Image Based Encryption System

The invention refers to a method for generating 2D barcodes, or generic tags, containing information encrypted and not decipherable except by those who generated them, which have a good visual impact integrated into the context, as opposed to current barcodes.

Title of the patent Encryption method and system based on images
Ownership Alma Mater Studiorum – University of Bologna
Inventors Daniele De Gregorio, Luigi Di Stefano
Protection International
Licensing status Available for licensing agreements
Keywords Convolutional neural network, Deep networks, Barcodes, Augmented reality, Watermarking
Filed on Feb 12, 2019

The proposed system uses Deep Convolutional Generative Adversarial Networks (DCGAN) to generate images, in black/white or color (with a style similar to an input image or randomly), that are apparently meaningless, but if read with a special device they reveal a hidden code.

The method implements a public key encryption system in which the part capable of generating images is considered as a Public Key and the part capable of interpreting images is considered as a Private Key.


In industry,  manufacturing and tertiary sectors, for:

  • replacement of QR Codes;
  • augmented reality;
  • robotic manipulation;
  • watermarking with product logos.



  • simplicity of implementation  as hardware;
  • readability of tags even in non-ideal conditions;
  • robustness of the decoder, trained with distortions;
  • lower aesthetic impact on products and their packaging;
  • unique and non-repeatable bar codes.