Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/37712
標題: An Effective Message Embedding Scheme for 3D Models
作者: Li, M.T.
王宗銘 
Huang, N.C.
Wu, K.C.
Jan, C.K.
Wang, C.M.
關鍵字: Message embedding;3D models;quantization;information hiding;steganography;point-sampled geometry
Project: Fundamenta Informaticae
期刊/報告no:: Fundamenta Informaticae, Volume 97, Issue 1-2, Page(s) 93-109.
摘要: 
We present an effective message embedding scheme for 3D models. We propose the unit length as the quantizer to generate an embedding order list and an embedding index list. Our scheme considers every two elements in the embedding order list as the order pair, and we embed 3 bits of 0 or 1 secret message into the index pair associated with the order pair. The message embedding is effective requiring, at most, adding 1 to, or subtracting 1 from, the index pair. This reflects a slight perturbation of a points coordinates where the magnitude of the perturbation is no greater than one unit length. Our algorithm achieves a high embedding capacity, being 4.5 times the number of points in the point cloud models. This amount of capacity allows us to convey a 502x502 resolution of the black-and-white image into a point cloud model consisting of 56,194 points for covert communication. The capacity magnitude is 50%-75% higher than that of the current state-of-the-art algorithms, yet the model distortion due to the message embedding is smaller than that of our counterparts. Our algorithm is robust against translation, rotation, and uniformly scaling operations. It is fast, simple to implement, and the message can be extracted without referring to the original point cloud model. We believe our scheme is appropriate for most point cloud models.
URI: http://hdl.handle.net/11455/37712
ISSN: 0169-2968
DOI: 10.3233/fi-2009-193
Appears in Collections:資訊科學與工程學系所

Show full item record
 

Google ScholarTM

Check

Altmetric

Altmetric


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.