A Reversible Data Hiding Scheme for 3D Point Cloud Model

This paper proposes a reversible data hiding scheme for 3D point cloud model. The method exploits the high correlation among neighboring 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 in the integer DCT domain is employed to modulate the watermark bit. After that, the modified coefficients are inverse transformed into coordinates in the spatial domain. In data extraction, we need to recreate the modified clusters first, and other operations are the inverse process of the data hiding. The original model can be perfectly recovered using the clusters information if it is intact. This technique is suitable for some specific applications where content accuracy of the original model must be guaranteed. Moreover, the method can be easily extended to 3D point cloud model authentication.