Fig. 1. Schematic of 3D integral imaging multilayer steganography

Fig. 2. Schematic of 3D ROI extraction technique based on spatial information segmentation

Fig. 3. Illustration of 3D ROI histogram-based watermark embedding

Fig. 4. Naked-eye 3D host image and watermarks for experiment

Fig. 5. Naked-eye 3D display principle and device

Fig. 6. 3D ROI extracting results based on different methods. (a) Cam-Net and (b) BasNet^[19] based on spatial texture segmentation; (c) Cam-Net based on sub-aperture image mapping^[15] and (d) hard-thresholding (The red box compares the ROI extracted by the four methods at a viewing angle of –14° along the y-axis, the green box compares the extracted salient texture)

Fig. 7. Watermarked 2D elemental image array and 3D naked-eye display: (a) "Rocket launcher" and (b) "Lung”

Fig. 8. Quantitative imperceptibility for six different 3D medical objects

Fig. 9. Light field display under six types of attacks and extracted text watermark

Fig. 10. Extracted multi-logo-multi-depth light field watermark under six attacks

Fig. 11. Details of recovered light field watermark

Fig. 12. Extracted single-logo-single-depth watermark under six attacks

Fig. 13. Extracted multi-logo multi-depth light field watermark embedded by single-layer SVD algorithm ^[20] under six attacks

Fig. 14. Extracted multi-logo-multi-depth watermark embedded by single-layer SVD algorithm ^[20] under six attacks