.png){kind=icon}
.png){kind=icon}
Return
A HEVC video steganalysis algorithm for transform unit partition modes
.png){kind=icon}
PRE
AI
DOI：10.1016/j.eswa.2025.129312.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• Recently, a new kind of video steganography based on transform unit (TU) partition modes are proposed. Existing video steganalysis methods struggle to effectively detect and analyze this type of steganography algorithm. To address this issue, a novel steganalytic approach is proposed in this paper. • By analyzing the difference between clean video and stego video, we find that TU-based steganography exerts a discernible influence on the partition structure of CTU. So, all of the CUs are categorized into eight different types based on their different depths, PU and TU partition modes. The area ratio occupied by different types of CUs are used to detect TU-based steganography. • Due to the uneven proportion of these eight types of CUs, some specific types of CUs exhibit a susceptibility to the influence of the TU-based steganography algorithm. So, the proportions of these CU types are also utilized for extracting the feature sets. • The proposed feature sets achieve great detection accuracy against TU-based steganography. Especially, the proposed feature sets have great steganalytic performance when quantization step (QP) is small.
Keywords:
video steganography
transform unit partition
steganalysis
coding tree unit
feature extraction
Journal
IF:
7.5
Papers:
2.9W
Citations:
10.2W
Organization
No organization information available