Cryptography Reference
In-Depth Information
under the precision tolerance τ , the local shape of the objects in the map has
been modified. The distorted shape degrades the quality of the watermarked
map, and the invisibility of the hidden watermark. This could possibly at-
tract the attention of attackers. The main causes of the problem is that the
algorithms choose the coordinates or their transform coe
cients such as the
cover data. This is an inconvenient form to present the shape information of
the map objects. Consequently, new cover data, which can be directly related
with the maps shape information, have to be selected instead of the raw co-
ordinates in order to conveniently control the shape distortions induced by
data embedding.
(a) Original map
(b) Watermarked map
Fig. 6.6.
An example of the distorted shape.
2. Robustness to Attacks
A successful attack generally has to remove the hidden message from the
watermarked data while retaining its fidelity. Due to the critical application
environment of 2D vector maps, even slight adjustment to the coordinates
could possibly destroy the maps fidelity. It is not a good choice for an at-
tacker to manipulate the watermarked map by rotating, translating, scaling,
adding random noise, etc. However, most existing algorithms used these at-
tacks. Some operations that can manipulate the map data without losing
its fidelity, including map interpolation, simplification and data reordering,
should be regarded as the more dangerous attacks. This will be emphasized
in the following scheme.
Cover Data Extracting
To reduce the shape-distortions, the cover data adopted in the scheme is ex-
pected to explicitly represent the shape information of the map objects, which
can not be directly satisfied by the raw coordinates. A distance sequence is
extracted from the original map and used as the new cover data. Given the
original map M
o
, its feature points F =f
1
,f
2
,are detected. These
separate M
o
into n subsets denoted as C
1
,,C
i
,,C
n
. The Douglas-
Peucker algorithm [16] is used in this scheme for feature point detection
