臺灣博碩士論文加值系統

　　目前機器視覺廣泛應用在各項領域，本文提出以機器視覺方法來執行物體尺寸精密量測的模式，以配合生產線上進行檢測的工作。文中以數位影像處理技術為基礎來結合量測之外，並在影像處理過程中提出如何對影像做均勻增強，使得二值化及邊緣偵測等後級處理部份能快速的執行和得到較佳的效果；亦對二值化影像進行細部瑕疵的編修，減少瑕疵點對影像的破壞，使其更符合原始影像的需要。另再針對量測較小型的工件進行探討，實驗中在量測系統上設計了較高放大倍率的影像擷取系統，以應較高的解析度需求，並另外設計了多處同步擷取影像的機構，克服須逐次移動量測的問題，以節省量測時間，符合線上檢測需求。
　　實驗應用時以電子槍極片間隙、光纖外徑和游標尺刻度的量測為例設計量測系統，系統解析度分別達到了9μm和6μm以內(CCD像素解析為510×492pixel)，實驗結果顯示其在重複性測試上也極為穩定。

　　Machine vision is currently applied in many fields. In this study, an on-line detection technique base on machine vision for precision dimension measurement is proposed. The measurements are based on the digital image processing technique and include a gray-level reset method to enhance image. After the image enhanced, the process of binarize and edge detection are more easy and the noise of image is little. For the needs of high image resolution and fast measurement speed, a mechanism that can measure different parts of the specimen synchronously with a larger resolution was designed. Clearance of electron gun grid, diameter of light fiber and scale of vernier are tested by the measurement system. Experimental results show that a system resolution within 0.009mm／pixel can be achieved and the repeatability is very stable.