IEEE 3361-2025

This document describes an evaluation method to assess the robustness of watermarking implementations in unstructured digital content, including images, video, audio, and documents. This document describes attacks and corruptions that may result in the quality compromise of the watermarking information embedded in digital content. Furthermore, the document specifies robustness requirements for implementations of digital watermarking in the digital content. Moreover, the document defines metrics used for the evaluation and the test cases for assessing the robustness against different corruptions or attacks. This document does not specify the evaluation for the robustness of the watermarking implementations in structured content, for example, database. The risks faced with watermarking implementations in structured content are different from those of unstructured content, and need different test cases.

Using the evaluation method, the relevant stakeholders of digital content utilizing watermarking technology can have a clear understanding of the robustness of the watermarking implementation. Also, based on the evaluation results, the relevant stakeholders can know the vulnerabilities in the watermarking implementations to help them make improvements.

New IEEE Standard - Active. An evaluation method to assess the robustness of invisible watermarking implementations in unstructured digital content, including images, video, audio, documents, and webpages, is described in this standard. Also described are distortion inferences that may result in the compromise of the watermark information embedded in digital content. Furthermore, the metrics used for evaluation and test cases for assessing the robustness against different distortion inferences are specified. This standard does not specify the evaluation for the robustness of the watermarking implementations in structured content, for example, a database. The risks faced with watermarking implementations in structured content are different from those of unstructured content and need different test cases.