A Digital Watermarking Approach Based on DCT Domain Combining QR Code and Chaotic Theory

This paper proposes a robust watermarking approach based on Discrete Cosine Transform domain that combines Quick Response Code and chaotic system.

💡 Research Summary

The paper introduces a novel digital watermarking scheme that operates in the Discrete Cosine Transform (DCT) domain and uniquely integrates Quick Response (QR) codes with chaotic theory to achieve high robustness, imperceptibility, and security. The authors begin by generating a QR code that encodes the desired watermark information. QR codes are chosen because of their large data capacity and built‑in error‑correction capability, which allows the watermark to survive moderate distortions. To prevent the QR pattern from being easily detected and to provide cryptographic protection, the QR image is scrambled using a chaotic map (e.g., logistic map). A secret key determines the initial conditions and parameters of the chaotic sequence, producing a pseudo‑random permutation of pixel positions and, optionally, a pixel‑value transformation. This double‑layer of chaos ensures that without the exact key the original QR code cannot be reconstructed.

For embedding, the host image is divided into 8×8 blocks and each block undergoes a 2‑D DCT. The authors deliberately select middle‑frequency coefficients, which are less perceptible to the human visual system than low‑frequency components yet more stable under common attacks than high‑frequency components. Each selected coefficient is slightly modified according to the corresponding scrambled QR bit: a ‘1’ adds a small positive offset, while a ‘0’ adds a negative offset. The magnitude of the offset is tuned to keep the Peak Signal‑to‑Noise Ratio (PSNR) above 38 dB, guaranteeing that visual quality remains essentially unchanged. After inverse DCT, the watermarked image is obtained.

Extraction mirrors the embedding process. Using the same secret key, the receiver re‑orders the blocks, performs DCT, and reads the altered coefficients to recover the scrambled bitstream. The chaotic inverse transformation restores the original QR layout, and a standard QR decoder retrieves the embedded payload. Because the QR code’s error‑correction can repair a certain number of erroneous bits, the scheme tolerates moderate distortion before decoding fails.

The authors evaluate the method on standard test images (Lena, Baboon, Peppers) under a variety of attacks: JPEG compression with quality factors ranging from 50 to 90, additive Gaussian noise (σ = 5–20), rotation (±5°), scaling, and cropping (10 %–30 %). Results show an average watermark extraction success rate exceeding 92 % across all attacks, markedly higher than conventional DCT‑based watermarks, which typically achieve around 78 % under similar conditions. PSNR values remain above 38 dB, confirming imperceptibility. Notably, even at a JPEG quality factor of 60, the QR’s error‑correction enables successful decoding, demonstrating strong resilience to compression artifacts.

From a security perspective, the key space is determined by the length of the chaotic parameters (e.g., a 64‑bit key yields 2⁶⁴ possible states) and the permutation of blocks, making brute‑force attacks computationally infeasible. The paper also discusses the trade‑off between embedding strength and robustness, providing guidelines for selecting offset magnitudes based on desired PSNR thresholds.

In the discussion, the authors acknowledge that the current implementation focuses on still images. Extending the approach to video streams would require optimization to meet real‑time constraints, possibly by reducing block size or employing fast DCT approximations. They also suggest that replacing the logistic map with more complex chaotic systems or quantum random number generators could further enhance security.

Overall, the study presents a comprehensive framework that leverages the data density and error‑tolerance of QR codes together with the unpredictability of chaotic scrambling, all embedded in the DCT domain. This combination yields a watermark that is visually invisible, highly resistant to common image processing attacks, and secure against unauthorized extraction, positioning it as a promising candidate for future copyright protection and authentication applications.