首页 | 本学科首页   官方微博 | 高级检索  
     

基于图像纹理和小波变换的聚类分割方法
引用本文:张勇,周文卫,陈立杨,石高峰. 基于图像纹理和小波变换的聚类分割方法[J]. 自动化信息, 2010, 0(7): 48-49,63
作者姓名:张勇  周文卫  陈立杨  石高峰
作者单位:西南交通大学电气工程学院,四川省成都市610031
摘    要:针对纹理图像,本文提出了一种基于图像纹理特征的非学习分割方法。采用小波变换和快速k-means聚类分割算法,减少了整个处理过程的运算量。为了保证分类算法的精确性,运用了总体流量变化最小(Total Variation Flow)[1]非线性去噪方法对图像进行预处理,从而将减小图像噪声污染带来的分割误差。在图像特征的提取上,运用Gabor滤波器原理生成滤波空间,并让图像通过滤波空间而生成特征向量空间。通过制定一个快速寻优策略,从而达到分割图像的目的。

关 键 词:非线性去噪  特征提取  图像分割  聚类分割算法  小波变换

A Method of Image Clustering Segmentation Based on Image Texture and Wavelet Transform
ZHANG Yong,ZHOU Wen-wei,CHEN Li-yang,SHI Gao-feng. A Method of Image Clustering Segmentation Based on Image Texture and Wavelet Transform[J]. Automation Information, 2010, 0(7): 48-49,63
Authors:ZHANG Yong  ZHOU Wen-wei  CHEN Li-yang  SHI Gao-feng
Affiliation:(School of Electrical Engineering, Southwest Jiaotong University, Chengdu 610031, China)
Abstract:Aiming at textured images, the authors propose a non-learning image segmentation method based on image texture fea- tures in this paper. The computation workload of the whole process is reduced by means of wavelet transform and fast k-means clustering segmentation algorithm. In order to ensure the accuracy of classification algorithm, a non-linear image-denoising method of " Total Variation Flow minimality" is adopted to implement the image-preprocessing, thus the segmentation error which is caused by image noise pollution will be decreased. In the image feature extracting process, a filtering space is generated by using Gabor filters principle and the test images will generate the feature vector space through the filtering space. A fast optimization searching strategy is formulated to achieve the purpose of image segmentation.
Keywords:Non-linear Denoising  Feature Extraction  Image Segmentation  Clustering Segmentation Algorithm  Wavelet Transform
本文献已被 维普 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号