基于颜色和纹理特征聚类的彩色图像分割  

Color image segmentation based on color and texture feature clustering

作　　者：赵蕾 刘本永 ZHAO Lei;LIU Benyong(College of Big Data and Information Engineering,Guizhou University,Guiyang 550025,China)

机构地区：[1]贵州大学大数据与信息工程学院,贵阳550025

出　　处：《智能计算机与应用》2024年第4期251-254,共4页Intelligent Computer and Applications

基　　金：国家自然科学基金(60862003)。

摘　　要：针对传统彩色图像分割算法在轮廓模糊和纹理丰富区域分割效果差的缺点,本文提出基于颜色和纹理特征聚类的彩色图像分割算法。首先,采用基于图的图像分割算法多次分割图像,以得到多组超像素块;其次,提取超像素块的颜色特征和纹理特征,将其融合为一个特征向量,并使用k-means聚类对每组超像素块的特征向量聚类,以获得多组分割结果;最后,使用线性组合的方法融合多组分割结果,得到最终的分割图像。在公开数据集BSD500上与经典聚类算法SFFCM、AFCF相比较,实验结果表明本算法优于这些经典算法。Aiming at the shortcomings of traditional color image segmentation algorithms in segmentation effect when an image contains areas with blurred contour and rich texture,an algorithm based on color and texture feature clustering is proposed.Firstly,the graph-based image segmentation algorithm is used to segment an image several times to get multiple groups of superpixel blocks.Secondly,the color feature and texture features of each superpixel block are extracted and fused into a feature vector,and the k-means clustering method is used to cluster the feature vectors of each group of super-pixel blocks to obtain multiple groups of segmentation results.Finally,the method of linear combination is used to fuse multiple groups of segmentation results to get the final segmented image.Compared with the classical clustering algorithms SFFCM and AFCF on the public data set BSD500,the experimental results show that this proposed algorithm outperforms the classical ones.

关 键 词：彩色图像分割 颜色特征 纹理特征 超像素 K-MEANS聚类 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]