[1]王忠勇贾萌侯中新.基于邢台学特征的颗粒图像分割和计数[J].郑州大学学报(工学版),2015,36(02):80-84.[doi:10.3969/ j. issn.1671 -6833.2015.02.018]
　WANG Zhong-yong,JIA Meng,HOU Zhong-xin,et al.Segmentation and Counting of Granular Objects Based on Morphological Features[J].Journal of Zhengzhou University (Engineering Science),2015,36(02):80-84.[doi:10.3969/ j. issn.1671 -6833.2015.02.018]
点击复制

基于邢台学特征的颗粒图像分割和计数()
分享到：

《郑州大学学报(工学版)》[ISSN:1671-6833/CN:41-1339/T]

卷:
36
期数:
2015年02期
页码:
80-84
栏目:
出版日期:
2015-04-30

文章信息/Info

Title:
Segmentation and Counting of Granular Objects Based on Morphological Features
作者:
王忠勇1贾萌2侯中新3
1.郑州大学信息工程学院,河南郑州450001; 2.河南高远公路养护设备股份有限公司,河南新乡453000
Author(s):
WANG Zhong-yong1; JIA Meng1; HOU Zhong-xin2; CHEN Zai-xu2
1.School of Information Engineering,Zhengzhou University,Zhengzhou 450001,China; 2.Henan Gao-Yuan road maintenance e-quipment Corp., Xinxiang 453000,China
关键词:
颗粒分割计数; 分水岭算法; 形态学特征; 区域连接图
Keywords:
segmentation and counting;  watershed algorithm;  morphological features; RAG
分类号:
TP391
DOI:
10.3969/ j. issn.1671 -6833.2015.02.018
文献标志码:
A
摘要:
针对黏连严重、分辨率和清晰度较低的颗粒图像分割和计数问题,提出了一种基于分水岭和形态学特征的新方法.首先使用分水岭算法分割图像,得到过分割的结果,接着通过本文定义的区域形态学特征,再根据加权马氏距离和区域连接图(Region Adjacency Graph)进行区域合并.采用实际图像进行的实验表明,该算法效果良好,对原始图像的灰度、对比度和噪声变化具有不变性,在准确率–查全率( Precision - recall)曲线的表现上优于现有方法.
Abstract:
For the task of segmentation and counting of granular objects with low quality images,a new meth-od which based on watershed and morphology is presented. First ,an oversegmented result is gotten by conduc-ting the watershed algorithm to the smoothed image. Then the region merge stage,which is directed by usingthe morphological features defined in this article,as well as weighted Mahalanobis distance and region adja-cency graph (RAG),takes place.Experiments with real images demonstrate the validity of the method,thatit is invariant under translation of gray scale,contrast and noise. And it performs better than any present meth-od ,in terms of precision-recall criterion.
更新日期/Last Update: