Abstract Quartz crucible is a widely used manufacturing equipment in silicon crystal manufacturing industry, and the detection of its quality has a great influence on the purity of subsequent silicon crystals. In this paper, a machine vision-based bubble detection statistical system for transparent layer of quartz crucible is designed, and an intelligent algorithm for measuring the diameter, area and number of bubbles is proposed. Firstly, the bubble image is filtered and enhanced pre-processing operation to enhance the difference between the bubble and the background, and then the binary contour of the bubble is extracted by using a bubble contour extraction algorithm that fuses the morphological algorithm. In order to improve the accuracy of bubble detection, a curvature-based corner detection algorithm is proposed for overlapping bubbles.The experimental results show that the algorithm can accurately and real-time detect the number and diameter of bubbles. Compared with the connected component algorithm(CCA) and edge chain code information (EDCI) algorithm, the proposed method has lower cost and higher detection efficiency.

中文翻译：

---

基于图像处理的石英坩埚透明层微气泡统计检测方法研究

摘要 石英坩埚是硅晶制造行业广泛使用的制造设备，其质量的检测对后续硅晶的纯度影响很大。本文设计了一种基于机器视觉的石英坩埚透明层气泡检测统计系统，提出了一种测量气泡直径、面积和数量的智能算法。首先对气泡图像进行滤波和增强预处理操作以增强气泡与背景的差异，然后利用融合形态学算法的气泡轮廓提取算法提取气泡的二值轮廓。为了提高气泡检测的准确性，提出了一种基于曲率的气泡重叠角点检测算法。实验结果表明，该算法能够准确、实时地检测气泡的数量和直径。与连通分量算法（CCA）和边缘链码信息（EDCI）算法相比，该方法具有更低的成本和更高的检测效率。