Theoretical framework for a practical evaluation and comparison of audio watermarking schemes in the triangle of robustness, transparency and capacity

Abstract

Digital watermarking is a growing research area to mark digital content (image, audio, video, etc.) by embedding information into the content itself. This technique opens or provides additional and useful features for many application fields (like DRM, annotation, integrity proof and many more). The role of watermarking algorithm evaluation (in a broader sense benchmarking) is to provide a fair and automated analysis of a specific approach if it can fulfill certain application requirements and to perform a comparison with different or similar approaches. Today most algorithm designers use their own methodology and therefore the results are hardly comparable. Derived from the variety of actually presented evaluation procedures in this paper, firstly we introduce a theoretical framework for digital robust watermarking algorithms where we focus on the triangle of robustness, transparency and capacity. The main properties and measuring methods are described. Secondly, a practical environment shows the predefined definition and introduces the practical relevance needed for robust audio watermarking benchmarking. Our goal is to provide a more partial precise methodology to test and compare watermarking algorithms. The hope is that watermarking algorithm designers will use our introduced methodology for testing their algorithms to allow a comparison with existing algorithms more easily. Our work should be seen as a scalable and improvable attempt for a formalization of a benchmarking methodology in the triangle of transparency, capacity and robustness.