摘要
应用扫描电子显微镜（ＳＥＭ）立体对成像技术和图像处理、计算机视觉方法建立了一套微结构形貌三维自动重建系统。在Ｂａｒｎａｒｄ算法基础上根据ＳＥＭ立体对成像的特点提出了一种基于边界特征的对应点匹配概率松弛迭代方法。系统中融合了基于区域的立体匹配原理，检出的特征定位准确。匹配过程引入多种几何及边界强度和位向约束，具有较高的匹配效率和精度。应用本系统进行了典型金属断裂表面的三维重建。
An autonomous system for the reconstruction of 3D topography of micro structures was established by the means of SEM stereo pairs,image processing and computer vision.Based on the principles of SEM stereo projection and Barnard's method for stereo matching,an edge feature based iterated relaxation corresponding algorithm was proposed.The principles of area based matching,epipolar constraint,edge strength and orientation constraints were introduced into the corresponding process.After the exact detection of the image edge feature,the results of correspondation and reconstruction were proved to be with high efficiency and good precision.Some typical fracture surfaces were reconstructed by this reconstruction system and further quantitative analysis can be made with these reconstruction results.
出处
《电子显微学报》
CAS
CSCD
1997年第1期49-56,共8页
Journal of Chinese Electron Microscopy Society
基金
国家自然科学基金
清华大学摩擦学国家重点实验室资助
关键词
三维重建
SEM立体
图像处理
扫描电镜
D reconstruction SEM stereo pairs image processing computer vision fracture surface
