首页|基于空间信息的鲁棒模糊C均值聚类的苗族服饰图像分割算法

基于空间信息的鲁棒模糊C均值聚类的苗族服饰图像分割算法

扫码查看
原文链接
  • 万方数据
  • 维普
针对苗族服饰图像中破损污渍、折叠痕迹、色彩差异大和噪声破坏等现象所导致的传统模糊C均值聚类(Fuzzy C-means,FCM)算法分割质量不佳问题,提出了基于空间信息鲁棒FCM算法,用于苗族服饰图像分割.通过均值滤波和中值滤波处理空间邻域信息,对应获得2种方法,并用一个加权参数调节模糊隶属度的稀疏性,旨在加强细节的提取和提高算法对噪声的鲁棒性.实验表明,对于被高斯噪声破坏的图像,基于均值滤波处理的改进算法,其划分系数提高约3.6％,划分熵降低约5.6％;对于被椒盐噪声破坏的图像,基于中值滤波处理的空间约束项的改进算法,划分系数提高约2.7％,划分熵降低约4.3％.该算法提高了对这类苗族服饰图像分割的质量,对于传统文化的传承具有非凡的意义.
Robust fuzzy C-mean clustering algorithm based on spatial information for segmentation of Miao costume image
Aiming at the poor segmentation quality of traditional fuzzy C-means clustering algorithm caused by damaged stains,folding marks,large color difference and noise destruction in Miao costume image,a robust FCM algorithm based on spatial information was proposed for Miao clothing image segmentation.By means of mean filtering and median filtering,two methods were obtained,and an adaptive weighted parameter was used to adjust the sparsity of fuzzy membership degree,which aimed to enhance the extraction of details and improve the robustness of the algorithm to noise.Experiments show that for images destroyed by Gaussian noise,the partition coefficient is increased by about 3.6％and the partition entropy is reduced by about 5.6％based on the improved algorithm of spatial constraint term processed by mean filter.For images destroyed by salt-and-pepper noise,the partition coefficient is increased by about 2.7％and the partition entropy is reduced by about 4.3％based on the improved algorithm of spatial constraint term processed by median filter.The algorithm in this paper improves the quality of this kind of Miao costume image segmentation,which has extraordinary significance for the inheritance of traditional culture.

Miao costume imagefuzzy C-means clusteringmean filteringmedian filteringsparsity of fuzzy membership degree

覃小素、黄成泉、彭家磊、陈阳、雷欢、周丽华

展开 >

贵州民族大学数据科学与信息工程学院,贵州贵阳 550025

贵州民族大学工程技术人才实践训练中心,贵州贵阳 550025

苗族服饰图像 模糊C均值聚类 均值滤波 中值滤波 模糊隶属度的稀疏性

国家自然科学基金贵州省省级科技计划项目贵州省研究生教育教学改革重点项目贵州省教育厅自然科学研究项目

62062024黔科合基础-ZK[2021]一般342黔教合YJSJGKT[2021]018黔教技[2022]015

2024

10.19333/j.mfkj.20230301008

毛纺科技
中国纺织信息中心 北京毛纺织科学研究所

毛纺科技

北大核心
影响因子：0.3
ISSN：1003-1456
年,卷(期)：2024.52(1)
  • 4