Database Reference
In-Depth Information
Restoring the Original Image.
The algorithm restores the watermark message (wm)
as follows: (1) If the LSB of
x
=1
add the LSB of
y
to wm and set the LSB of
x
and
y
to 0, (2) If the LSB of
x
=0
add the LSB of
y
to wm and set the LSB of
x
and
y
to 1. For restoring the original image, Coltuc's scheme then performs the inverse
transformation depicted in Equation 3.
x
=
2
3
x
+
1
3
y
, y
=
1
3
x
+
2
3
y
(3)
3
Fingerprint Pattern Data - A Motivating Example
Fingerprints, that are traces of an impression from the friction ridges of any part of a
human hand, have a long tradition in crime detection [15]. In general, the latent fin-
gerprints detected at a crime scene are used for fingerprint pattern identification (a.k.a.
dactyloscopy
) in order to find the criminal [2].
As with today, new approaches for
contactless
latent fingerprint detection are subject
to intensive research, especially in the field of signal or image processing [21,13,20].
As a result, the detected fingerprints are available in digital rather than physical form.
This, in turn, leads to problems that are common for digital data, especially regarding
IT security aspects like
authenticity, integrity
or
confidentiality
.
Furthermore, for the legal usage of digital fingerprints, we have to ensure another im-
portant aspect: the
chain-of-custody
. To this end, it is necessary to guarantee the trace-
ability of fingerprint data throughout all transformation steps, and as a result, that each
derived data item can be tracked back to the original data items. This is especially nec-
essary if the digital fingerprint data is used as a proof at court. Here, two questions have
to be answered without a doubt: Firstly, does any derived data item (used as proof) stem
from the original digital fingerprint? And secondly, can we ensure the trustworthiness
of the source provenance information? Unfortunately, neither standard DB mechanisms
such as access control nor current data provenance approaches support us in answer-
ing these questions. Hence, we must introduce an appropriate approach, which we will
sketch in the following section.
4
Using Watermarking for Data Provenance
In this section, we present an approach that allows a database system to use existing
invertible watermarking schemes to a) gather provenance information for digital fin-
gerprint data and b) ensure the trustworthiness of the provenance information. First, we
present the database infrastructure in which the presented approach should be used. Sec-
ond, we explain our approach and how it ensures reliability of provenance data. Finally,
we show parts of the implementation with help of a simplified running example.
4.1
Architecture and General Data Flow
In Figure 4, we present an abstract overview of our DB architecture and how it is em-
bedded in an infrastructure for fingerprint verification. The original data results from
