A Database Watermark Embedding Method Based on PSO–GA Hybrid Optimization and Dynamic Difference Expansion

To overcome the limitations of excessive embedding distortion, susceptibility to local optima, and compromised synchronization in database watermarking, we propose PSOGA-DDE, a reversible watermarking method. We integrate a parallel PSO–GA hybrid model with a stratified embedding strategy to ensure global optimality and uniform watermark distribution. By employing dual-threshold-controlled dynamic difference expansion, the embedding strength is adaptively adjusted to maintain strict reversibility. Experimental results on UCI datasets demonstrate that our method achieves a lower Mean Absolute Error (MAE) compared to the representative FASADEW while maintaining comparable robustness. Against tuple and attribute deletion attacks, the proposed method exhibits enhanced stability and lower error rates over the classic DEW baseline, providing a flexible and effective solution for secure data management.