Optimizing Steganographic Fidelity: Content-Aware Syndrome Trellis Code

Syndrome Trellis Code (STC) stands out as one of the nearly optimal steganographic coding methods to date. Its superior efficiency and performance have garnered significant attention. However, STC has a limitation: it struggles to handle unevenly distributed host signals and messages, preventing it from achieving the theoretical minimum distortion. To address this flaw, we introduce an enhanced STC scheme called Content-Aware STC (CA-STC). In our work, we propose modifying the codebook based on the statistical relationship between host signals and messages. This adjustment aims to reduce overall distortion. Additionally, we introduce a new parameter to strike a balance between complexity and embedding efficiency in the proposed method. Simulation results, including scenarios involving random data, demonstrate that our approach outperforms STC in terms of global distortion, effectively adapting to diverse scenarios. Code available: https://github.com/shx-lyu/CA-STC/.