白色透過光を用いたランプチューブ欠陥自動検出装置の開発

抄録

In this study, the technology for automatically detecting cracks, impurities, cavities, and surface scratches in a sodium lamp tube during production is developed. The specimen is illuminated from below with white transmitted light and the top of the specimen is viewed from above with a CCD linear image sensor. The viewed image is transferred into computer memory and smoothed with a 3 × 3 pixel smoothing operator. The above-mentioned four types of defects are detected using a subtracted image which is obtained by subtracting the original image from the smoothed image. The detected defects are classified into three groups based on their longitudinal length and the rotational length. Most impurities and cavities fall into the 1st group, and most cracks and scratches into the 3rd group. However, some impurities, cavities, cracks and scratches fall into the 2nd group where they cannot be classified with this machine. The defects in this group are subsequently classified into the 1st group or the 3rd group based on observation with the naked eyes. As a result, the 1st group contains impurities and cavities, and the 3rd group contains cracks and scratches. It is verified experimentally that this method is effective for the detection and the classification of the defects into two groups.