognition and the recognition rate, a new algorithm of cell image recognition has been studied in this article. Different from other methods who have only once image processing on the whole image of the cell morphology, this algorithm has two image processing that focus on cell morphology and nuclear morphology of original image. And finally we got two high-definition binary images of the entire cell and of the nucleus. Then we extracted characteristic parameters, and...