DEEP RESTORATION OF INVISIBLE QR CODE FROM TPVM DISPLAY

QR (Quick Response) codes are widely used in daily lives for their convenience in conveying information from offline to online. However, black-white QR code image takes up space and is not visually appealing when embedded into videos. Owing to the high resolution and high refresh rates of display screens, we are able to embed QR code to be invisible on displays based on temporal psychovisual modulation (TPVM). In this work, we proposed a method to improve the invisibility of QR code via TPVM. Furthermore, we propose a fast restoration-recognition method to scan invisible QR codes by smartphones using deep convolutional neural networks. Extensive experiment results show that our proposed methods effectively improve the indivisibility of QR code and can efficiently recognize the invisible QR code correctly and stably. Our methods can be applied in video copyright protection and Internet shopping without affecting the visual quality for audiences.