快速空间邻域信息的中智模糊聚类分割算法

Neutrosophic C-means algorithm based on local information for image segmentation

为了克服传统的模糊C-均值聚类算法抗噪性能差的局限性,在中智模糊聚类基础上提出了一种新的基于邻域信息的中智模糊聚类图像分割算法。将中智集合引入模糊C-均值聚类算法,转化为一个优化问题。通过建立局部邻域信息约束的函数考虑像素之间的相互联系进行图像分割。通过对灰度图像添加不同的加性和乘性噪声进行分割测试,其测试结果表明,该算法得到的图像分割结果更稳定、边界更平滑且具有较强的噪声抑制能力。

To overcome the limitation of the traditional fuzzy C-means clustering algorithm, a new algorithm based on neighborhood information is proposed to solve the poor noise performance. The idea is to introduce the fuzzy C-mean clustering algorithm into an optimization problem. Image segmentation is performed by establishing the function of local neighborhood information constraints to consider the correlation between pixels. By adding different additive and muhiplicative noises to the gray image, the test results show that the proposed algorithm is more stable and smooth, and has better noise suppression ability.