基于向量流场节点的图像分割算法

Image Segmentation Algorithm Based on Nodes of Vector Flow Field

人工干预使蛇模型只能用于半自动的图像分割,该文在梯度向量流(GVF)蛇模型的基础上提出一种基于流场节点与最小路径方法的全自动图像分割算法。在图像的GVF场上检测出流场节点,以节点为种子,采用多标记快速扫描法获得一个初始分割,采用区域合并得到最终分割结果。实验结果证明了该算法的鲁棒性和有效性。

Snakes model is usually used for semi-automatic image segmentation for the existence of human interaction. In this paper, on the basis of Gradient Vector Flow（GVF） snakes, a fully automatic image segmentation algorithm based on the analysis of flow field and the minimal path method is proposed. It detects nodes of GVF field, sets the nodes as seeds, acquires an initial segmentation by multiple-label fast sweeping method, and uses a region merging to get the segmentation result. Experiments demonstrate the robustness and effectiveness of the algorithm.