基于模糊连接度的图像分割及算法

提出了一种基于模糊连接度图像分割的方法,在模糊连接度分割的基础上增加了最优路径(即与种子点的所有路径中连接度最大的路径)上各点相对于种子点的属性相似度的检验,使之能在待分割对象边界比较模糊的情况下取得理想的分割结果.同时提出了一种基于图像扫描机制的算法,它以种子点为中心,逐个计算邻居点相对于种子点的模糊连接度,该算法充分利用模糊连接度和属性相似度的性质,能简单、快速地找到两点间的最优路径.实验结果表明,该方法快速、有效地提高了图像分割的质量.

An Image Segmentation and Its Algorithm Based on Fuzzy Connectedness

In this paper, a modification to the fuzzy connectedness image segmentation is presented. Through checking the property affinity between the seed pixel and the pixel along the optimal path which has the largest fuzzy connectedness from the pixel to seed pixel, good results can be achieved, especially for those objects with blurred boundary. Additionally, an image-scanning mechanism algorithm for detecting optimal paths is proposed to calculate the fuzzy connectedness between pixels and the seed pixel one by one. This algorithm can make full use of the properties of fuzzy connectedness and property affinity, and detect the optimal path between two pixels effectively. Experimental examples show that the new method is simple, fast, and works well for some images.