摘要
多进制小波分析是二进制小波理论的推广和延伸,冈此其在很多方面具有比二进制小波更优良的特性.本文提出了一种基于多进制小波变换的纹理特征提取方法,通过对小波系数的标准差作为纹理测度以生成特征向量,利用C-均值聚类算法进行纹理分割.实验结果好于二进制小波.
M-band wavelet analysis is the extension of 2-band wavelet theory, and there is more detailed information than that based on 2-band wavelet. In this paper, a novel method on texture feature extraction based on M-band wavelet transformation is presented. The Feature Vectors are calculated by using standard deviation of the wavelet coeffcient, then we use the C-means clustering to segment the texture image directly. The experimental results show the efficiency of M-band wavelet better than it of 2-band wavelet.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
2004年第3期286-290,共5页
Pattern Recognition and Artificial Intelligence
基金
国家自然科学基金(No.60172067)
广东省自然科学基金重点项目(No.036608)
广州市科技计划(No.2003J1-C0201)资助项目
关键词
小波变换
多进制
特征提取
纹理分割
C-均值聚类
Wavelet Transformation
M-band
Feature Extraction
Texture segmentation
C-means Clustering
