International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 12 | Dec 2025
p-ISSN: 2395-0072
www.irjet.net
A Study on the Adequacy of Digital Watermarking in Establishing Ownership Rights Pavan A C1, M T Somashekara2 1Assistant Professor, Department of Computer Science and Engineering, PES University, Bengaluru, India 2Associate Professor, Department of Computer Science and Applications, Bangalore University, Bengaluru, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Over the years, researchers have developed
This class of attacks highlights a significant vulnerability that arises when the original image is used in the detection process. It prompts an important question: can we ensure ownership verification without referencing the original image during detection? In this paper, we argue that it is indeed possible. We present a set of necessary and sufficient conditions that a general watermarking scheme must satisfy to establish rightful ownership. Furthermore, we show that a modified version of the scheme proposed in [1] adheres to these conditions and performs effectively in this context.
numerous methods aimed at undermining or defeating digital watermarking systems. These developments indicate that the presence of a robust, non-removable watermark alone is not enough to establish ownership with certainty. This paper aims to identify and formalize the critical conditions a watermarking scheme must meet to reliably determine rightful ownership. Our evaluation shows that several prominent image watermarking approaches fall short of these requirements. However, we demonstrate that a modified version of a known watermarking technique [1] performs effectively under the proposed criteria.
2. NECESSARY AND SUFFICIENT REQUIRETo begin with, we identify the essential conditions that a watermarking scheme must fulfil to prevent subversion. The first and most straightforward requirement is that the scheme should be resistant to watermark removal—referred to hereafter as REQ1. Specifically, any attempt to remove the watermark should cause noticeable degradation in image quality. If this condition is not met, an attacker could simply erase the watermark and embed their own, thereby making a false claim to ownership.
Key Words: Digital image watermarking, Image Normalization, Robustness, Ownership Rights
1. INTRODUCTION A wide range of techniques for embedding invisible watermarks into images have been introduced in recent literature. Most of these approaches focus on demonstrating robustness against common image processing operations or intentional attempts to remove the watermark. However, studies [2, 3, 4, 5] have revealed that the ability to embed a robust watermark does not necessarily guarantee that the watermarking scheme can reliably establish ownership.
A second, more nuanced condition is that the legitimate owner must possess a version of the image that contains no watermark other than their own. We refer to this as REQ2. Importantly, the owner does not necessarily need to retain the original, unmarked image; a version bearing only their watermark is sufficient, provided that no other watermark can be detected within it. We refer to such a version as the pseudo-original, to distinguish it from the actual, unwatermarked original. If REQ2 is not met, it becomes possible for an attacker to claim that their watermark is also present in the pseudo-original, thereby creating a situation in which both the attacker and the rightful owner appear to have equal proof of ownership.
The attack strategy described in [3] targets watermarking schemes that rely on the original image during the detection phase. To illustrate this attack, consider the following scenario: a user, Good (G), embeds a watermark into his original image (O) and distributes the watermarked version (O′). An adversary, Bad (B), who has access to O′, intends to falsely claim ownership. Assuming an additive watermarking scheme that requires the original image during detection (such as the scheme in [6]), B subtracts his own watermark from O′ to generate a new image F, which he then presents as his original. Because F is derived in this manner, the difference O′ – F will correlate strongly with B’s watermark. Similarly, O – F will also exhibit this correlation. As a result, if F is treated as the original during detection, both O and O′ appear to contain B’s watermark. This means B can produce the same level of evidence to support his ownership claim as G can, effectively undermining the integrity of the watermarking scheme. Another variant of this attack, targeting a modification of the method in [6], is discussed in [2].
© 2025, IRJET
|
Impact Factor value: 8.315
The above discussion establishes that both REQ1 and REQ2 are individually necessary to assert rightful ownership. We now argue that these two conditions, taken together, are also sufficient. REQ2 ensures that the legitimate owner has access to a pseudo-original that is uncontaminated by other watermarks. Consequently, any adversary attempting to claim ownership must also produce such a pseudo-original, which is infeasible. Assuming the owner only distributes watermarked copies, REQ1 guarantees that any unauthorized attempts to remove or replace the watermark
|
ISO 9001:2008 Certified Journal
|
Page 164