摘 要本研究計劃旨在全自動微鑽針影像量測系統的建立，在相關的微鑽針影像量測系統研究，大多只是著墨在特徵的擷取及計算瑕疵上，少有人針對人為調整所造成的誤判率研討，因此本研究針對需要人為調整的光源調整、二值化設定及規格調校，此三大部份進行自動調校，避免人為因素誤差的產生。微鑽針檢測時，旋轉角度並非是固定的，而在此特性下，一般量測程式常使用相似度比對， 利用相似度比對抓取中心點座標及平行軸線的十字交角，而相似度比對會在大量檢測生產時，造成百分點的比對誤差，因此, 本文研究可取代相似度比對的演算法，減少百分點的比對誤差率。AbstractThis research project aimed at the establishment of automatic micro-drills image measurement system. In the related research about the micro-drills image of measurement system, we put most emphasis on the feature extraction and the calculation of the defects. Rare do people deliberate upon the false positive rate seminars caused by men. Therefore, this study focuses on the adjustments of light sources, the binary image settings and the regulations for specifications which require men adjustments. In order to prevent the bias caused by human, we decided to progress and revise these three parts automatically.When testing the micro-drills, the rotating angle isn’t fixed. Under this character, regular measurement systems often use it to match the similarities. Using the similitude to obtain the center of coordinates and the cross angle that parallels to the axis. Whereas, the matching up will result in errors of percentage points while checking on the mass producing. Therefore, this research can replace the algorithm of pattern matching and reduce the error rate of percentage points.
