利用基于小波的尺度共生矩阵进行纹理分析

摘要

In this paper, we proposed a new method for image feature extraction within scale space. The new method captures the relation of features between different scales, but not the features within a single scale space. Compared with the traditional methods, the proposed method can represent the scale property of texture better. In practice, we first construct a scale-based concurrent matrix (SCM) which reflects the relation between different scales; and then using the matrix calculate some useful measurements as the features for texture classification. Experiments also show that the proposed method can get more accurate results for texture classification than the traditional texture classification methods.%提出了一个在尺度空间提取特征的新方法，它的特点是提取尺度之间的依存关系，而非尺度间的独立特征，和传统方法相比，它更全面、更准确地刻画了纹理的尺度特性。具体做法是，首先构造一个反映尺度依存关系的矩阵(本文称之为尺度共生矩阵)，然后在此基础上进行特征提取分类。实验结果表明：用基于尺度共生矩阵的分类方法可以得到较好的分类结果。