Extracting Paper Sticky Notes with Visual-inertial Odometry of ARKit
伊藤栄俊, 大囿忠親, 新谷虎松


Post-it notes are popular on meetings, such as brainstorming or card sorting. On brainstorming, we arrange paper cards for grouping. The positional relationship of paper notes is important. However, there is a problem between an image resolution and positional relationship of paper notes. Extraction of the paper notes content requires high resolution image data. As a capture point is close to paper notes, positional relationship of paper notes is less clear. More and more paper notes make that problem tangible. We present a method to solve the trade-off problem about the positional relationship on extracting paper sticky notes. We developed three extract mode: Single shot, Semi auto, and Full auto. Our method extends the area where we use for brainstorming. The result of experimental evaluation shows that our system can project virtual cards at the same position as the original accurately.