摘要
本文针对纹理图像的特点，提出了基于主能量通道子波变换的纹理图像多级分割方法。该方法以最大能量为线索，建立起由细到粗的多分辨率子波分解金字塔，分割按照其逆过程，即由粗到细，每个分辨率上分割特征的优化选择、分割方案—模糊Ｃ聚类的确定，都有利于纹理图像的分割，最后由低向高传递分割结果，从而得到较精细的分割图像。
For textured image,this paper suggests a hierarchical segmentation scheme based on wavelet transform performed on the sub band image with the most power.The scheme includes constructing a fine coarse multiresolution wavelet pyramid by energy and segmenting textured image by the inverse process of wavelet decomposition,i.e,the process from coarse to fine resolution.At every resolution,the feature optimal selection and using of Fuzzy C clustering are all helpful to segmentation.Finally,the segmentation information at the low level is propagated to the next higher level of resolution.Thus,a fine segmentation result of textured image can be obtained.
出处
《系统工程与电子技术》
EI
CSCD
1998年第8期23-27,共5页
Systems Engineering and Electronics
关键词
图像处理
特征选择
滤波器
纹理分割
Wavelet transform,Texture segmentation,Feature selection,Fuzzy C clustering.
