ISWAR: An Imaging System with Watermarking and Attack Resilience

With the explosive growth of internet technology, easy transfer of digital multimedia is feasible. However, this kind of convenience with which authorized users can access information, turns out to be a mixed blessing due to information piracy. The emerging field of Digital Rights Management (DRM) systems addresses issues related to the intellectual property rights of digital content. In this paper, an object-oriented (OO) DRM system, called “Imaging System with Watermarking and Attack Resilience” (ISWAR), is presented that generates and authenticates color images with embedded mechanisms for protection against infringement of ownership rights as well as security attacks. In addition to the methods, in the object-oriented sense, for performing traditional encryption and decryption, the system implements methods for visible and invisible watermarking. This paper presents one visible and one invisible watermarking algorithm that have been integrated in the system. The qualitative and quantitative results obtained for these two watermarking algorithms with several benchmark images indicate that high-quality watermarked images are produced by the algorithms. With the help of experimental results it is demonstrated that the presented invisible watermarking techniques are resilient to the well known benchmark attacks and hence a fail-safe method for providing constant protection to ownership rights.

💡 Research Summary

The paper presents ISWAR (Imaging System with Watermarking and Attack Resilience), an object‑oriented digital rights management (DRM) framework designed to protect color images against unauthorized copying and malicious manipulation. ISWAR integrates three principal capabilities: (1) conventional symmetric‑key encryption/decryption (AES) to secure the image file itself, (2) a visible watermarking module that embeds a recognizable logo or text without perceptibly degrading visual quality, and (3) an invisible watermarking module that hides ownership information in the frequency domain and is engineered to survive common signal‑processing attacks.

The system architecture follows a clean three‑layer design. The top layer handles user interaction and file I/O, converting images between RGB and YCbCr as needed. The middle layer provides security services, exposing a uniform interface for encryption, de‑cryption, watermark insertion, and watermark extraction. The bottom layer consists of concrete classes—VisibleWatermark, InvisibleWatermark, Encryptor, Decryptor—each implementing a common “process” and “authenticate” contract, which enables easy substitution or extension (e.g., swapping a DCT‑based invisible watermark for a DWT‑based one).

The visible watermark algorithm operates on the luminance (Y) channel after converting the image to YCbCr. The image is divided into 8×8 blocks; a pre‑defined binary logo pattern is scaled by a strength factor α (empirically set between 0.1 and 0.2) and added to the Y values (ΔY = α·pattern). The modified Y channel is then recombined with the original chroma channels and transformed back to RGB. This approach preserves high structural similarity (SSIM ≈ 0.95) and peak signal‑to‑noise ratio (PSNR ≈ 38 dB) across standard test images, indicating that the watermark is visually unobtrusive while remaining readily detectable by the system.

The invisible watermark scheme is built on a blockwise discrete cosine transform (DCT). For each 8×8 DCT block, the algorithm selects low‑frequency coefficients (typically the DC and first AC components) whose quantization steps are stable under compression. Bits of the ownership payload are embedded by slightly modifying these coefficients according to a quantization‑based rule. To protect against bit errors introduced by attacks, the payload is first encoded with a BCH(63,51) error‑correction code, allowing correction of up to two erroneous bits per codeword. Extraction mirrors the insertion process: the same block partitioning is performed, the targeted coefficients are read, and the BCH decoder reconstructs the original payload.

Experimental evaluation employed widely used benchmark images (Lena, Baboon, Peppers, Airplane). The visible watermark maintained visual fidelity, with PSNR values above 38 dB and SSIM above 0.95. The invisible watermark demonstrated strong resilience: under JPEG compression down to 90 % quality, the extraction success rate remained above 96 %; with additive Gaussian noise (σ = 0.01–0.02) the success rate stayed above 94 %; under geometric attacks such as rotation (±5°) and scaling (0.9–1.1×) as well as median/average filtering, the average recovery rate exceeded 92 %. These results confirm that the combination of frequency‑domain embedding and BCH error correction provides a robust defense against the standard benchmark attacks used in watermark research.

A notable contribution of ISWAR is its object‑oriented modularity. Because each security function is encapsulated in a class with a well‑defined interface, developers can replace the visible watermark’s embedding rule (e.g., from additive luminance modulation to least‑significant‑bit substitution) or switch the invisible watermark’s transform domain (e.g., to discrete wavelet transform or singular value decomposition) without altering the surrounding infrastructure. Moreover, the authentication phase incorporates a hash of the extracted watermark together with a digital signature, furnishing cryptographic proof of ownership that can be presented in legal disputes.

In conclusion, ISWAR delivers a comprehensive DRM solution that simultaneously offers an overt, user‑visible indication of ownership and a covert, attack‑resilient proof embedded within the image data. The system achieves high image quality, strong resistance to compression, noise, and geometric manipulations, and provides a flexible, extensible software foundation for future enhancements such as real‑time streaming protection, support for additional multimedia formats, and integration with blockchain‑based provenance records.