摘要
Ｍａｌａｔ非正交小波变换通常用于光滑图像的边缘提取，本文将其改进后，推广到图像纹理特征的提取和纹理图像的分割，取得了良好的效果．基于小波变换的纹理图像分割的算法中，小波变换尺度数的选取和纹理聚类数的确定是其难点．对此，本文作了详细的讨论，提出了一些独特、有效的解决方法．
Gradient wavelet transform is usually used to extract edges in an image. It is improved in order to extract image features, which are used to do texture segmentation in this paper. Experiments show that the algorithmis performed well. It is necessary but difficult to give the 010 number of wavelet scales and the number of clusters. We study the problem carefully and give some special and effective approaches to decide the two numbers.
出处
《东南大学学报（自然科学版）》
EI
CAS
CSCD
1999年第1期44-48,共5页
Journal of Southeast University：Natural Science Edition
