International Core Journal of Engineering 2020-26 | Page 142
2019 International Conference on Artificial Intelligence and Advanced Manufacturing (AIAM)
Reversible Database Watermarking Based on
Differential Evolution Algorithm
Miaochan Zhao * Chuanxian Jiang Jiangbing Duan
Guilin University of Technology
College of Information Science and
Engineering
Guilin, China
e-mail:[email protected] Qiongtai Normal University
School of information
Haikou, China
* e-mail: [email protected] Anhui Si Chuang Electronics Co., Ltd.
Anhui, China
e-mail:[email protected]
Abstract—To battle against the problem of invisibility and
robustness of existing reversible database watermarking
algorithms, we propose a reversible database watermarking
method based on differential evolution algorithm. In the image
preprocessing stage, the edge detection operator is introduced,
and the image edge feature points are extracted to guide the
watermark embedding, which improves the invisibility of the
database watermark. In addition, by introducing a differential
evolution algorithm to select the optimal embedding position of
the database, the watermark is embedded in the case of the
minimum fitness function, which improves the watermark
invisibility. Finally, in the coding mode, the set of sites
consisting of database tuples and attribute columns is used as a
population individual to improve the robustness of the
watermark. The experiment proves the effectiveness of the
proposed method and the proposed algorithm has good
performance than the reversible database watermark based on
genetic algorithm and Simulated annealing algorithm.
and tracking when identifying database sources after
modifying the database source.
Although researchers have made great progress in the
research of reversible database watermarking technology, in
general, it is still in the initial stage, and there are still many
places to be improved. At present, most database watermarks
embed watermarks in the whole column of the database. For
the problem that the whole column deletion attack is less
robust and difficult to recover, this paper proposes a
reversible database watermarking technique based on
differential evolution algorithm, and uses differential
evolution algorithm to select the optimal for each data block.
Embedding the loci, the results show that the algorithm is
improved in both robustness and invisibility.
II. E DGE D ETECTION O PERATOR AND D IFFERENTIAL
E VOLUTION A LGORITHM
Keywords—Reversible watermarking; Differential Evolution
algorithm; database; Roberts operator
A. Edge detection algorithm
The Roberts operator greatly reduces the amount of data
and eliminates information that can be considered undesired.
The Roberts operator uses the difference between the two
pixels in the diagonal direction to approximate the gradient
amplitude to detect the edge, and the effect of detecting the
vertical edge is better. Other direction edges. The specific
algorithm is as follows. The difference between the diagonal
pixels is expressed as:
I. I NTRODUCTION
In the contemporary digital society, with the
advancement of information technology, the use of
information systems is increasing, and relational databases
for storing information are widely used and shared. As a
collection of data, the database stores a lot of sensitive
information and is one of the main branches of information
security. In recent years, many industries have experienced
large-scale data leakage problems, and protecting
information security is an important task.
' x f
At first, watermarking technology was used to protect the
copyright of data owners and prevent theft, tampering and
illegal dissemination of criminals. Database watermarking
technology was proposed in combination with relational
database in digital watermarking technology. The database
watermark was first introduced by Agrawal and Kiernan in
2002. The LSB position is embedded in the numeric attribute
values in the database, but this will cause data distortion.
Therefore, Sion et al. proposed a strategy for marking
numerical attributes by considering the tolerance range of
attribute values. Saman Iftikhar et al. [2] proposed a robust
reversible database watermarking algorithm in 2015,
introducing the concept of mutual information. In 2017,
Francocontreras J et al. [5] proposed a relational database
robust lossless watermark based on circular histogram
modulation, which can be used for database authentication
978-1-7281-4691-1/19/$31.00 ©2019 IEEE
DOI 10.1109/AIAM48774.2019.00031
' y f
f ( i , j ) f ( i 1, j 1)
f ( i , j 1) f ( i 1, j )
(1)
(2)
Among them f ( i , j ) , f ( i 1, j ) , f ( i , j 1) ,
f ( i 1, j 1) re-present the pixel value of one pixel in the
image, and the pixel point represented by it The relative
relationship is as follows:
Fig 1. Pixel distribution
120