Reversible Data Hiding for Encrypted 3D Mesh Models With Secret Sharing Over Galois Field

Reversible data hiding in encrypted 3D models (RDHEM) is an emerging steganography technique, capable of both encrypting the cover model to ensure confidentiality and embedding additional messages for covert communication. However, the embedding capacity provided by recent RDHEM methods is still at a low level. In this paper, an adaptive vertex grouping strategy is proposed, which can divide the vertices in the cover 3D model into groups. Then, the multi-MSB prediction and Huffman coding are exploited to compress the data volume of vertices. Through proper vertex grouping and efficient data compression of the model vertices, the embedding capacity of the RDHEM can be effectively improved. Additionally, two schemes for 3D model encryption are provided. One is based on a secret sharing method over the Galois field and the other leverages the stream cipher technique. Experimental results show that the embedding capacity of the two proposed schemes significantly outperforms state-of-the-art schemes.