基于BF-WS的肝脏CT图像自动分割
BF-WS based automatic segmentation for liver CT images
吴福理 1鲁锦樑 1胡同森1
作者信息
- 1. 浙江工业大学计算机科学与技术学院,浙江杭州310023
- 折叠
摘要
由于血液的流动以及肝脏收缩与舒张,在腹部CT图像中存在一定量的噪声,会产生局部梯度极大值区域、弱边界、伪影等不良现象,加之腹部内器官分布紧密且灰度值接近,使得肝脏分割尤为困难.为此,首先采用改进的双边滤波算法,自适应的对腹部CT图像进行平滑滤波处理;其次,将自适应双边滤波算法和标记控制的分水岭算法有机融合起来,采用基于中值的多阈值最大类间方差算法提取肝脏标记,并运用分水岭分割算法进行精确分割;最后,分析对比不同的肝脏分割方法,表明该方法能够更有效地去除噪声,保留目标边界,提高肝脏分割的准确率.
Abstract
Because of blood flow,liver systolic and diastolic,there is a certain amount of noise in abdomen CT image,which probably produce the local gradient maximum region,weak boundary,artifacts and other undesirable phenomena.At the same time,abdominal organs and similar gray values make liver segmentation is filled with challenges.Therefore,first of all,an improved bilateral filtering algorithm is proposed,abdomen CT images are preprocessed adaptively;Secondly,the improved bilateral filtering algorithm is combined with a marked watershed algorithm,and liver markers are extracted through the improved multi-threshold otsu algorithm based on the median;Finally,experiments show that can not only eliminate noise but also preserve the edge information.It will improve the accuracy of liver segmentation.
关键词
肝脏分割/自适应双边滤波/基于中值的多阈值最大类间方差算法/标记分水岭算法Key words
liver segmentation/adaptive bilateral filtering/median-based multi-threshold otsu algorithm/marked watershed algorithm引用本文复制引用
基金项目
国家自然科学基金(31471416)
浙江省自然科学基金(LY14C130015)
出版年
2015
NETL
NSTL
维普
万方数据