摘要
Snake模型已被广泛地应用于医学图像的分析之中.在应用该模型时,如何选取合适的初始轮廓是一个难题,因为初始轮廓的选择对它的分割速度和分割效果都有着重大的影响.笔者提出了一种使用基于标识分水岭算法来确定Snake模型的初始轮廓的方法,该方法先设置两条初始曲线,再通过标识分水岭算法来确定之间的最大连续高度点作为初始轮廓线.实验结果证明,它在简化确定初始轮廓的操作的基础上还提高了算法的效率,易应用于临床的系统中.
Snake model has been widely used in applications of medical image analysis. In the application of this model, how to select the appropriate initial contour is a big problem. The selec- tion of initial contour has great impact on the image segmentation speed and effect. This paper introduces a method of initial contour selection of snake model based on watershed algorithm. The algorithm starts with the two selected curves and ends with the initial contour which is deter- mined by the most consecutive points between the two initial curves. The experimental result shows that this method not only simplifies the operation of determining the initial contour, but also improves the efficiency of the system. It can be easily applied in clinical systems.
出处
《浙江工业大学学报》
CAS
2008年第4期374-377,共4页
Journal of Zhejiang University of Technology
作者简介
胡同森（1954-），男，江苏南京人，教授，研究方向为数字图像处理．
