![](bg2.png)
original contents without modifying any values, it
may endure more changes during distribution.
However, this technique is usually not flexible
because it cannot embed external information.
Moreover, it is usually time-consuming due to the
high computational complexity implementing
computer vision techniques.
In this paper, we propose a novel copyright
protection technique for digital images using
positive features of both watermarking and
fingerprinting techniques. The proposed method
uses external copyright messages and original digital
contents to generate extra scrambled information.
Random number generator is used to randomly
select the elements of digital contents. The generated
extra scrambled information contains the
information of both external copyright messages and
original digital contents. The generated information
can then be stored in a certified database with initial
seeds for random number generator. The original
digital contents can be distributed without having
any changes in element values. If there is a need to
identify either the external copyright message or the
original contents, the stored extra scrambled
information can be used for identification. The
proposed technique generates and saves extra
scrambled information as a function of external
copyright message and original contents, and
distributes the original contents. Therefore, there is
no degradation in fidelity on the distributed contents.
In addition, in generating the extra scrambled
information we use a simple and fast spatial domain
insertion method so that the proposed technique has
a very low computational complexity.
This paper is organized as follows. In section 2,
related techniques including watermarking and
fingerprinting models briefly described. The
proposed technique is explained in detail in Section
3. Experimental results and conclusion are described
in Section 4, 5, respectively.
2 RELATED WORKS
2.1 Watermarking
In typical watermarking systems, a watermark
message which contains information on
manufacturers or authorized users is embedded in
the digital contents and distributed to the outside
world. The embedded watermark message should
not be perceivable not to degrade the quality of the
digital contents. The embedded watermark message
can be extracted to resolve legal disputes on the
digital contents such as illegal copy, modification
and distribution.
Fig. 1 shows the conceptual block diagram of a
typical watermarking system. It consists of a
embedder and a detector. The embedder embeds the
watermark message to the original content and the
watermarked contents are distributed. The
watermarked contents can be modified by various
signal processing operations or malicious attacks.
The watermark detector is to detect a watermark
message form the corrupted version of the digital
contents. Various techniques have been developed
for the better performance depending on the specific
applications (Nikolaidis, 2001) (Fazam, 2001).
Figure 1 : General watermarking system.
The properties required for a watermarking
system can be varied according to the applications.
Some properties including fidelity, robustness,
computational complexity, and informed/blind
detection can be found in (Cox, 2002).
2.2 Fingerprinting
A fingerprinting technique (ISO/IEC 21000-11,
2004) is can be also used to protect digital contents
from illegal uses as in the watermarking system. The
main difference between two methods in that a
fingerprinting system extracts features from digital
contents and uses the features for identifying the
digital contents while a watermarking system
embeds an external watermark message in the digital
contents. Therefore, this fingerprinting technique
does not embed any information on the digital
contents.
The conceptual block diagram of a fingerprinting
system is shown in Fig. 2. It consists of two
fingerprint generators and a comparison part. The
technique usually employs computer vision
techniques to generate an invariant fingerprint from
the digital contents.
VISAPP 2006 - IMAGE FORMATION AND PROCESSING
28