A novel algorithm for defect inspection of touch panels

• We develop a novel algorithm to inspect the defects on touch panel.
• Based on periodic patterns of touch panel, an adaptive model is learned online to extract defects.
• The experimental results indicate that our proposed method achieves accurate detection with efficient computation.
• The users pay very little effort for the testing of different panel products.

Automatic optical inspection plays an important role to control the appearance quality of wide range of products in the product process. Recently, the high popularity of smartphones and information appliances drives significant demand of touch panels. However, the traditional frequency-based method which exploits the line structure feature of texture images is not effective for the defect detection of touch panels. The paper presents a novel spatial domain algorithm to inspect the defects on touch panel. By utilizing the characteristics of periodic patterns of the sensing circuits, an adaptive model for each pattern is learned online to effectively extract defects. The experimental results indicate that our proposed method achieves accurate detection with efficient computation. In addition, the users pay very little effort for the testing of different panel products.