| 摘要: |
| 由于血管截面有可能出现凹状，导致种子点位置偏移到血管外，造成分割结果出现错误.为解决该问题，提出一种基于最小生成树的种子点选取算法.利用前一帧分割结果随机取点，并保证随机点均匀覆盖整个分割区域；再运用Kruskal算法对所有随机点构建最小生成树，并以最小生成树的树重心作为当前帧的种子点；最后以区域生长算法对当前帧进行分割.实验结果表明，本算法能较为准确地定位血管截面的中心点，保证血管分割的准确性. |
| 关键词: 最小生成树 凹形 区域生长 血管分割 图像分割 |
| DOI：10.3969/J.ISSN.1000-5137.2021.04.004 |
| 分类号:TP391.41 |
| 基金项目:国家自然科学基金（61373004） |
|
| Research on blood vessel section video segmentation based on tree centroid |
|
WANG Xiaohan, PENG Boxing, GUO Huinan, ZHAO Xinyue, MA Yan
|
|
College of Information, Mechanical and Electrical Engineering, Shanghai Normal University, Shanghai 200234, China
|
| Abstract: |
| The position of the seed could be outside of blood vessel due to the concave shape of the blood vessel section, which lead to wrong segmentation result. To solve the aboveissue, a seed selection algorithm based on minimum spanning tree was proposed in this paper. Firstly, random points wereselected from the segmentation results of previous framewhich were uniformly distributed in the entire segmentation region. Secondly, Kruskal algorithm was used to construct the minimum spanning tree for all random points, and the tree centroid was taken as the seed of the current frame. Finally, the current frame was segmented with region growing algorithm. The experimental results showed that by this algorithm the center point of the cross-section of blood vesselscould be accurately located and the accuracy of vessel segmentation could be ensured. |
| Key words: minimum spanning tree concave shape region growing blood vessel segmentation image segmentation |
