Open AccessData-Loss resilience video steganography using frame reference and data ensemble reconstruction
Author(s) -
Feng Yong Li,
Jiang Yu,
Yan Ren
Publication year - 2019
Publication title -
mathematical biosciences and engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.451
H-Index - 45
eISSN - 1551-0018
pISSN - 1547-1063
DOI - 10.3934/mbe.2019228
Subject(s) - computer science , vandermonde matrix , steganography , distortion function , robustness (evolution) , redundancy (engineering) , information hiding , data redundancy , frame (networking) , distortion (music) , artificial intelligence , data mining , computer vision , algorithm , theoretical computer science , embedding , decoding methods , computer network , amplifier , biochemistry , eigenvalues and eigenvectors , physics , chemistry , quantum mechanics , bandwidth (computing) , gene , operating system
In this paper, we propose a robust video steganographic method, which can efficiently hide confidential messages in video sequences, and ensure that these messages are perfectly reconstructed by recipient. To apply proposed scheme to video sequences, we must be faced with two nontrivial problems: (a) how to effectively minimize the total steganographic distortion for each video frame? (b) how to recover the hidden messages if some frames are lost or damaged? We tackle the first question by designing a new distortion function, which employs two continuous adjacent frames with the same scene as side-information. The second question is addressed by data sharing. In this mechanism, the original data is expanded and split into multiple shares by using multi-ary Vandermonde matrix. Since these shares contain a lot of data redundancy, the recipient can recover the hidden data even if some frames are damaged or lost during delivery. Extensive experiments show that proposed scheme outperforms the state-of-the-arts in terms of robustness and diverse attacks.