隨著工業科技的發展以及高度自動化生產技術的進步，在檢測的技術和效率上也有相對的提昇。而近年來自動光學檢測更是掀起了一波熱潮，其技術就是運用機器視覺於自動化生產線上的檢視與量測。因為有不少的檢測方法是必須接觸待測物才能量測，這使得待測物中微小結構因而受到破壞，尺寸或形狀也因此受到變形，最終導致量測上的不精確或誤判；反觀機器視覺檢測具有非接觸性、非破壞性的特性，且可達到即時性、全面性的檢測能力，因此如何量測這類全域性元件的外型尺寸便是一個值得研究的課題。 本研究中我們以CCD及自動對焦鏡頭搭配三維運動機構建立一套全域性外形尺寸的量測系統。本研究將主要使用變焦CCD擷取待測物之影像，並應用霍氏轉換 (Hough Transfer) 演算法處理該影像，最後轉換出待測物的外型輪廓尺寸並與理論值做比對，進而評估全域性影像量測的準確性。 The new developments in industrial technology and advancement in automatic production technology, have resulted in the improvement of inspection both in method and efficiency. Besides, Automatic Optical Inspection has been a hot topic in recent years; it actually applies in Machine Vision to automatic production line on inspection and measurement. Many measurement methods require contact to the specimens, therefore, they not only destroy specimens’ tiny structure but also change their sizes and shapes. Sometimes, they even cause inaccuracy and misjudgment on measuring. In contrast, Machine Vision’s features are not only non-contact and non-destructive, but also enable real time and full coverage inspection ability. Due to these advantages, it’s a study-worthy topic on how to inspect this kind of global device shape size. In this research, we built a set of measurement system of global shape’s size by CCD and auto-focus lens, which go along with a 3-D motion device. Moreover, we applied various measurements on shapes and sizes, including Hough Transfer and Stereo image Theorem principle to evaluate the accuracy of certain component which is applied to global measurement. As the results, we can evaluate the accuracy of the system for basic profile measurement like triangle, square and circle of global components.
