摘要
局域短裂纹数统计是疲劳短裂纹扩展研究的关键技术之一,本文提出并实现了一种基于分水岭算法的局域短裂纹数统计方法。针对传统分水岭算法容易产生过分割的缺点,在分析含短裂纹群材料试件锥束CT图像中短裂纹群的形态特征的基础上,综合运用一系列数学形态学方法进行预处理,有效抑制噪声,增强图像对比度,提取标记;之后,运用距离变换进一步优化预处理结果,最后,使用标记控制的分水岭算法实现分割并统计局域短裂纹数。实验结果证明,该算法能有效地抑制过分割现象,得到较好的统计结果,是实用、准确和有效的。
Measuring the number of short fatigue cracks is the key technique in the research of fatigue cracks development. A method for measuring the number of short fatigue cracks number based on watershed segmentation is presented in this paper. Aiming at the over segmentation problem existing in traditional watershed segmentation, we first analyzed the shape features of short fatigue cracks in cone beam volume CT image. Than a preproeessing algorithm that use many different morphology functions in combination is proposed to effectively reduce the noise, enhance the image contrast and get the marker. After that, a distance transform is performed to further optimize the preproeessing result. Finally, a marker-controlled watershed segmentation algorithm is used to measure the number of short fatigue cracks. Experimental results show that the proposed method effectively reduces the over segmentation and is accurate and effective.
出处
《机械科学与技术》
CSCD
北大核心
2007年第9期1126-1129,共4页
Mechanical Science and Technology for Aerospace Engineering
