FaceCode: An Artistic Face Image with Invisible Hyperlink

With the widespread use of mobile social software, QR code plays an important role in acquiring messages from offline to online. However, the plain black and white blocks of QR code cannot attract users in social software, thus some popular software like Wechat beautify the QR code to improve its visual appeal. Inspired by the mentioned fact, this paper proposes a novel 2D image code named FaceCode. FaceCode is a Convolution Neural Network (CNN) based framework to embed information into a picture and beautify it simultaneously. For an input picture, FaceCode firstly uses a style transfer neural network to make the picture artistic, and then an embedding network cascading to the style transfer network embeds the personal information of users into the picture. The information is invisible but can be recognized by a phone camera. With the help of FaceCode, the users can hide their personal information in their profile pictures in social software. Experiments show that FaceCode can generate personalized profile pictures while maintaining the accuracy of recognizing the hidden information.