Hybrid Steganography Using APVD and DCT for Dual-Domain Security

This study proposes and evaluates a novel hybrid steganographic method that integrates Adaptive Pixel Value Differencing (APVD) from the spatial domain and Discrete Cosine Transform (DCT) from the frequency domain to enhance security, imperceptibility and robustness in digital image steganography. By strategically leveraging the distinct advantages of both domains, our dual-domain approach first embeds a portion of the secret data via APVD, which excels at maintaining visual quality in smooth regions, and subsequently embeds the remaining data into carefully selected mid-frequency DCT coefficients, which offer superior resilience to compression and filtering attacks. Experimental evaluation on standard grayscale image datasets demonstrates that the proposed hybrid method significantly outperforms standalone APVD and DCT techniques in terms of Peak Signal-to-Noise Ratio (PSNR), payload capacity, and robustness against common image processing attacks like JPEG compression and noise addition. These results establish that a synergistic combination of spatial and frequency domain techniques can overcome the inherent limitations of single-domain approaches, offering a more secure and resilient framework for covert communication.