摘要
为了准确有效地解决骨髓细胞的计算机识别分类问题,提出一种基于灰度阈值和彩色空间的骨髓细胞识别分类算法。该算法基于计算机图像处理技术,采用平滑、去噪等一系列的预处理得到平滑的骨髓细胞图像,通过对细胞图像的HSI颜色空间的分析,应用H通道和S通道的阈值分割方法分别将红细胞,白细胞的细胞核和细胞浆分割出来,并对有核细胞的胞核和胞浆提取形态特征和彩色光密度特征作为特征矢量,利用BP神经网络实现对骨髓细胞的分类识别。将该算法应用于临床采集到的150例骨髓细胞图像中,实验结果表明,该算法能较好地分类识别出各类骨髓细胞,具有较高的分类识别准确率。
In order to resolve the problem of recognizing bone marrow cells accurately and effectively, an identifying and classifying algorithm was proposed using gray level and color space. After image processing, smooth marrow cells images were gained by usingsmoothness, image erosion and so on. According to the analysis of the HSI color space of marrow cell images, an adaptive threshold segmentation method using H channel and S channel was developed to recognize the red cells, the nucleus and the cytoplasm of the bone marrow cells. To classify each kind of marrow cells using the Back Propagation neural network, the morphological characteristics and the colorful light density characteristics as characteristic vectors were extracted from the karyon and cytoplasm of the residuary nucleated cells. The algorithm were testified in 150 clinical collected cases of bone marrow cell images. The results proved that the proposed algorithm was valid and efficient in recognizing bone marrow cells and had relatively high accurate rates on identification and classification.
出处
《中国生物医学工程学报》
CAS
CSCD
北大核心
2009年第4期549-553,共5页
Chinese Journal of Biomedical Engineering
基金
山东省自然科学基金重点项目(Y2005C68)
关键词
骨髓细胞
HSI颜色空间
特征提取
BP神经网络
marrow cells
HSI color space
characteristic extraction
back propagation neural network
