材料科学
抛光
光学显微镜
显微镜
光学
刮擦
边缘检测
GSM演进的增强数据速率
人工智能
光电子学
图像处理
计算机科学
复合材料
扫描电子显微镜
图像(数学)
物理
作者
Lichao Guan,Tiancai Lei,Qingfeng Yin,Hailong Cui,Jiexiong Ding
摘要
KDP (Potassium Dihydrogen Phosphate) crystal is widely used in inertia control fusion and other high-tech fields as a high quality non-linear optical material. Result of the surface quality correlating with the crystal optical properties, detecting surface defects on polished device is an essential part. In this paper, optical microscopy and image processing technology are used in the KDP crystal surface defects detection after MRF and surface cleaning. Firstly, the surface image is acquired by optical microscopy. Uneven illumination exists in the surface image, so the background extraction technology is presented to eliminate the impact of uneven illumination on the defect extraction. 2D maximum entropy threshold segmentation is applied to extract the defects. To identify residues and scratches defects, the features are utilized including the irregularity of residual defect edge, the linearity of scratch defect edge and the residue defect attached to the scratch defect shows the discontinuities and the curvature on straight edge. According to the features, canny operator is used to extract defects edge and improved straight recognition algorithm by freeman chain code is used to detect completed information of residues and scratches. Finally, the scratch defects are counted with width and length and using area to get a statistical result of the residue defects. Experimental results show that the method can accurately detect KDP crystal surface defects in different states after polishing and cleaning.
科研通智能强力驱动
Strongly Powered by AbleSci AI