一种基于边缘提取的交互式图像分割算法

An interactive image segmentation algorithm based on edge extraction

提出了基于边缘提取的交互式图像分割算法,该算法将图像映射为无向图,使用拉普拉斯零交叉点、边缘强度和动态轨迹长度构造能量模型,并为无向图中的边赋予能量代价。根据能量代价,引入角点信息,在交互得到的控制点间搜索最优路径,迭代此过程,实现分割。实验结果表明,该算法具有较高的精度和效率,能较好地克服噪声影响,适用于灰度及彩色图像。

The algorithm presented in this paper maps the image to undirected graph, and structures energy model with Laplace zero intersection, edge strength, dynamic path length, and endows energy cost for edges in the undirected graph. According to energy cost and fusion angular point information the algorithm searches the optimization path between the control points obtained by interaction, and realizes the segmentation by iterating this process. The experiment results demonstrate the algorithm has high precision and efficiency, and effective in anti-noise. The algorithm is suitable for gray and color image.