一种基于脑血管图像特征的快速边界跟踪算法

Rapid boundary tracking algorithm based on cerebrovascular image feature

在处理复杂的脑血管图像时,经典的边界跟踪算法暴露出边界精度不高、边界不够平滑,且速度不尽人意等缺点。提出了一种新的快速边界跟踪算法,该算法在分析脑血管边缘垂直细节远多于水平细节的特征基础上,结合方向记忆选择搜索方向,并在不同的搜索方向上赋予不同的权值,最终得到下一个边界点。实验表明:该算法提取的脑血管边界平滑、速度快,适合脑血管图像的边界提取,为下一步的脑血管形状特征提取及表示提供了精确的数据准备。

When dealing with complex brain vascular images, classic boundary tracking algorithm exposed some disadvantages such as boundary was not high enough smooth, accuracy was limited and its speed was unsatisfied. This paper proposed a new rapid boundary tracking algorithm to solve those problems. Analysed the feature of cerebral vascular image edge that vertical details were more than horizontal details. Based on this feature, determined next border point with memory direction of the search options and different weights to different search directions. Experimental results show the proposed algorithm can get smooth boundary and its performance is enhanced greatly. For the next phase of cerebral vascular shape and feature extraction, the new algorithm provides accurate data.