Abstract
High definition video application has drawn a lot of interest both from academy and industry. The relevant latest video coding technology, H.265/HEVC has been a promising area for video steganography. In this paper, we present a novel and efficient video steganography method based on transform block decision for H.265. In order to improve the visual quality of carrier video, we analyze the embedding error of data hiding with modifying partitioning parameters of CB, PB and TB, and modify the transform block decision to embed secret message and update corresponding residuals synchronously. In order to limit embedding error, we utilize an efficient embedding mapping rule which can embed N (N>1) bits message and at most modify one bit transform partitioning flag. Our experimental results show that the proposed method can achieve better visual quality, larger embedding capacity and less bit-rate increase than state-of-the-art researches.
What Problem Does This Paper Address?
Steganography—the art of hiding information within a "carrier" medium—is increasingly difficult in modern video codecs like H.265/HEVC. Because these codecs are designed for maximum compression efficiency, any modification to the bitstream to hide data often results in noticeable visual artifacts or a significant spike in the file size (bit-rate).
Previous methods focused on modifying motion vectors or transform coefficients. However, these often lead to "distortion drift," where a small error in one frame propagates through subsequent frames. Other methods attempted to modify Coding Block (CB) or Prediction Block (PB) structures. However, because these are high-level decisions, changing them forces the encoder to recalculate almost all other parameters, leading to poor performance and visual degradation.
Methods
The authors propose a method centered on the Transform Block (TB) decision. The TB is the smallest unit in the HEVC partitioning hierarchy where the actual residual error (the difference between the predicted and actual frame) is transformed and quantized. By targeting the TB rather than the CB or PB, the authors found a "sweet spot" where modifications are localized, preventing cascading distortion errors.
Key components of their approach include:
- Adaptive TB Partitioning: The algorithm modifies the binary flags that decide whether an 8x8 block is split into smaller 4x4 blocks. Since TBs are at the end of the prediction chain, modifying them has a minimal impact on other coding parameters.
- Hybrid Embedding Scheme: The system switches between two modes based on a threshold ($\Gamma$):
- Single-embedding: Hides one bit per candidate block.
- Collaborative-embedding: Uses matrix encoding to hide multiple bits ($N > 1$) by modifying at most one partitioning flag. This significantly boosts capacity while minimizing distortion.
- Residual Synchronization: When a partitioning decision is changed to hide data, the algorithm re-calculates the transform and quantization for that block. This "syncs" the video data with the new structure, making the hidden data nearly invisible.
- Fast Traversal Algorithm: To maintain encoding speed, they implemented a quadtree traversal that skips non-optimal partitioning paths, ensuring the steganography process doesn't add excessive computational lag.
Experimental Evaluation
The researchers evaluated their method using standard video sequences covering a range of resolutions, including BasketballPass (416 × 240), BlowingBubbles (416 × 240), KristenAndSara (1280 × 720), and ParkScene (1920 × 1080). These were tested across different Group of Pictures (GOP) structures, including All-Intra, IPPP, and IBBB.
Key Results
The proposed method was compared against recent state-of-the-art (SOTA) research, demonstrating superior performance across three critical metrics:
- Visual Quality: The average PSNR (Peak Signal-to-Noise Ratio) discrepancy was only 0.82 dB, meaning the stego-video is almost indistinguishable from the original compressed video.
- Capacity: The method achieved an average embedding capacity of 38.55 kbits/s, outperforming previous methods that struggled to maintain high capacity without compromising video quality.
- Bit-rate Control: By re-optimizing residuals after embedding, this method keeps the bit-rate increase remarkably low, averaging just 0.98%, whereas competing methods often saw increases of 1.5% to 2% or more.
Conclusion and Future Work
This research demonstrates that targeting the Transform Block decision in H.265/HEVC provides a highly efficient and visually "stealthy" channel for steganography. The use of matrix encoding allows for high data capacity with minimal structural noise.
However, the authors acknowledge that this is currently a fragile steganography method. The hidden data is likely to be lost if the video is subjected to "active attacks" such as re-encoding, scaling, or frame rate conversion. Future work will focus on improving the robustness of the method to ensure the secret message can survive common video processing operations.
Summary updated