塑膠瓶蓋(Plastic Cap)的應用普遍存在於多種日常食用飲品中,而大部分的飲品都沒有做瓶口的封口設計,因此瓶蓋的密封圈為飲品品質的重要關卡,密封圈有缺陷,將可能會容易導致瓶裝內液體細菌感染,造成食物中毒;現今社會講究包裝整潔,塑膠瓶蓋的表面印刷也間接影響了顧客購買欲望,因此表面印刷的優劣影響了包裝的品質。 本研究之目的在運用機器視覺對塑膠瓶蓋的密封面及印刷面之瑕疵做檢測與分析,主要以機器視覺配合適當之光源取得瓶蓋之單面影像,再藉由影像處理之技術強化瑕疵之型態,將瑕疵擷取並做分析取得其特性,對於密封面與印刷面分別建立自我學習系統,簡化瑕疵檢測的參數設定。目前實驗結果顯示,已經突破密封面與印刷面的自我學習系統設計,只需要少部分的參數設定,即可進行瑕疵檢測。完成取像及判斷良莠的設計後,硬體上配合I/O訊號將瑕疵品從輸送帶上剔除,並搭配自動化控制元件,建立出一套瓶蓋自動化檢測系統。
Plastic caps are the most commonly seen bottle caps used in beverage and food containers. They are widely used to seal freshness of beverage or liquids in bottles. Threads are usually grooved inside the caps for easy twist-off caps and sealing rings prevent the liquids from bacterial infection. Companies print logos or pictures on the top surface of plastic cap, such that the quality of printing also indirectly affects the customers purchase. Inspection of plastic caps, including the surface printing, thread, and sealing ring, is a great issue during the caps production currently. The objective of this study is to use machine vision to inspect the defect of the sealing area and the printing surface of a plastic cap. An automated inspection system, which includes two CCD camera, lighting source, sensors, and a cap transporter, is constructed, and a digital image processing software is designed to learn good caps and screen out the defective ones. The experimental results show that the proposed inspection system can self-learn the features of a good surface printing, and effectively detect the defective caps under very few parameter setting, while the major defects in the sealing ring and thread area such as malformation, contamination, overfill, incomplete, scratches, can be successfully identified under the rate of 1200 piece per minute.