Image Processing-Based Crystal Diameter Detection for Czochralski Method

In order to real-time and accurately detect the crystal diameter in making process, the diameter detected method in making monocrystal with Czochralski method has been analyzed, and the process for detecting the diameter of monocrystal has been divided into three stages, crystal guiding stage, crystal transitional guiding stage, and equal-diameter growing stage. Furthermore, the edge feature information detection algorithm has been proposed, and the circle fitting method according to detected feature information has been put up to get the crystal diameter. Moreover, the application software to detect monocrystal diameter has been developed in light of the detection principle and method proposed in this paper. At last, some experiments have been done with the developed software, the experiment data have been analyzed, and the result indicates that the average error of the detected crystal diameter in the same stage is in the range of five pixels, so the method proposed in this paper is feasible.