.png){kind=icon}
.png){kind=icon}
Return
Virtualized three-dimensional reference tables for efficient data embedding
.png){kind=icon}
PRE
AI
DOI：10.1016/j.jvcir.2024.104351.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Data embedding methods utilizing a three-dimensional reference table (3DRT) modify pixels to embed digits from various bases using the 3DRT. However, the current 3DRT-based methods are constrained to specific bases and necessitate a physical 3DRT for both embedding and extraction processes. This paper introduces a novel approach that constructs the 3DRT using groups of anisotropic cubes to minimize embedding distortion. The 3DRT is virtualized by representing it as a two-coefficient equation, eliminating the need for a physical 3DRT during embedding and extraction. This virtualization significantly reduces computational complexity, enabling embedding and extraction through straightforward calculations. Furthermore, virtualization decreases the storage space required for the 3DRT. Experimental results demonstrate that the proposed method achieves high image quality and embedding capacity. Specifically, at embedding rate of 2 and 3 bits per pixel, the method produces quality scores of 46.99 dB and 40.91 dB, respectively, across 200 test images.
Keywords:
Three-dimensional reference table
Data embedding
Virtualization
Journal
IF:
3.1
Papers:
414
Citations:
5.6K