摘要
为解决目标相互粘连的金相图像分类问题,提出了一种基于小波分解和分形理论的金相图像识别方法。该方法通过对原图像作一级小波变换,分别提取小波变换后的各输出子图像的分形维数和相关维数构成纹理特征值,对金相图像进行分类识别,结果表明取得了较好的效果。
A method of metallographic image recognition based on wavelet decomposition and fractal theory was presented The aim of the method was to solve the difficulty problem of metallographic image, in which objects were hardly isolate. After a tree-structured wavelet transform was employed, the features of texture, which were composed of fractal dimensions and relative dimensions, were extracted from the each frequency channel of wavelet transform output, the experiment of recognizing metallographic had been done and the result was satisfactory, which verified the effect of the method.
出处
《理化检验（物理分册）》
CAS
2008年第4期184-188,共5页
Physical Testing and Chemical Analysis(Part A:Physical Testing)
关键词
金相图像
小波分解
分形
纹理特征
Metallographic image
Wavelet decomposition
Fractal theory
Texture feature
