臺灣博碩士論文加值系統

本文針對精密對位系統所產生的對位標記(fiducial mark, FM)影像(image)，進行有對位標記和無對位標記(unfiducial mark, UFM)之偵測與定位的分析。首先選擇影像中的對位標記作為參考對位標記(reference fiducial mark, RFM)，然後再擷取RFM上的特徵(feature)，最後再依據RFM來搜尋目標影像(target image, TI)中的FM，並定位出FM在影像上的座標(coordinate)。
本文將以影像中的FM或UFM做為自動精密對位系統的主要對位目標，並且以提高對位的精密度與效率作為本文的研究目的。有對位標記的影像中，其中參考對位標記的取得，本文提出自動候選對位標記的偵測；對位標記搜尋的方法主要以在工業界中常遇到之四種對位標記影像偵測，包括有傳統對位標記搜尋、遮蔽物十字標記的偵測、不同標記顏色之偵測以及旋轉十字標記的偵測，其搜尋方式係採用粗糙搜尋(rough search)法至細緻搜尋(fine search)法的方式，以加快搜尋的速度。無對位標記影像中沒有明顯且容易取得的標記，因此必須自行決定參考對位標記；其搜尋與定位的影像有同心圓對位標記的搜尋、直線交點標記的偵測、即時成像的直線的偵測以及標記邊緣的偵測。
本研究的實驗，除了進行自動候選對位標記的偵測之外，同時為了驗證本文定位方法的精確度與效能，也分別針對有對位標記與無對位標記影像的實驗結果，與康耐視的機器視覺系統(Cognex VisionPro)的結果做定位差之比較，實驗證明本文方法可以有效且精確將影像做定位。

In this paper, with regard to fiducial mark (FM) images produced by a precision fiducial system, we conduct analyses of fiducial mark & unfiducial mark (UFM) detection and positioning. First of all, we select the fiducial marks in an image as a reference fiducial mark (RFM). Then we capture the features on RFM. Finally, we search for FM in the target images based on RFM and locate the coordinates of FM in the images.
In this paper, we use FM or UFM in images as the main fiducial target of the automatic precision fiducial system. The purpose of the study is to improve fiducial precision and efficiency. For images with FMs, this paper proposes automatic candidate fiducial mark detection to obtain the RFMs. The searching methods of FMs mainly involve the four fiducial mark image detection often encountered in the industry; namely, traditional fiducial mark search, shelter cross mark detection, different mark color detection and rotational cross mark detection. In order to speed up the search speed, the searching methods combine rough search and fine search. NFM images do not have apparent and readily available marks. It is necessary to decide the RFM. The methods of searching and positioning in images include concentric fiducial mark search, line intersection mark detection, real-time imaging line detection, and mark edge detection.
Regarding the experiments in this study, other than automatic candidate fiducial mark detection, in order to verify the accuracy and performance of the positioning method proposed in this paper, we also compare the positioning differences of experimental results in images with FMs and UFMs to the results of Cognex VisionPro. The experiments show that the proposed method can efficiently and accurately locate positions in images.

摘　要 I
Abstract II
致　謝 IV
第一章 緒論 1
1.1 研究背景 1
1.2 研究動機 1
1.3 研究目的 2
1.4 論文架構 3
第二章 相關研究探討 5
2.1 不連續性 5
2.1.1 Sobel邊緣偵測(edge detection) 6
2.1.2 Laplacian 邊緣偵測 6
2.1.3 Canny 邊緣偵測 7
2.2 相似性 8
2.2.1 門檻值(threshold) 8
2.2.2 區域標籤 (component labeling) 8
2.3 標記比對 10
2.3.1 差值平方和(Sum of Squared Difference, SSD) 10
2.3.2 絕對誤差和(Sum of Absolute Difference, SAD) 10
第三章 標記自動偵測 12
3.1 影像預處理 12
3.2自動對位標記偵測 12
第四章 有對位標記之影像 15
4.1傳統對位標記搜尋 16
4.1.1粗糙搜尋 17
4.1.2對位標記的定位方法 18
4.1.3 細緻搜尋 19
4.2遮蔽物十字標記的定位 20
4.3不同標記顏色之偵測 21
4.4旋轉十字標記偵測 21
第五章 無對位標記之影像 23
5.1同心圓對位標記搜尋 23
5.1.1 內圓偵測與切割 24
5.1.2 外圓偵測與切割 25
5.2直線交點標記偵測 27
5.3即時直線偵測 28
5.4標記邊緣偵測 30
5.4.1 雜訊邊緣偵測 30
5.4.2底部邊緣偵測 32
5.4.3大面積樣本之邊緣偵測 33
第六章 實驗結果 34
6.1參考對位標記之自動偵測 34
6.2對位標記之影像資料庫 35
6.2.1自行設計圖庫 35
6.2.2全研科技實驗影像圖庫 36
6.3有對位標記之搜尋 39
6.3.1傳統對位標記搜尋 39
6.3.1.1對位的精密度 39
6.3.1.2對位系統的效能 41
6.3.2遮蔽物十字標記的定位 43
6.3.3不同標記顏色之偵測 46
6.3.4旋轉十字標記偵測 48
6.4無對位標記之搜尋 49
6.4.1同心圓對位標記搜尋 49
6.4.2直線交點標記偵測 50
6.4.3即時直線偵測 53
6.4.4標記邊緣偵測 55
6.4.4.1雜訊邊緣偵測 55
6.4.4.2底部邊緣偵測 56
6.4.4.3大面積樣本之邊緣偵測 56
第七章 結論 57
參考文獻 58

[1]M. Tichem and M. S. Cohen, “Subμm Registration of Fiducial Marks Using Machine Vision”, IEEE Transactions on Pattern Analysis and Machine Intelligence”, Vol. 16, No.8, pp.791-794, 1994.
[2]H. Keith 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, No.1, pp.17-30, 1988.
[3]X. Fernandez, and J. Amat, “ Research on Small Fiducial Mark Use for Robotic Manipulation and Alignment of Ophthalmic Lenses ”, pp.1143-1146, , 1999.
[4]N. Guil, J. Villalba, and E. L. Zapata, “A Fast Hough Transform for Segment Detection”, IEEE Image Processing, Vol. 4, No.11, pp.1541-1548, 1995.
[5]S.K. Tsau, D.Y. Hong, H.W. Lee , C.M. Ming Chang and C.H. Lin, “Multiple Alignment Stage for the Automatic Precision Alignment System“, International Symposium on Computer, Consumer and Control,pp.926-929,2012 .
[6]Y.C.Lin, Y.Y.Chiu, H.W.Lee, B.Y.Jhan and C.H.Lin, “The Study of Automate Locate Special Fiducial Marks,” The Sixth International Conference on Genetic and Evolutionary Computing,, Japan ,pp.51-54,2012 .
[7]C.H. Lin, Y.C.Lin, K.H Teng, “Detection System for Multiple Fiducial Marks”,全研論文獎,F-M05,2012.
[8]R. C. Gonzalez, and E. W. Richard, “Digital Image Processing”, Prentice-Hall, 2002.
[9]J. F. Canny, “A Computational Approach to Edge Detection”, IEEE Transaction on Pattern Analysis and Machine Intelligence, Vol. 8, No. 6, pp. 679–698, 1986.
[10]L. Ding and A. Goshtasby, “On the Canny Edge Detector”, Pattern Recognition, Vol. 34, pp. 721–725, 2001.
[11]C.H.Lin, Y.K.Chan, C.C.Chen, “Detection and Segmentation of Cervical Cell Cytoplast and Nucleus,” International Journal of Imaging Systems and Technology, Vol. 19, pp.260–270, 2009.
[12]P., F.A. Vanzella, W. Torre, and V., “Edge Detection Revisited,” IEEE transactions on systems, man, and cybernetics-part B: CYBERNETICS, Vol. 34, NO. 3, pp.1500–1518, 2004.
[13]C.H.Lin and Y.J.Syu, “Fast Segmentation of Porcelain Images Based on Texture Features,” Journal of Visual Communication and Image Representation, Vol. 21, pp.707–721, 2010.
[14]J. Haddon and J. Boyce, “Image segmentation by unifying region and boundary information,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 12, No. 10, pp. 929-948, October, 1990.
[15]C.H.Lin and C.C.Chen, “Image Segmentation Based on Edge Detection and Region Growing for ThinPrep-Cervical Smear,” International Journal of Pattern Recohnition and Artificial Intelligence, Vol. 24(7), pp.1061–1089, 2010.
[16]Z. Hou, Q. Hu and W. L. Nowinski, “On minimum variance thresholding”, Pattern Recognition Letters, Vol. 27, pp. 1732–1743, 2006.
[17]F. Y. Shih and S. Cheng, “Automatic seeded region growing for color image segmentation”, Image and Vision Computing, Vol. 23, pp. 877–886, 2005.
[18]M., D. and Shah, J., “Optimal approximations by piecewise smooth function and associated variational problems”, pp.577–684, 1898.
[19]R. B and O. R, ”2D feather detection via local energy “, Image and Vision Computing . pp.353-368, 1997.
[20]T. M and H. M, ”Fast corner detection “, Image and Vision Computing ,Vol. 16 pp.75-88,1998.
[21]M.Z. Brown, D. Burschka and G.D. Hager,” Advances in computational stereo”, vol. 25, pp. 993-1008, 2003.
[22]D.Scharstein and R.Szeliski, “A Taxonomy and Evaluation of Dense Two-Frame Stereo Correspondence Algorithms,”International Journal of Computer Vision archive, vol.47(1-3), pp.7-42, 2002.
[23]S.A. Hojjatoleslami and J. Kittler, “Region Growing: A New Approach”, IEEE Transaction Image Processing, vol. 7, no. 7, pp. 1079-1084, 1998.
[24]視覺軟體VisionPro-康耐視視覺工具可搭配使用攝影機、板卡和周邊設備,美商康耐視股份有限公司台灣分公司, http://tw.asiamachinery.net/onlineExpo/product_detail.asp?id=DLOG2013&ProID=1671, 搜尋日期：2013年8月21日。