基于贪心算法的快速模糊连接度图像分割

摘要

提出一种基于Dijkstra的贪心算法来实现模糊连接度的快速计算方法。每次寻找全局最优解，并从两个方面减少迭代次数来加快运算速度。首先，通过统一根节点模糊连接度的方式，去掉原始Dijkstra算法的更新操作，每个像素点一次即可完成模糊连接度的计算。其次，每一次迭代可以实现多个像素点模糊连接度的计算。通过真实医学图像测试表明，与其他改进算法相比，其分割速度明显加快，且不影响其分割精度。%In this paper, we present a Dijkstra-based greedy algorithm to realise the fast calculation method of fuzzy connectedness.It searches the global optimal solution every time and decreases the times of iteration from two aspects to speed up the computation.First, it removes the update operation of original Dijkstra’ s algorithm by the way of unifying the root node fuzzy connectedness, once per pixel point can the calculation of fuzzy connectedness be completed.Secondly, every iteration can realise the computation of fuzzy connectedness of multiple pixel points.It is showed by real medical image test that the new method is obviously faster in segmentation speed than other improved algorithms, and the accuracy of the segmentation doesn’ t be affected.