Reversible Data Hiding for 3D Point Cloud Model

Abstract

This paper proposes a reversible data hiding scheme for 3D point cloud model. It exploits the high correlation among neighbor vertices to embed data. It starts with creating a set of 8-neighbor vertices clusters with randomly selected seed vertices. Then an 8-point integer DCT is performed on these clusters and an efficient highest frequency coefficient modification technique is employed to modulate the watermark bit. In data extraction, the modified clusters are retrieved first, and other operations are the inverse process of the data hiding. The original model can be perfectly recovered if it is intact. Our scheme is suitable for some specific applications where content accuracy of the original model must be guaranteed. Moreover, the method can be used for 3D point cloud model authentication, e.g., hiding a hash sequence.