基于清晰度的细胞显微图像分割和计数方法

Definition Based Cell Microscopic Image Segmentation and Counting Algorithm

目的针对细胞显微图像分割和计数困难的问题,提出一种基于清晰度的细胞显微图像分割和计数方法。资料与方法首先对细胞显微图像进行预处理,然后对图像进行离散余弦变换,并截断高频信号部分,再与原图做差以区分清晰部分和模糊部分,结合细胞图像局部聚合度较高的特性,利用区域生长方法提取完整目标,最后进行分析和计数。结果实验结果表明,针对细胞显微图像分割和计数不准确、速度慢的问题,利用该方法分割计数的准确率>90%,平均一幅图片的处理时间<100 ms。结论用清晰度的方法可以快速、准确地分割细胞显微图像并进行分析计数,对快速、准确地掌握细胞个数及密度有重要意义。

Purpose To propose a definition based algorithm for segmenting and counting cell microscopic images. Materials and Methods Cell microscopic images were first pretreated and then transformed using discrete cosine transformation（DCT）. The high frequency part was truncated and re-converted to differentiate clear and blurred images. The clear foreground regions were obtained. The intact objective was extracted using region growing method. Statistics and analysis of cell number were then conducted. Results This algorithm showed good performance in cell microscopic image counting with accuracy of over 90% at less than 100 ms/image. Conclusion Definition based method is fast and accurate in cell microscopic image segmentation and counting.