一种视觉显著性引导的模糊聚类图像分割方法  

Visual Saliency-guided Fuzzy Clustering Method for Image Segmentation

作　　者：白雪飞[1] 韩晓静 王文剑[1,2] BAI Xuefei;HAN Xiaojing;WANG Wenjian(School of Computer and Information Technology, Shanxi University, Taiyuan 030006, China;Key Laboratory of Computational Intelligence and Chinese Information Processing of Ministry of Education, Shanxi University, Taiyuan 030006, China)

机构地区：[1]山西大学计算机与信息技术学院,山西太原030006 [2]山西大学计算智能与中文信息处理教育部重点实验室,山西太原030006

出　　处：《郑州大学学报（理学版）》2022年第2期1-7,共7页Journal of Zhengzhou University:Natural Science Edition

基　　金：国家自然科学基金项目(61703252,62076154);国际科技合作计划项目(201903D421050);中央引导地方科技创新项目(YDZX20201400001224)。

摘　　要：传统的模糊C均值聚类算法利用图像的灰度、颜色、纹理、强度等底层特征进行聚类,实现图像的分割,它容易受到噪声的影响,且计算量大,不能提供理想的彩色图像分割结果。针对这些问题,提出一种视觉显著性引导的模糊聚类图像分割方法。首先使用显著性检测对图像进行初始化分割,得到带有区域级标注信息的引导图,然后将引导图作为指导信息,引导模糊聚类算法对图像进行细分割。在公共数据集上的实验结果表明,本文方法与其他改进的FCM算法和深度网络分割模型相比,可以取得较好的分割效果,有效减少了分割时间。The traditional fuzzy C-means clustering algorithm uses the low-level features of image,such as grayscale,color,texture and intensity,to achieve image segmentation.These methods are susceptible to noise and has a large amount of calculation,so they can not provide ideal color image segmentation results.To solve these two problems,a visual saliency-guided fuzzy clustering image segmentation method was proposed.Firstly,the image was initialized and segmented by saliency detection,and the guide graph with region-level labeling information was obtained.Then,the guide graph was used to guide the fuzzy clustering algorithm to segment the image.Experimental on the three common data sets showed that,compared with other improved FCM algorithm and deep network segmentation model,the proposed method could achieve better segmentation results and reduce the segmentation time effectively.

关 键 词：模糊C均值聚类 显著性检测 彩色图像分割 聚类分割 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]