【摘要】 以半导体材料单晶生长为背景,利用图像处理与分析技术对单晶生产过程进行监测研究。晶体生长是具有相变与移动边界的复杂分布参数系统,但是单晶炉所采用的监测控制手段仍然是相当基本的,以至阻碍了高纯度、高均匀性、高完整性和大直径半导体单晶材料的生产。在对现场摄取的单晶生长图像序列进行边缘检测和二值化等预处理的基础上,优化设计了自动阈值分割算法,有效地提取了几何特征量,直观给出了单晶生长固体-液体界面的图像信息。对计算机视觉检测和分析应用于单晶生产作了有益的探索。更多还原

【Abstract】 Based on crystal growth of semi-conductor materials, and the producing process is monitored by the technology of image processing and analyzing. Crystal growth is a system of complicatedly distributed parameters. The detecting and controlling methods used in signal crystal furnace have been relatively low leveled, the large diameter crystal production of semiconductor materials of high purity, homogeneity and integrity becomes difficulty. Based on the edge detection and binarization of series crystal growth images shot in-situ, the automatic threshold segmentation algorithm is optimized to extract effectively the geometric eigenvalues, and the image information of solid-liquad interface is showed directly. The text makes the beneficial quest in the production of crystal by the use of computer vision measure and analyzing.更多还原