高速高解析度觸控面板玻璃瑕疵光學檢測系統設計

本研究致力於開發一套透明觸控面板玻璃表面瑕疵自動化光學檢測系統，以取代當前生產線上仍以人工目視為主流之檢測方式，所開發完成的系統可提供人眼無法檢視之高解析度檢測需求及穩定的檢測品質，達到高生產品質規格要求與降低人事成本等目標。本研究建立了多個通用型視覺檢測模組化單元子系統，各單元可依照解析度規格需求進行調整，亦可整合數個子系統應用於大尺度玻璃面板表面瑕疵全域檢測，系統使用黑白線型互補式金屬氧化物半導體元件（Complementary Metal-Oxide-Semiconductor，CMOS）擷取大面域檢測影像，結合自行研究開發之高亮度線性光源構成光機檢測探頭模組，採用陣列式排列高功率發光二極體，此探頭最高可達到1.75微米水平解析度之檢測設備；軟體以圖形處理器（Graphics Processing Unit，GPU）為檢測計畫提高效率，利用GPU的平行運算技術分割圖像進行自應性影像處理，可大幅提高大數據運算處理之效率與智慧化功能。　　本研究透過分析GPU分配資源使編程達到運算效率最大效率，在解析度3.5微米下最大檢測面域 43*229 mm，影像數據量逾800M個像素點，智慧化選擇參數可達到7.325s的運算效率，若採用使用者輸入參數可達到3.731s的運算效率。

關鍵字

自動化光學檢測； GPU ；平行運算；觸控面板玻璃

並列摘要

The present study develops an automated optical inspection system for detection of surface defects in touch panel glass surface. High resolution and repeatability for quality in-line inspection is achievable at reduced personnel costs. Multiple visual inspection units with adjustable optical resolution are offered by the developed system to suit different inspection requirements. Therefore, integration of several units can be implemented for large-scale defect detection of glass panel surface. The optical inspection unit uses single line CMOS camera to capture images. High-luminance light source consisting of closely arrayed parallel high-power LED is developed to provide uniform luminance on the surface of the sample. Each inspection unit can have up to 1.75 μm horizontal resolution. To improve the software efficiency for detection, this project uses graphical processing unit (GPU) for parallel computing. It helps computing to become fast and smart. The detection system can achieve maximum efficiency by analyzing the resource distribution of GPU. For a resolution of 3.5 μm, the maximum detection surface of present inspection unit is 43 mm x 229 mm, which is equivalent to an image data of over 800 megapixels. Using smart parameter settings, the detection system finishes computing in 7.325 s. However, if the parameter settings are set by the user, the operational efficiency can be improved to reach 3.731 s.

並列關鍵字

AOI ； GPU ； parallel computing ； tough panel glass

參考文獻

[4] E. Lindholm, J. Nickolls, S. Oberman, & J. Montrym, “NVIDIA Tesla: A unified graphics and computing architecture,” Proceedings of the IEEE Micro, Vol. 28, Issue 2, Pages 39-55, 2008.

[7] Lei Pan, Lixu Gu, and Jianrong Xu, “Implementation of medical image segmentation in CUDA,” Paper presented at the 5th Int. Conference on Information Technology and Applications in Biomedicine, ITAB 2008 in conjunction with 2nd Int. Symposium and Summer School on Biomedical and Health Engineering, pp.82~85, 2008

[9] Hong-Dar Lin and Duan-Cheng Ho, “Detection of pinhole defects on chips and wafers using DCT enhancement in computer vision systems,” International Journal of Advanced Manufacturing Technology, Vol.34, No.5-6, pp.567-583, 2007.

[10] Jung-Hun Kim, Suk Ahn, Jae Wook Jeon, and Jong-Eun Byun, “A high speed high-resolution vision system for the inspection of TFT LCD,” IEEE International Symposium on Industrial Electronics, Vol.1, pp.101-105, 2001.

[11] Kyu-Bong Lee, Min-Seok Ko, Joon Jae Lee, Tak-Mo Koo , and Kil-houm Park, “Defect detection method for TFT-LCD panel based on saliency map model,” TENCON 2004, IEEE Region 10 Conference, pp.223-226, 2004.

國際替代計量

高速高解析度觸控面板玻璃瑕疵光學檢測系統設計

全文下載

主題瀏覽