Evaluation Metrics used in 3D Mesh Watermarking: A Study

Abstract

Three Dimensional (3D) models are watermarked to protect their ownership information and to maintain the integrity while transmitted over networked communication. These watermarking operations change the mesh details during data embedding, which in turn affects the perceptual quality or geometrical details of the mesh. There are two different situations where a mesh watermarking technique needs to be evaluated. One of them is to make sure that the watermark embedding function induces minimal perceptual distortion to the original model and the second situation is assessing the similarity between the extracted watermark and the original one to estimate robustness of the watermarking scheme. However surface mesh quality is assessed in two ways: subjective way (observation directly from the human) or objective way where some metrics are formulated by researchers to measure the mesh quality that can meet the human observation. This study covers a detailed discussion of various objective mesh quality assessment metrics proposed for detecting the mesh degradation due to different classes of watermarking techniques. Robustness estimation metrics to assess the similarity of watermarks are also discussed.