本研究旨在開發一套能應用於實際加工機中來協助裝載切削刀具輔助系統，在製程中刀具尺寸的設定，是由人員量測初步值輸入控制器後，再透過加工機中的刀具量測儀進行更精密的測量來得到實際刀具尺寸，而其中當初步的尺寸設定錯誤時，可能會造成刀具與量測儀之間發生碰撞。在現行做法裡刀具的安裝設定與確認的檢查仍依賴人工檢測以及經驗法則判斷，這些一連串的前置作業不但太過依賴人工，且長時間的工作存在著人為疏失的風險產生。
本次研究將藉由電腦視覺技術為基礎開發裝載切削刀具輔助系統，針對主軸上的刀具影像同時進行初步的刀長量測以及確認刀具類型。在初步的刀具量測中，藉由所擷取到的刀具影像進行影像分割並從中取得刀具輪廓進行特徵分析與尺寸計算。透過此系統所測量的初步刀長值可確保量測誤差在 0.5 公厘以內。此外， 本研究以 Faster R-CNN 類神經網路模型來建構出刀具類型分類器，在辨識常用 刀具時其辨識率可達 90%以上。最後經由測試結果得知，系統只需要耗費 2 秒鐘便可得到初步的刀具尺寸與刀具類型的確認，並且同時傳輸至控制器上，相對於傳統人工方式顯得更有效率。

The study aims to develop a computer vision system to aid the tool loading operation for the CNC machining center. CNC machining is a prevailing machine tool for either mass or small production. Although the process can be operated automatically, the presetting is often done manually, especially for customized or small production. These works include loading workpiece and cutting tools, keying in tool information into the controller, and setting work coordinate. However, the manual operation could result in dramastic damage during the machining process because of the human error. The mistakes such as loading wrong type of cutting tools and typing in wrong values of cutting tool size can result in dramastic loss like collision of tool setter or unqualified products, even damage the machine tool. The technology of computer vision is used in this study to develop a system for the tool loading. Two functions of tool length and tool type detections were developed. In the detection of tool length, the technology of edge detection was employed, while Faster R-CNN was utilized in the recognition of tool type. The detection is conducted while the cutting tool is loaded on the spindle. The error of the tool length is within 0.5 mm, which is enough for the tool setter to measure the precision tool setting length. And the accuracy of the tool type recognition can be upto 90% with trained cutting tools. The detection results are uploaded to the CNC controller directly. The detection of each cutting tool can be done around 2 seconds. This investigation is expected to avoid the human error when keying in the tool information into the controller, or it could be used for the double-checking in the automatic operations.

[1] Yunardi, R. T., and Pujiyanto, W., 2015, “Contour-based Object Detection in Automatic Sorting System for a Parcel Boxes,’’ Proceedings of the IEEE International Conference on Advanced Mechatronics, Intelligent Manufacture, and industrial Automation, ICAMIMIA 2015, Surabaya, Indonesia, October 15-17, 2015, pp.38-41.
[2] Golkar, E., and Prabuwono, A. S., 2013, “Vision Based Length Measuring System for Ceramic Tile Borders,” Proceedings of the 4th International Conference on Electrical Engineering and Informatics, ICEEI 2013, Phuket, Thailand, September 21-22, 2013, pp.771-777.
[3] Zhang, X., Tsang, W., Tian, X., Yamazaki, K., and Mori, M., 2008, “Automatic segmentation of the apparent contour for 3D modeling of cutting tools from single view,” Proceedings of the 4th International Symposium on Visual Computing, ISVC 2008, Las Vegas, Nevada, United states, December 1-3, 2008, pp.772-781.
[4] Zhang, X., Tsang, W. M., Mori, M., and Yamazaki, K., 2010, “Automatic 3D model reconstruction of cutting tools from a single camera,” journal of Computers in Industry, 61(7), pp. 711-726.
[5] Zhang, X., Shi, Y., Zhang, B., and Si, C., 2018, “A study of on-machine micro milling cutter condition inspection based on machine vision,’’ Journal of Micro and Nano-Manufacturing, 6(3), ASME Paper No. JMNM-17-1065, pp.1-7.
[6] Guo, S., Zhang, J., Jiang, X., Yin, P., and Lei, W., 2011, “Mini Milling Cutter Measurement Based on Machine Vision,” Proceedings of the International Conference on Advanced in Control Engineering and Information Science, CEIS 2011, Dali, China, August 18-19, 2011, pp. 1807 – 1811.
[7] Zoller, “Tool measuring technology,” accessed June 28, 2020, https://www.bermat.com.br/images/catalogos/aparelho-de-medicao-e-inspecao-de-ferramentas-pombasic.pdf.
[8] 洪朝陽，應用電腦視覺於微型鑽頭與鍍膜銑刀的量測，國立交通大學工業工程與管理學系，碩士論文，新竹，台灣，2004。
[9] 廖家鉦，端銑刀自動化量測之研究，中華大學機械工程學系，碩士論文，新竹，台灣，2011。
[10] 陳宇叡，具自動分類與磨耗識別之刀具量測系統，國立高雄科技大學電機工程研究所，碩士論文，高雄，台灣，2018。
[11] 兵耿豪，立體視覺應用於車刀尺寸量測，國立中正大學機械工程學系，碩士論文，嘉義，台灣，2013。
[12] Ping, G. H., Chien, J. H., and Chiang, P. J., 2018, “Verification of turning insert specifications through three-dimensional vision system,” International journal of Advanced Manufacturing Technology, 96(9), pp. 3391–3401.
[13] Chiang, P. J., and Ping, G. H., 2013, “Stereo Vision for the Measurement of Turning Tool Size,” Proceedings of the IEEE 6th International Workshop on Computational Intelligence and Applications, IWCIA 2013, Hiroshima, Japan, July 13, 2013, pp. 173-177.
[14] Yann, L., Ligon. B., Yoshua, B., and Patrick, H., 1998, “Gradient-based learning applied to document recognition,” Proceedings of the IEEE, 86(11) pp. 2278-2324.
[15] Girshick, R., Donahue, J., Darrell, T., and Malik, J., 2014, “Rich feature hierarchies for accurate object detection and semantic segmentation,” Proceedings of the IEEE 27th Conference on Computer Vision and Pattern Recognition, CVPR 2014, Columbus, Ohio, United States, June 23-28, 2014, pp. 580-587.
[16] Girshick, R., 2015, “Fast R-CNN,” Proceedings of the IEEE 15th International Conference on Computer Vision, ICCV 2015, Santiago, Chile, December. 7-13, 2015, pp. 1440-1448.
[17] Girshick, R., Ren, S., He, K., and Sun, J., 2017, “Faster R-CNN: towards real-time object detection with region proposal networks,” Journal of IEEE Transactions on Pattern Analysis and Machine Intelligence, 39(6), pp. 1137-1149.
[18] Redmon, J., Divvala, S., Girshick, R., and Farhadi, A., 2016, “You Only Look Once：Unified, Real-Time Object Detection,” Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2016, Las Vegas, Nevada, United States, June 26 – July 1, 2016, pp.779-788.
[19] Kaehler, A., and Bradski, G., 2008, Learning OpenCV 3：Computer Visionin C++With the OpenCV Library, O'Reilly, California, United States.
[20] Singla, N., 2014, “Motion Detection Based on Frame Difference Method,” International Journal of Information and Computation Technology, 4, pp. 1559-1565.
[21] Canny, J., 1986, “A computational approach to edge detection,” Journal of IEEE Transactions on Pattern Analysis and Machine Intelligence, 8(6), pp. 679-698.
[22] Masound, A. A., and Bayoumi, M. M., 1995, “Using local structure for the reliable removal of noise from the output of the LoG edge detector,” Journal of IEEE Transactions on System, Man, and Cybernetics, 25(2), pp. 328-337.
[23] IMAQ, 2000, “IMAQ Vision Concepts Manual,” National Instruments Corporate Headquarters Part Number 322916A-01.
[24] Duda, R. O., and Hart, P. E., 1972, “Use of the Hough Transformation to Detect Lines and Curves in Pictures,” Journal of Communication of the ACM, 15(1), pp. 11-15.
[25] Simonyan, K., and Zisserman, A., 2015 “Very deep convolutional networks for large-scale recognition,” Proceedings of the 3rd International Conference on Learning Representations, ICLR 2015, San Diego, California, United states, May 7-9, 2015, pp. 1884-2021.
[26] Otsu, N., 1979, “A threshold selection method from gray-level histograms,” Journal of IEEE Transactions on Systems, Man and Cybernetics, 9(1), pp. 62-69.
[27] Zack, G. W., Rogers, W. E., and Latt, S. A., 1977, “Automatic measurement of sister chromatid exchange frequency,” Journal of Histochemistry and Cytochemistry, 25(7), pp. 741-753.
[28] “TensorFlow,” accessed June 28, 2020, https://www.tensorflow.org.
[29] Lin, Tzuta., “LabelImg,” accessed June 28, 2020, https://github.com/tzutalin/labelImg.