基于多分辨率小波变换的复合材料晶粒边缘的检测

材料晶粒的边缘检测是材料三维重构的一个基础性问题,本文简要介绍了利用小波变换探测图像边缘的原理和材料SEM图像的特点,根据Mallat小波理论中图像边缘对应于小波变换模的局部极大值点的理论,采用二次B样条小波对复合材料SEM图像进行了多分辨率小波变换,从低分辨率到高分辨率逐步搜索边缘,并根据不同分辨率噪声的小波系数变化较大的特点,抑制了噪声.成功的检测出FeAl-ZrO2复合材料第二相晶粒的边缘.文章的最后给出了实验结果并且对实验结果进行了讨论.

Material Grain Edge Detection Based on Multiscale Wavelet

In this paper, the important role of edge detection in 3 dimensions reconstruction of composite is discussed, and then the principle of edge detection by using wavelet transform and the characteristic of SEM micrograph of composite briefly introduced. According to Mallat's wavelet theory that image edge corresponds to the local wavelet transform modulus maxima, a edge detection algorithm based on multiscale quadric spline wavelets transform is proposed: first formed the edge maps at several scales; second based on a top-down maxima searching and noise filter-out criterion, gave the best edge position to that obtained in the lowest scale of the multiscale wavelets. Thus we succeed in extracting single pixel contour of the second phase of the composite. Finally we present and discuss the result obtained with the FeAl-ZrO_2 composite micrograph.