| 基于改进的K-means聚类图像分割算法 |
| |
| 引用本文： | 周新建,涂宏斌. 基于改进的K-means聚类图像分割算法[J]. 无损检测, 2007, 29(5): 258-261,265 |
| |
| 作者姓名： | 周新建 涂宏斌 |
| |
| 作者单位： | 华东交通大学,CAD/CAM研究室,南昌,330013 |
| |
| 摘 要： | 介绍一种采用K—L变换、二维向量小波和改进的K—means结合的图像分割算法。阐述和分析了二维向量小波变换和Laws纹理能量测度，得出图像每个像素点可用9个纹理特性来描述的结论。运用K—means算法的思想对其进行改进，最后得出分割后的图像。试验结果证明，运用以上处理方法可显著提高分割速度和精度。
|
| 关 键 词： | 信号处理 图像分割 聚类分析 小波变换 表面缺陷 |
| 文章编号： | 1000-6656（2007）05-0258-04 |
| 收稿时间： | 2006-11-21 |
| 修稿时间： | 2006-11-21 |
Image Segmentation Based on a Modified K-means Algorithm |
| |
| ZHOU Xin-jian,TU Hong-bin. Image Segmentation Based on a Modified K-means Algorithm[J]. Nondestructive Testing, 2007, 29(5): 258-261,265 |
| |
| Authors: | ZHOU Xin-jian TU Hong-bin |
| |
| Affiliation: | Research Lab CAD/CAM, East China Jiaotong University, Nanchang 330013, China |
| |
| Abstract: | An image segmentation method combined K-L transformation, two dimensional multi-wavelet and modified K-means algorithm was introduced. Two dimensional multi-wavelet and laws texture measurement were analyzed. A conclusion was got that every pixel had nine texture measure for describing image. A modified K-means algorithm applied the thought of K-means was proposed, and the target image was acquired. The experiment results proved that the FKM method was usefull because of its accuracy and computing velocity. |
| |
| Keywords: | Signal processing Image segmentation Clustering analyzing Wavelet transform Surface defect |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
