摘要
介绍了常用的图像分割方法及其优缺点和金相图像分析中的图像分割方法。针对金相图像分割中存在的问题,提出了两步分割的图像分割方法。这种方法是先用阈值分割法把目标从背景中分割开来,然后采用数学形态学、打孔、找凹点和连分割线的方法进行粘连分割。与已有的金相图像分割方法相比,这种方法具有较强的强健性、自适应性和非监督性。
In this paper, the traditional methods of image segmentation and their advantages and disadvantages are introduced first. Then the methods of image segmentation for metallographic analysis are introduced. To solve the problems existing in metallographic analysis, a two-step method of image segmentation is proposed. In the method, the objects are segmented first by thresholding. Then the joint or overlapped objects are separated by mathematic morphology, hole opening, concave point finding, and cutting line determination. Compared with the existing segmentation methods in metallographic analysis, this method is robust, adaptive, and unsupervised.
出处
《理化检验（物理分册）》
CAS
2005年第5期236-239,共4页
Physical Testing and Chemical Analysis(Part A:Physical Testing)
关键词
金相图像分析
图像分割
数学形态学
阈值分割
粘连分割
Metallographic image analysis
Image segmentation
Mathematic morphology, Thresholding separation
Join separation
