Title:
REVERSIBLE NEURAL NETWORK-BASED LARGE-CAPACITY IMAGE STEGANOGRAPHY AND RECOVERY METHOD, AND SYSTEM
Document Type and Number:
WIPO Patent Application WO/2022/166073
Kind Code:
A1
Abstract:
A reversible neural network-based large-capacity image steganography and recovery method, and a system. The purpose of the method is to embed one or more hidden images into a single cover image, and to recover all of the hidden images from a stego image. In the method, an image steganography model that supports bi-directional mapping is designed. The model is formed by cascaded reversible modules containing a cover branch and a hidden branch, and forward mapping is to embed the hidden images into the cover image to synthesize same into a stego image, and reverse mapping is to separate a cover image and hidden images out of a single stego image and recover same. According to the method, the reversibility of the model is fully utilized, and forward steganography process and reverse recovery process share all parameters, so that high-quality stego images and recovered images can be obtained concurrently, and the capacity of steganography is effectively improved.
Inventors:
LU SHAOPING (CN)
WANG RONG (CN)
ZHONG TAO (CN)
WANG RONG (CN)
ZHONG TAO (CN)
Application Number:
PCT/CN2021/100758
Publication Date:
August 11, 2022
Filing Date:
June 18, 2021
Export Citation:
Assignee:
UNIV NANKAI (CN)
International Classes:
G06T1/00
Foreign References:
| CN112884630A | 2021-06-01 | |||
| CN107087086A | 2017-08-22 | |||
| CN109818739A | 2019-05-28 | |||
| US20080199093A1 | 2008-08-21 |
Attorney, Agent or Firm:
BEIJING GAOWO LAW FIRM (CN)
Download PDF: