摘要
基于高纯度的熔融石英坩埚的良好特性,在太阳能行业单晶硅制作的环节得到广泛应用,透明层气泡的大小以及分布规律等特征极大的影响了石英坩埚的使用周期和性能。主要研究距离石英坩埚内壁表面0.72mm处的气泡含量,首先提取彩色图像的R,G,B通道,分别对其进行基于可变阈值的细胞学习自动机边缘检测,获取气泡边缘图像,其次对气泡进行填充并应用形态学算法,最终通过分析气泡结构分别对非重叠和重叠气泡进行计数统计。并与canny算子、sobel算子和N.N.Misra提出的显微计算机视觉算法等检测结果进行对比。结果表明基于可变阈值的细胞学习自动机边缘检测算法在坩埚内壁气泡计数方面效果更优,准确率高达97.16%。
Based on the good characteristics of high purity molten Shi Ying crucible, it is widely used in the production of monocrystalline silicon in solar industry. The size and distribution of bubbles in transparent layer greatly affect the service life and performance of Shi Ying crucible. This paper mainly studies the bubble content at a distance of 0.72 mm from the inner wall surface of the Shi Ying crucible. Firstly, the R, G and B channels of the color image were extracted, and the edge detection of the cell learning automata based on variable threshold was carried out respectively to obtain the bubble edge image. Secondly, the bubble was filled and morphological algorithm was applied. Finally, the non-overlapping and overlapping bubbles were counted and counted respectively by analyzing the bubble structure. The detection results were compared with canny operator, Sobel operator and micro-computer vision algorithm proposed by N.N. Misra. The results show that the edge detection algorithm of cell learning automata based on variable threshold is more effective in counting bubbles on the inner wall of crucible, and the accuracy rate is as high as 97.16%.
作者
赵谦
史凌云
ZHAO Qian;SHI Ling-yun(School of Communication and Information Engineering,Xi*an University of Science and Technology,X i,an Shanxi 710054,China)
出处
《计算机仿真》
北大核心
2020年第5期429-433,共5页
Computer Simulation
基金
陕西省科技计划工业攻关项目(2017GY-073)
西安市碑林区应用技术研发项目(GX1811)。
关键词
可变阈值
细胞学习自动机
连通域
石英坩埚
气泡
Variable threshold
Cellular learning automata
Connected domain
Quartz crucible
Bubble
作者简介
赵谦(1977-),男(汉族),陕西省西安市人,副教授,硕士研究生导师,主要研究领域为图像处理、视频目标跟踪;史凌云(I997-),女(汉族),陕西省西安市人,硕士研究生,主要研究领域为图像处理。
