尿沉渣图像的细胞分割和特征提取  

Segmentation and Feature Extraction of biological particles in Urinary Microscopic Images

作　　者：段华琼[1] 梁琰[2] 钱基业[2] 张伟文[3] 

机构地区：[1]成都理工大学工程技术学院计科系,四川省乐山市614007 [2]重庆大学计算机学院,400030 [3]重庆大学机械工程学院,400030

出　　处：《微计算机信息》2009年第9期306-308,共3页Control & Automation

摘　　要：针对尿沉渣图像成像的特殊性以及图像中各有形成分特点,对图像处理和特征提取带来的影响,提出了一种基于8-邻域灰度差分和canny算法的细胞分割算法以及基于canny算子和环投影细胞特征提取方法,以便有效进行细胞定位、细胞封闭轮廓提取以及常见有形成分中红、白细胞特征的提取。试验结果表明,该方法有效地克服了因光照不均等不确定因素的干扰,提高细胞定位准备率,较好地勾勒了细胞轮廓,刻画了红、白细胞的特点。Due to the poor resolution ＇and strong variability of objects, and due to the noisy and highly variable background, urinary microscopic image processing is very challenging. In this paper, a cell segmentation method based on 8-neighborhood gray level difference and canny＇s operator, and also a cell feature extraction method based on canny＇s operator and Ring-Projection are proposed to detect the object of interest and extract effective feature in characterizing the red blood cell and white blood cell. The experimentation results show that this method performs well in cell detection and feature extraction in the noisy and variable illumination background.

关 键 词：尿沉渣图像 CANNY算法 细胞分割 细胞特征提取 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]