University of Konstanz
Graduiertenkolleg / PhD Program
Computer and Information Science

Colloquium of the Department and the PhD Program

title

Data Hiding in Digital Images

speaker

Prof. Dr. Xingpeng Zhang, University of Konstanz, Germany

date & place

Wednesday, 30.11.2011, 16:15 h
Room C 252

abstract

Data hiding is a technique to imperceptibly embed additional data into digital multimedia for various purposes, such as copyright protection, covert communication and content recovery. A data-hider always hopes to lower the distortion caused by data hiding or to maximize the embedded payload with a given distortion level, in other words, to achieve a good "rate-distortion" performance. This talk will introduce several recent efficient data-hiding schemes based on multi-layer construction. Firstly, by exploiting Wet Paper codes and Hamming codes, a family of data-hiding methods can be derived from an existing binary embedding method. If the binary embedding method is near optimal, the performance of the derived methods will still stay rather close to the performance bound of binary embedding with arbitrarily small embedding rates. Secondly, after performing the binary embedding in LSB of cover samples, we can embed some additional secret data into the second LSB by increasing or decreasing the cover samples by one. When the performance of the binary embedding method in LSB achieves the performance limit, the performance of the double-layered scheme can also achieve the upper bound of 1 embedding. Thirdly, if the allowable modification range on cover samples is [-2, +2], the LSB, second LSB and third LSB of cover samples can be controlled to carry the secret data by using a triple-layered construction. It is also shown the rate-distortion performance of the triple-layered method approaches the theoretical limit of 2 embedding.