自動化的瑕疵檢測應用於機械零件的圖形識別

Automated Flaws Inspection and Its Application in Pattern Recognition of Mechanical Parts

張勻碩(Yun-Shuo Chang)

指導教授 ： 陳文雄

國立暨南國際大學/科技學院/電機工程學系/碩士(2013年)

摘要

近年來機械自動化工程隨著科技的進步，產品製造的功能性也必須跟著改變。許多傳統產業上所需要的零件越來越精簡化，所以零件加工就成為了一個重要的環節，要如何將機械製造所生產出來的零件精準度提高便是現在傳統中小企業工廠面臨的一大問題。
現在台灣傳統產業製造以電腦數值控制（Computer Numerical Control；CNC）工具機為多數，在機械製造過程中往往對於品質的控管是相當不易的，因為利用人為來檢測零件的時間長、效率低、漏檢率高、視覺圖像訓練不易、人員成本也高，所以「品質檢測」是每家工廠都會面臨到的一個重要課題。因此本論文提出一個自動化的瑕疵檢測技術並結合電腦圖學與機械視覺來做一個全新概念的機械零件的圖形識別檢測系統。
本文採用OpenCV的電腦視覺程式庫來做出機械圖形比對的功能，並且利用相機或攝影鏡頭，結合C++與MYSQL資料庫設計出一套自動化的瑕疵檢測系統。另外一個重點是使用這套系統能夠更方便且快速地判斷瑕疵物件，並不會花費太多時間在品質檢測階段，這將節省傳統產業的人力資源。

關鍵字

瑕疵檢測 ； 機械視覺 ； 電腦圖學

並列摘要

Recently, with technological advance in mechanical automation engineering in the field of manufacturing have to change. Many mechanical parts are required to be more and more precise. To improve the precision becomes an important issue.
Nowadays, most machines in traditional industries are computer numerical control (CNC) based. The quality control of product process is always not easy. Because manual inspection is time-consuming, inefficient, with high undetected rate. The image recognition training is difficult and personnel costs are high. So, the quality control becomes an important issue. This paper proposes to apply pattern recognition and mechanical vision to the automated flaw detection system.
We use OpenCV for pattern recognition. It utilizes camera, C++ programming, and MySQL database to devise an automated flaw detection system. This system can detect flaw objects very quickly and save a lot of time in quality control. It will help and improve human resource of traditional companies.

並列關鍵字

Flaw Detection ； Mechanical Vision ； Pattern Recognition

參考文獻

[1] “Computer Numerical Control,” http：//en.wikipedia.org/wiki/Numerical_control.

[2] “DSI CNC三軸向重複定位精度檢測,” http：//www.youtube.com/watch?v= WCjGTIOaML0.

[3] A. O. Fernandes, L. F. E. Moreira, and J. M. Mata, “Machine Vision Applications and Development Aspects,” IEEE Int'l Conf. on Control and Automation, pp. 1274–1278, 2011.

[4] M. Alasdir, J. H. Wang, and C. S. Tseng, “Introduction to Digital Image Processing with Matlab,” Baker & Taylor Books, April, 2004.

[5] N. Alvertos, D. Brzakovic, and C. Gonzalez Rafael, “Camera Geometries for Image Matching in 3-D Machine Vision,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol.11, pp. 897–915,1989.