A Robust Method With Adaptive Inpainting for Edge Detection of Single Silicon Crystal

• Published in: IEEE sensors journal Vol. 23; no. 5; pp. 4500 - 4507
• Main Authors: Zhang, Xinyu, Liu, Ding, Lin, Pengyi, Ji, Ruirui
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.03.2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Crystal growth; Crystals; curvature driven diffusions; Diameters; Differential equations; Differential operators; Edge detection; edge inpainting; fractional differential operators; Image edge detection; Integrity; Maintenance engineering; Operators (mathematics); Silicon; Single silicon crystal growth; Temperature measurement
• ISSN: 1530-437X; 1558-1748
• DOI: 10.1109/JSEN.2022.3176682

The improvement of the edge detection accuracy and integrity of the single silicon crystal and bright halo is the key to measuring the crystal diameter. This paper proposes an edge detection method with adaptive image inpainting. Firstly, according to the image characteristics, an evaluation function is designed to obtain the optimal order of fractional differential operators for the Canny method. Then, the initial edges of the single silicon crystal and bright halo are obtained with the Canny method based on these operators, and many details of the edges can be obtained. Meanwhile, an improved curvature driven diffusions (CDD) model is proposed for repairing the initial edge, which introduces the gradient information of edge into the CDD model to repair the edge breaks in the initial edges and enhance the edge integrity. Finally, a series of experiments are conducted to validate the effectiveness and performance of the proposed method on the images of practical engineering applications. The experimental results show that the proposed method performs significantly better than the existing methods in the edge detection accuracy and integrity of the single silicon crystal and highlight halo.