本論文主要研究一套透過影像處理技術及光學字元辨識(Optical Character Recognition,OCR)技術進行擷取自動化設備的生產參數,目的為了解決傳統生產設備無法輸出供外界系統使用的數位資料。本研究使用影像處理技術,經過色彩空間轉換、影像幾何轉換、影像二值化及型態學處理等;經過處理的影像分割字型樣式,並透過 OCR 技術進行字元辨識。為了提升系統字元辨識率,本研究探討不同的環境光源下,增加字元訓練的樣本數,可有效提升 OCR 技術的準確率。總結而言,本系統可以應用於傳統生產設備之資訊擷取,提高生產自動化效能,同時降低人力收集資訊所耗費 的時間與成本。
In this thesis, image processing techniques and the Optical Character Recognition (OCR) were used for automation equipment to capture production parameters, with the objective to solve the problem that conventional production equipment could not directly output digital data as inputs to exterior systems. In this study, image processing techniques included color space conversion, geometric transformation, binary image processing, and morphological image processing, etc. Then, character patterns were segmented in the processed image,followed by the character recognition using the OCR technology. Furthermore, this study investigated the effect of different environment illumination and increased the number of training character samples, which could effectively increase the accuracy of the OCR technology. In conclusion, our system could be used for conventional production equipment, leading to improve production automation performance and reduce time and cost to manually collect information.