图像信息驱动的抑制式粗糙模糊聚类分割算法

粗糙模糊聚类方法需要手动设置阈值确定粗糙聚类的上、下近似且对图像中的噪声较为敏感。为了减少人为干预，实现粗糙模糊聚类在图像分割中的深度应用，本文提出一种图像信息驱动的抑制式粗糙模糊聚类分割算法。方法中设计了基于超像素区域信息的自适应阈值策略，用于有效确定粗糙聚类的上下近似，将图像空间信息引入到粗糙模糊聚类，构造了融合空间信息的粗糙模糊聚类目标函数，克服方法对于图像噪声的敏感性，此外，为进一步提升聚类性能，将模糊聚类中的抑制式学习思想引入到粗糙下近似集中像素的模糊隶属度的修正，实现了粗糙和模糊思想的深度融合。本文算法是更具混合智能机理的粗糙模糊聚类图像分割算法，实验结果表明了本文算法的有效性。 

Image Information Driven Suppression Rough Fuzzy Clustering Segmentation Algorithm

Rough fuzzy clustering methods need setting the threshold manually to determine the upper and lower approximation for being sensitive to the noise. To reduce human intervention, this paper proposed a suppression rough fuzzy clustering segmentation algorithm driven by image information realizing the deep application of rough fuzzy clustering. The method designed an adaptive threshold strategy based on super-pixel region information to effectively determine the upper and lower approximation. The image spatial information was introduced into rough fuzzy clustering to construct the objective function for overcoming the sensitivity of the method with noise. Besides, the idea of suppression study in fuzzy clustering was introduced for correcting the fuzzy membership degree in rough approximation set, which realized the deep fusion of rough and fuzzy ideas. This algorithm is a rough fuzzy clustering image segmentation algorithm with more hybrid intelligent mechanism. The experimental results show the effectiveness of this algorithm.