本研究藉由機器視覺配合影像處理方法，發展一套應用於加工平台上的光學影像定位與物件量測系統，此系統具有以下幾個主要功能：
(1)影像定位：利用指定的定位點，配合跨影像特徵點搜尋與匹配，計算得到特定特徵或加工處的定位位置；(2)物件特定尺寸量測：利用影像處理取出特定物件邊緣，配合跨影像特徵點搜尋與匹配，計算處理得到物件尺寸；(3)彈性基準定位：由使用者指定物件上所需要的加工基準位置，配合特徵搜尋匹配，達成可適應多次平台移動的來回定位。
本研究使用一個高倍率（最大4.5倍）的可見光相機，搭配XYZ三軸移動平台，使用校正片以獲取真實尺寸的參考比例尺，拍攝物件為多個不同尺寸的塊規。首先進行相機對焦，對不同焦距的校正片影像進行傅立葉轉換(FT)至頻率域後，由頻譜圖計算得到最佳焦距。然後透過固定倍率下已知的條紋間距，與影像尺寸換算後獲取真實尺寸影像比例尺。接著移動平台拍攝多張物件影像，選定基準影像中的定位點後，使用SURF(Speeded Up Robust Features)特徵演算法，搜尋不同張影像中的特徵點，跨影像進行匹配，達成跨影像定位。特定尺寸量測方面，利用影像處理方法將欲得知的輪廓幾何形貌取出後，再由特徵點匹配所得到物件移動資訊，相配合計算即可得到尺寸。藉由上述步驟，本研究提出一個達微米精度等級的系統，並且具有適應真實加工情況的能力，測試結果在4 mm的樣本平均量測誤差為1.6 μm，10 mm的樣本平均量測誤差則為3.3 μm。

In this study, we develop a set of optical image localization and object measurement systems applied to the processing platform by computer vision and image processing. The system has the following main functions:(1) image positioning: the use of the specified positioning point, with the cross-image feature points to search and match, calculate the specific characteristics or processing location of the location (2) object specific size measurement: the use of image processing to remove the edge of a specific object, with the use of the specified location, with cross-image feature points to search and match, calculate the specific characteristics or processing location; (3) elastic reference positioning: by the user to specify the required object on the processing of the reference position, with the characteristics of search matching, to achieve a number of times positioning of objects and the platform.
This study uses a high magnification (4.5 times maximum) of the visible light camera, with XYZ three-axis mobile platform, the use of calibration film to obtain the real size of the reference scale, shooting objects for a number of different sizes of block gauge. First, the camera focus, the different focal length of the calibration film image Fourier transform (FT) to the frequency domain, calculated by the spectrum of the best focal length. And then through the fixed magnification under the known fringe spacing, and image size conversion to obtain real size image scale. Then use the SURF (Speeded Up Robust Features) feature algorithm to search for the feature points in different images, and then match the images across the images to achieve cross-image positioning. Specific size measurement, the use of image processing methods to be aware of the contour geometry removed, and then by the feature point matching the information obtained by moving objects, can be obtained by matching the size. Based on the above steps, this study presents a system with a micron accuracy level and which has the ability to adapt to the real processing situation. The test results shows that 4 mm sample average measurement error is 1.6 μm and 10 mm sample average measurement error is 3.3 μm.

摘要　I
Abstract　III
致謝　V
目錄　X
圖目錄　XIII
表目錄　XXIII
第一章 緒論　1
1.1 前言　1
1.2 研究動機　1
第二章 文獻回顧　4
2.1 現有定位與量測方法　4
2.2 對焦演算法　17
2.3 輪廓偵測　19
2.4 特徵點探尋與匹配　27
2.5 影像模板匹配 29
第三章 研究方法 31
3.1 實驗裝置與軟體平台 33
3.2 研究測試樣本 38
3.3 影像比例尺 43
3.4 對焦演算法 46
3.5 物件輪廓獲取與紀錄 48
3.6 特徵點匹配與追蹤 53
3.7 定位或量測尺寸 58
第四章 結果與討論 61
4.1 影像對焦 61
4.2 物件輪廓提取 64
4.3 特徵搜尋與匹配 71
4.4 特徵點位移紀錄與計算 76
4.5 物件尺寸量測 80
4.6 量測誤差探討 83
4.7 模擬加工情況之物件量測 85
4.8 物件特徵搜尋能力與表面粗糙度 101
第五章 結論 106
5.1 結論 106
5.2 未來展望 109
參考文獻 111

[1]万敏. 彭静文, 李卫东, 张乐, "一种基于特征距离的模具定位方法 " China Patent, 2015.
[2]RENISHAW. Equator™ 工作原理. Available: http://www.renishaw.com.tw/tw/equator-explained--13465
[3]K. Okarma and M. Grudziński, "The 3D scanning system for the machine vision based positioning of workpieces on the CNC machine tools," in 2012 17th International Conference on Methods & Models in Automation & Robotics (MMAR), pp. 85-90, 2012.
[4]F. Li, A. P. Longstaff, S. Fletcher, and A. Myers, "Rapid and accurate reverse engineering of geometry based on a multi-sensor system," The International Journal of Advanced Manufacturing Technology, vol. 74, pp. 369-382, 2014.
[5]K. Lu and W. Wang, "A multi-sensor approach for rapid and precise digitization of free-form surface in reverse engineering," The International Journal of Advanced Manufacturing Technology, vol. 79, pp. 1983-1994, 2015.
[6]H. Srinivasan, O. L. A. Harrysson, and R. A. Wysk, "Automatic part localization in a CNC machine coordinate system by means of 3D scans," The International Journal of Advanced Manufacturing Technology, vol. 81, pp. 1127-1138, 2015.
[7]M. Corsini, P. Cignoni, and R. Scopigno, "Efficient and Flexible Sampling with Blue Noise Properties of Triangular Meshes," IEEE Transactions on Visualization and Computer Graphics, vol. 18, pp. 914-924, 2012.
[8]S. Y. Lee, Y. Kumar, J. M. Cho, S. W. Lee, and S. W. Kim, "Enhanced Autofocus Algorithm Using Robust Focus Measure and Fuzzy Reasoning," IEEE Transactions on Circuits and Systems for Video Technology, vol. 18, pp. 1237-1246, 2008.
[9]J. Canny, "A Computational Approach to Edge Detection," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI-8, pp. 679-698, 1986.
[10]Wikipedia. Gaussian blur. Available: https://en.wikipedia.org/wiki/Gaussian_blur
[11]I. Sobel. (2015). History and Definition of the so-called "Sobel Operator", more appropriately named the Sobel-Feldman Operator. Available: https://www.researchgate.net/publication/239398674_An_Isotropic_3_3_Image_Gradient_Operator
[12]Wikipedia. (2016). Sobel operator. Available: https://en.wikipedia.org/wiki/Sobel_operator
[13]A. A. Paul V. C. Hough, Mich., "NETHOD AND MEANS FOR RECOGNIZÍNG COMPLEX PATTERNS " the United States Patent, 1962.
[14]R. O. Duda and P. E. Hart, "Use of the Hough transformation to detect lines and curves in pictures," Commun. ACM, vol. 15, pp. 11-15, 1972.
[15]A. shack. (2016). The Hough Transform. Available: http://aishack.in/tutorials/hough-transform-basics/
[16]M. S. Chris Harris, "A combined corner and edge detector," presented at the Proceedings of Fourth Alvey Vision Conference, United Kingdom, 1988.
[17]S. Jianbo and C. Tomasi, "Good features to track," in 1994 Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, pp. 593-600, 1994.
[18]D. G. Lowe, "Object recognition from local scale-invariant features," in Proceedings of the Seventh IEEE International Conference on Computer Vision, vol. 2, pp. 1150-1157, 1999.
[19]H. Bay, A. Ess, T. Tuytelaars, and L. Van Gool, "Speeded-Up Robust Features (SURF)," Computer Vision and Image Understanding, vol. 110, pp. 346-359, 2008.
[20]N. Dalal and B. Triggs, "Histograms of oriented gradients for human detection," in 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05), vol. 1, pp. 886-893, 2005.
[21]C. Cortes and V. Vapnik, "Support-vector networks," Machine Learning, vol. 20, pp. 273-297, 1995.