A scheme of hiding large-size image into small-size image based on FCdDNet.

Area of Science:

• Computer Science
• Image Processing
• Cryptography

Background:

• Current information hiding techniques offer substantial capacity, yet there's a demand for greater data embedding.
• Existing methods can embed two color images within a single color image carrier.

Purpose of the Study:

• To develop a novel information hiding scheme with enhanced capacity for embedding large color images into smaller ones.
• To improve upon existing methods by utilizing an improved Fully Convolutional Densely Connected Network (FCdDNet).

Main Methods:

• An improved FCdDNet serves as a shared backbone for both the hiding and extraction networks.
• The hiding and extraction networks are trained adversarially in pairs, promoting mutual enhancement.
• The proposed scheme allows for the hiding of large-size color images within small-size color images.

Main Results:

• The developed scheme achieves a significantly larger information hiding capacity, embedding data four times the scale of the carrier image.
• The visual quality of the steganographic image is preserved.
• Extracted images demonstrate a high degree of fidelity and restoration.

Conclusions:

• The proposed information hiding scheme based on improved FCdDNet offers superior capacity compared to existing methods.
• The technique ensures high visual quality and accurate data recovery.
• Potential applications include secure image transmission, data authentication, and covert communication.