摘要
为了有效地分割肝脏区域,提出一种改进的基于标号搜索算法的自动分割肝脏算法。在基于标记搜索算法的基础上,将梯度大小作为权重因子加入梯度标号图,同时将肝脏轮廓与灰度值分布相联系,能有效地操作与轮廓有关的像素,很好地保留了肝脏区域。通过计算每个候选像素最小代价函数来确定最佳路径,获得一个准确的肝脏轮廓。实验表明,改进算法能有效地分割肝脏,极大地提高了分割肝脏的准确性。
In order to effectively segmentateand calculate volume of the liver, this paper proposes an improved algorithm to auto-matic segmentateand of liver.Based on the labeling search algorithm take the gradient magnitude as a weighting factor, will add the gradient into labeled graph, while the liver contour and the gray value distribution phase, can operate effectively with contour pixels. Through the calculation of each candidate pixel minimum cost function to determine the optimal path and obtain an accurate liver contour. After the division of the liver section, the use of its thickness and interlayer information and pixel size to automatically real-ize the liver volume measurement. Experiments show that the improved algorithm can effectively segmentate liver, greatly improves the accuracy of segmentation of liver.
出处
《自动化与仪器仪表》
2014年第1期142-146,共5页
Automation & Instrumentation
基金
甘肃省自然科学基金项目(0803RJZA015)
甘肃省研究生导师科研项目(1135-02)
关键词
肝脏分割
形态学过滤
变形轮廓
K均值聚类
标记搜索算法
Liver segmentation
Morphological filtering
Deformable contouring
K-mean clusteringLabeling search algorithm
作者简介
陈秀兰（1971－）,女,江苏丹阳人,硕士研究生,副教授,主要研究方向为图像处理、数据库挖掘.
