多元對位標記影像之快速對位與檢測技術

過去十幾年來，自動對位技術已經被廣泛用於各種工業領域。根據常見的對位狀況，本研究將對位影像分為「有對位標記」與「無對位標記」兩種類型，再以這些影像做為自動精密對位系統的主要處理目標。本文中主要提出自動對位標記偵測與對位標記搜尋的方法。針對自動化對位系統，特殊的對位標記通常會由使用者自行擷取做為後續對位基準標記，而自動對位標記偵測係自動偵測對位影像中較具唯一性之特徵做為參考標記，使得目標對位元件不需額外設計任何專門做為對位用途的標記進行對位；對位標記搜尋則是提出粗糙搜尋(Rough Search)法至細緻搜尋(Fine Search)法，可以加快搜尋的速度並以子像素搜尋(Sub-Pixel Search)演算法進行對位標記的微調提昇對位標記對位的精密度。根據影像特性不同還另外提出自動十字對位標記對位演算法及直線交點座標之自動對位。本研究以Visual C# 2012 及EmguCV 2.4.9為開發環境，建立出一套可以由使用者選取參考對位標記，自動依程序取像、計算及傳送座標偏移量資訊之對位系統，進行對位標記即時對位實驗。在本文實驗透過人工目視對位及視覺系統(Vision Pro)對位結果的量測證明本研究方法的執行效率最佳。預計針對目前LCD製程產業將能有效幫助降低缺及製造成本。

關鍵字

粗糙搜尋；細緻搜尋；子像素搜尋；對位標記；十字標記對位；自動對位

並列摘要

Over the past decade, the technology of automatic alignment has already widely used in industry. According to the common alignment situations, the digital images can be divided into two types： “image with fiducial marks" and " image without fiducial marks". We take these images as the objectives of this paper for Automatic Alignment System. In this paper we present an automated detection of fiducial marks and fiducial marks searching method. For Automatic Alignment System, special marks are commonly referred to as fiducial mark extracted by user. Automated detection of fiducial marks is to detect the unique feature as the reference mark on the image. It can make no additional special design at the fiducial panels for the purpose of alignment； Fiducial marks searching method is the Rough Search to Fine Search, it can make searching process more efficient. Sub-Pixel Search is also proposed to improve the precision of the alignment. We also proposed the crossmark automatic searching methods and the automatic alignment methods based on a point where lines intersect by considering content-based characteristics of various images. In our work, we build an Automatic Alignment System with Visual C# 2012 and EmguCV 2.4.9 for real-time experiment. Fiducial mark can be chosen by user conveniently. Image is captured autimaticaly for coordinate calculation and offset information transfer. In our experiments the alignment results which is measured by artificial visual alignment and Vision Pro process show the best performance. The results of this work are expected to aid in optimization of the LCD manufacturing process to minimize defects and the cost of goods manufactured.

並列關鍵字

Rough Search；Fine Search；Sub-Pixel Search；fiducial mark；Crossmarkalignment；Automatic alignment

參考文獻

[6] 全研科技股份有限公司, XXY平台型錄規格表, 彰化, 台灣全研科技股份有限公司.

[2] R. C. Gonzalez and R. E. Woods, “Digital Image Processing”, Prentice-Hall, 2002.

[3] C. H. Lin, Z. Ying ,and W. Changyun, ”Visual Servo Auto-focusing using Recursive Weighted Least-Squares for Machine Vision Inspection”, Electronics Packageing Technology Conference, 2003.

[4] D. M. Tsai, M. C. Lin, “Machine-vision-based identification for wafer tracking in solar cell manufacturing”, Robotics and Computer-Integrated Manufacturing, vol.29, pp.312-321, 2013.

[7] H.K. Nishihara and P.(A) Crossley, “Measuring Photolithographic Overlay Accuracy and Critical Dimensions by Correlating Binarized Laplacian of Gaussian Convolutions”, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol.10(1), pp.17-30, 1988.

國際替代計量

多元對位標記影像之快速對位與檢測技術

全文下載

主題瀏覽