Dual-key-based adaptive watermark embedding for light field 3D images.

Area of Science:

• Computer Vision
• Digital Image Processing
• Information Security

Background:

• Existing watermarking methods struggle to leverage viewpoint data in light field 3D images.
• This limitation leads to reduced invisibility and robustness of embedded watermarks.

Purpose of the Study:

• To propose a novel light field 3D dual-key-based watermarking network (3D-DKWN).
• To enhance both the invisibility and robustness of watermarks in light field 3D images by utilizing viewpoint information.

Main Methods:

• A pixel mapping algorithm is used to generate a disparity sub-image from the light field 3D image.
• An encoding key (EK) is generated from the disparity sub-image.
• Adaptive watermark embedding is performed on the disparity sub-image, generating a steganographic key (SK).
• The watermarked light field 3D image is reconstructed.

Main Results:

• The proposed 3D-DKWN effectively utilizes viewpoint information from light field 3D images.
• Significant improvements in watermark invisibility were observed compared to previous methods.
• Enhanced robustness of the embedded watermark was achieved.

Conclusions:

• The novel 3D-DKWN offers a superior approach for watermarking light field 3D images.
• Effective utilization of viewpoint information is crucial for robust and invisible watermarking.
• This method addresses key limitations in current 3D image watermarking techniques.