Abstract: The effective and fast image processing techenique is important for surface disfigurement dectecting system based on machine-vision for arc segments ceramic magnet.Because mathematical morphology takes on characteristics of making image data simple and making parallel processing easy，this paper provides a method of detecting surface disfigurement with mathematical morphology for arc segments ceramic magnet.By choosing suitable structrue elements，image data with surface disfigurement are operated with methods of morphology and then the outlines of disfigurement are extracted.The test results indicate that this method is effective，and takes on high precision ratio and fast computing processing.

Key words: arc segments ceramic magnet, disfigurement dectecting, mathematical morphology, image processing

摘要： 高效、快速的磁瓦图像处理方法是基于机器视觉技术磁瓦表面缺陷检测系统的关键技术之一。由于数学形态学具有简化图像数据，保持它们基本的形状特性，同时天然具有并行实现的结构特点。提出了一种将数学形态学运用到磁瓦表面缺陷检测中的方法。通过选择合适的结构元素，对缺陷图像进行多种形态学运算，提取出缺陷轮廓，并作为缺陷面积计算的基础。实验结果表明，该方法效果良好，缺陷的检测准确率高，计算处理的速度快。

关键词: 磁瓦, 缺陷检测, 数学形态学, 图像处理