A Design and Implementation of Graphical User Interface (GUI) for

Three-Dimensional Model Watermarking

Chen-Chung Liu, Pei-Chung Chung
Department of Electronic Engineering,
National Chin-Yi University of technology,
Department of Electronic Engineering,
National Chin-Yi University of Technology
No. 35, Lane 215, Sec. 1, Chung-Shan Rd., Taiping, Taichung, 41111, Taiwan, R.O.C.
ccl@ncut.edu.tw


Abstract

In Recent times, with the fast development of
internet and the large development of digital media
information processing and content distribution,
digital information included three models is easy to
transmit and duplicate unauthorized reproduction to
become a serious problem. There is a critical demand
for techniques to protect the copyright of the original
digital data and to prevent unauthorized duplication
or tampering. Generally there are two good
technologies for the intellectual property and
copyright protection; cryptography and
watermarking.
MATLAB is a data visualization software
package and programming environment. It has a
graphical user-interface programming tools such that
the problem in MATLAB is solved interactively and
the results displayed graphically. MATLAB also has
a great GUI (graphical user interface) development
environment for creating applications. In this paper, a
graphical user interface (GUI) program has been
developed and implemented for three dimensional
model watermarking. The present GUI is a graphical
mouse-driven program that allows the user to create,
simulate and analyze 3D model watermarking
systems interactively. The program code consists of a
set of files, which can be modified to match other
purposes as well. This program may also serve as a
tool for real-time monitoring and process control.

Keywords: MATLAB, graphical user interface,
watermarking, 3D model.

1. Introduction

Recently, with the fast development of internet
and the rapid development of digital media
information processing and content distribution,
digital information is easy to transmit and duplicate
unauthorized reproduction becomes a serious
problem. There is an urgent demand for techniques to
protect the copyright of the original digital data and
to prevent unauthorized duplication or tampering.
Generally there are two good technologies for the
intellectual property and copyright protection;
cryptography and watermarking.
The cryptography technology cut off the access
of the unauthorized person after the multimedia
information is encrypted [1- 5]. However, it cannot
prevent the unlawful action of an authorized person
and cannot solve the problem that some copyright
owners assert their ownerships for one content. To
solve the problems of the cryptography, there have
been much researched in watermarking technology,
which is the end-step in information security and
protects the copyright of owner by embedding the
watermark into the multimedia information. A lot of
research has been carried out to protect the copyright
protection of image, video, and audio. Digital
watermarking is a technique designed to hide
information in a certain type of digital data.
Embedded watermarks can be used to enforce
copyright, data authentication or to add information
to the data. Ideally, the watermark should not
interfere with the intended purposes of the data. In
last decades, most of the research on watermarking
has concentrated on audio signals, images, or video
sequences [6- 9]. The watermarking algorithm for
3D models are few because the watermarking
technique for 3D model has many difficulties for the
following reasons: (A) compared with images, only a
small amount of data (ie. vertices) is available for
watermark embedding; (B) no unique representation
nor implicit ordering of 3D model data exists; and (C)
no robust transformation field could be used to
embed watermark [10]. In recent years, 3D graphic
models, such as VRML, MPEG-4, and 3D
geometrical CAD, have become very popular leading
the development of 3D watermarking algorithms to
protect the copyright of 3D graphic models [11].
Ohbuchi [12] proposed several watermarking
algorithms for 3D models: triangle similarity
quadruple (TSQ) embedding algorithm, tetrahedral
volume ratio (TVR) embedding algorithm, and mesh
density pattern embedding algorithm. However, these
algorithms are not sufficiently robust against attacks.
Beneden [13] also described a watermarking system
that is based on affine registration of meshes in order
to compensate for affine transformations and used it
in the watermarking detection procedure. Although
this algorithm is robust against the randomization of
points, mesh altering, and polygon simplification, it
is not robust against cropping attacks. Kang kang Yin
et al. [14] proposed a mesh watermarking scheme for
triangular meshes. In their scheme watermark
information is embedded into a suitable coarser mesh
which consists of the low-frequency components.
The scheme is not robust against crop operation.
The term cluster is an unusual aggregation of
events that are grouped together in time or space [15].
Cluster analysis is one of the basic tools for
exploring the underlying structure of a given data set.
The primary objective of cluster analysis is to
partition a given data set of multidimensional vectors
(patterns) into so-called homogeneous clusters such
that patterns within a cluster are more similar to each
other than patterns belonging to different clusters.
Cluster seeking is very experiment-oriented in the
sense that cluster algorithms that can deal with all
situations are not yet available; each approach has its
own merits and disadvantages [16]. Further
information on clustering and clustering algorithms
can be found in the literature [17- 20]. Clustering has
been applied in a wide variety of engineering and
scientific disciplines such as medicine, psychology,
biology, sociology, pattern recognition, and image
processing. We believe that clustering for 3D model
vertices will have much effect in the 3D model
watermark. In this proposed, we develop a cluster
algorithm based on vertices connectivity and the 3D
model reconstruction with star- topology to construct
a robust watermarking scheme for 3D model.
For programmers, to develop a visualization
control system always needs higher programming
techniques. For examples, one want to use the
Window-based computer as the platform of a control
system, he has to solve the problem of image
extraction and the interface between control signal
and system drive. On the other hand, one should
consume great much time for programming and test
if he want to use the C language or the Java language
to process the visual images. The claim for
easy-to-use tools is a real factor in creating a GUI on
top of the existing subroutines. By seeing a graph and
being able to control the data processing, the process
becomes much easier. In this paper, we use develop a
GUI on Matlab for 3D model watermarking. We hid
the available algorithms from the users, so the users
could concentrate on the data analysis and the
generation of the desired experiment results instead
of dealing with technical details. A usable GUI is
primarily practical, rather than theoretical. Our
proposed GUI is a graphical mouse-driven program
that allows the users to create, model, simulate and
analyze 3D model watermarking systems easily and
interactively [21- 23].
To explore the utility and demonstrate the efficiency
of the proposed scheme, simulations under various
conditions are conducted. The experiment results
show that our proposed scheme is a robust
watermarking scheme for 3D model. The remainder
of this paper is organized as follows: In Section 2,
program structure is described. In Section 3, The
Block of Clustering an Input 3D model is presented.
The Block of Finding Suitable Vertices for
Watermarking is illustrated in Section 4. The
watermark embedding Block is presented in Section
5. Watermark Extraction Block is explained in
section 6. Section 7 presents the Quality Analysis
Block of Extracted Watermark. Finally, Section 8
concludes this paper.

2. Program structure

Each current program is frequently composed
of two main parts: the program kernel and the GUI.
The GUI is a kind of bridge connecting the user and
the main program, allowing them to interact. In the
case of programs for computer simulation, the data
flow between the software and the user can be
considered as a sequence of three phases:
pre-processing, processing and post-processing.
In order to offer a 3D model watermarking
GUI with different ways of the button assignments
and the button functions, we develop a system based
on the block interpretation to construct 3D model
watermarking GUI and set the functions of the
buttons. The 3D watermarking GUI allows a user to
embed watermarks into a 3D model and to extract
watermarks from a watermarked 3D model. The 3D
watermarking GUI program is a set of m-files and
consists of four major parts (an m-file is a
user-defined function or script file composed of
existing MATLAB commands and functions): the
clustering of an input 3D model, the finding of
suitable vertices for watermarking, the extracting of
watermarks from a watermarked 3D model, and the
quality analysis of extracted watermarks. Figure 1
shows the Block diagram of the algorithm for
showing a GUI on the screen through the
combination all of the blocks.
{ } m
{ } m of length
{ } m of length


Figure 1. Block diagram of the 3D model watermarking
A robust watermarking scheme for 3D models
must be extremely secure without reducing the visual
quality of the cover 3D model and must be robust to
against the attacks of translating, rotating, and
scaling after the watermark. In order to construct a
superior watermarking scheme for 3D model, several
schemes are used in this paper to achieve the goal.
The overall watermark embedding process is shown
in Figure 2.

N
C C , ,
1
K
N
C C , ,
1
K { }
e
V
{ } m of length
Key
W
{ } m
H
PN
Model D
d Watermarke
3
Model D
Original
3

W: Watermark
TA: Total Automorphism
ABHS: Arrange the Bits of Hashed Watermark image into a
Sequence
CCV: Clustering the Model with the Connectivity of Vertices
RST: Reconstruct the 3D Model with Star-Topology Elements
EVS: Embedding Vertices Selection
ASKM: Embed the Watermark with Angle Shift Keying
Modulation

Figure 2. The flow chart of finding the candidate

vertices for watermark embedding.

The watermark W is transformed to H by toral
automorphism (TA) using a pseudo random sequence
(PN) generated by a private key to enhance the
security [24]. The hashed watermark H is then
arranged to a binary random sequence {m}. On the
other hand, the host 3D model X is divided into
several sub-models (clusters) C
1
, C
2
, , C
N

according the structure of the host 3D model. The
starting vertex of each cluster is found and the
reference vertex of the 3D model is selected from
these starting vertices. Each cluster is reconstructed
with star- topology elements and candidate vertices
for watermark embedding are collected. In the same
time, the 3D model is translated such that the
reference vertex is coincide with the original point of
the globe coordinate system, then the coordinates of
the candidate vertices are transformed into spherical
coordinate. These candidate vertices are then
arranged into a sequence {
c
} according to their
weights calculated from their spherical coordinates.
PN is used to selected watermarking vertex
m
from
{
c
} by embedding vertex selection (EVS).In the
ASK modulation, the feature angle of the
watermarking vertex
m
is modulated into by
secret bit m. After EVS and ASK, contains the
secret information is then embedded into host 3D
model X at the selected candidate vertex by
watermark bits embedding (WBE). The above
embedding process is repeated for each secret bit in
{m} and its corresponding vertex. The details of each
block of our watermark embedding algorithm are
described in the following.
3. The Block of Clustering an Input 3D
model

Recently, three dimensional (3D) models are
used in various applications, such as computer
graphics, virtual reality, 3D animation and synthetic
imaging systems. Many representations have been
proposed for 3D models [25].In particular, triangular
meshes are frequently being used to represent 3D
model surfaces. A triangulated meshes 3D model is
represented by its topological, geometry and
attributes list. The topological list describes the
connectivity relations among vertices and the
incidence relation between triangles and vertices.
The geometry list specifies the locations of the
vertices. The attribute list generally consists of colors,
normal vectors and texture information, which are
needed to paint and shade the model. Geometry and
attribute list are specified by floating-point numbers,
whereas topological list are represented by integer
indices. A common scheme for representing and
storing polygon meshes is to use a list of vertex
geometry coordinates to store the geometry and a list
of vertex indices for each face to store mesh
connectivity. Edges are implied and not explicitly
stored (edge is a line segment that connected two
adjacent vertices). [26]
The coordinates of 3D model vertices should
be changed when people embed watermarks into 3D
model vertices. On the other hand, vertices are
connected to adjacent vertices with edges to form
triangle meshes, connected triangle meshes are
collected to form a simple 3D model, and several
simple 3D models are grouped to construct a 3D
model. For the convenience of finding candidate
vertices for watermark embedding, one has better to
decompose the 3D model into none-overlapped
simple 3D models according to the 3D models
original structure.
Clustering of data in multi- dimensions has
been applied in a variety fields like as image
segmentation, pattern recognition, and so on.
Clustering is the process of partition the data into
groups of items such that items within a group are
similar to one another and different from those in
other groups, the similarity between items is
determined based on their features. We cluster a 3D
model into clusters (simple 3D models) based on the
connectivity relation of vertices from the topology
list of the 3D model. Figure 3 shows the result
clusters of a 3D model teapot with 4 objects: body,
handle, spout, and lid.

(a)

(b) (c)

(d) (e)

Figure 3. The result clusters of a 3D model teapot
with 4 objects: (a) integrated teapot, (b)
cluster body, (c) cluster handle, (d) cluster
spout, and (e) cluster lid.


4. The Block of Finding Suitable Vertices
for Watermarking

A three dimensional model is often represented
with its surface that composed by triangles. In a
triangular mesh representing 3D model, each edge is
either shared by two triangles, called as an interior
edge, or belongs to a single triangle, called a
boundary edge. A closed loop formed by linking up
such boundary edges forms a boundary of the mesh.
Note that a 3D model may have multiple boundaries.
We call two triangles as adjacent triangles if they
share an edge. A star- topology is a polygon that
admits a triangulation in which all triangles have a
common vertex, called as center vertex. The other
vertices of the polygon are called as front-vertices.
The front-vertices and the edges between them
construct the front of a star- topology. The number of
front-vertices d is the degree of the center of the star-
topology [27].
When we make a revision against to a vertex,
the relative positions among the revised vertex and
those vertices that connected to the revised vertices
are actually changed. If this kind of chain-reaction
does not be overcome, then the 3D model may be
di st or t ed and t he i nf or mat i on whi ch al r eady


embedded at other vertices before shall be destroyed.
In order to overcome this question, we must avoid
embedding watermark bits on adjacent vertices. In
order to get the goal, we reconstruct the simple 3D
model using star- topology as the elemental element
to divide a simple 3D model into non- overlapped
star- topologies such that any two adjacent star-
topologies have several sharing front- vertices. Then,
we can embed watermark at the center vertex of each
star- topology.
We use a kind of adjacent vertices searching
method to reconstruct the triangular mesh 3D model
with the star- topology 3D model. The searching
method is shown in figure 4 and is described as
follows
1. To cluster the vertices of a 3D model into several
clusters according to whether the vertices are
connected or not. The number of clusters N is one
for a simple 3D model, and the number of clusters
N is more than one for a compound 3D model that
is constructed with several simple 3D models.
2. To find a feature vertex that has maximum vertex-
length among maximum degree vertices in the i-th
cluster to do as the starting vertex
I
s
of the i-th
cluster, i = 1, 2, . . ., N. The feature vertex of a
cluster is also the center vertex of the first star-
topology of this cluster.
3. To connect these front- vertices of that feature
vertex to form the first star- topology for the cluster.
4. To construct new star- topologies that are adjacent
to the existent star- topology.
5. To repeat step 4 till each object of the 3D model
is completely composed with star- topologies.
6. To find a vertex that has maximum vertex- length
among maximum degree vertices among these
starting vertices of a 3D model does as the
reference vertex
r
of the SB mouel.
7. These centers of star- topology except the
reference vertex of a 3D model are collected to be
the candidate
8. vertices for watermark embedding. Figure 5
shows examples of candidate vertices selected
with our proposed method.

Model D 3
N
C C C , , ,
2 1
K
s
N
s s
v v v , , ,
2 1
K
CVW

CCV: Cluster the 3D Model with the Connectivity of Vertices

SF: Starting Vertex Finding
CST: Construct the 3D Model with Star-Topology
CVW: Candidate Vertices For Watermarking

Figure 4. The flow chart of searching method of

candidate vertices for watermarking.



(a)
(b)

Figure 5. Candidate vertices (points in blue) and
starting vertices (points in red): (a) simple
3D model Dragon; (b) compound 3D model
dolphin with twelve 285 objects.

5. The watermark embedding Block

We input the length of the watermark bit
sequence and a seed into the PN sequence generator
to produce a PN sequence. The embedding vertex is
selected (EVS) one by one from the candidate
vertices sequence {
c
} according to the PN sequence.
We choose a triangle from the star- topology of the
embedding vertex as the embedding triangle; the
embedding triangle has the minimum area among the
triangles that shearing the embedding vertex. In the
embedding triangle, we define that the shorter side

c
s between the two sides sharing the embedding
triangle as the feature side and its corresponding
angle as the feature angle.
In the angle shift keying modulation (ASKM),
the angle interval [u
o
, 9u
o
] aie equally divided into
n
10 4 , n = 1, 2, , 5 subintervals shown as in
figure 6. The feature angle is modulated into the
modulated feature angle
i
according to the section
that its original position placed.

1 1
0 0 ) 1 4 ( k
) 2 4 ( k
) 1 4 ( + k
) 2 4 ( + k
) 3 4 ( + k
k 4
Figure 6. The signal constellation of the ASKM

After EVS and ASK, contains the secret
information is then embedded into host 3D model X
at the selected embedding vertex by watermark bits
embedding (WBE). In the watermark bits embedding,
the embedding vertex
c
(x
c
, y
c
, z
c
) is translated to
'
c
(x'
c
, y'
c
, z'
c
) along the feature side
c
s such that
the modulation feature angle equals to
i
. The above
embedding process is repeated for each secret bit in
{m} and its corresponding vertex, then the
watermark embedding is finished.

6. Watermark Extraction Block

The watermark extraction is the inverse
process of a watermark embedding process; it is to
recover the original watermark image from a
watermarked 3D model. Most of The retrieval algorithm
is identical to the embedding process until it reaches the
angle shift keying modulation process described in the
previous section. The flow chart of the watermark
recovering process is shown in figure 7.

N
C C , ,
1
K
N
C C , ,
1
K { }
e
V
{ } m of length
Model D
d Watermarke
3
Key
W
{ } m
H
PN


ITA: Inverse Total Automorphism
AWBI: Arrange the extracted watermark bit sequence into a
2D image
CCV: Clustering the Model with the Connectivity of Vertices
RST: Reconstruct the 3D Model with Star-Topology Elements
EVS: Embedding Vertices Selection
ASKD: Extracting the Watermark with Angle Shift Keying
Demodulation
W:Watermark Extracted from Watermarked 3D Model

Figure 7. The flow chart of the watermark recovering
process.


The watermarked 3D model X is divided into
several sub-models (clusters) C'
1
, C'
2
, , C'
N

according the structure of X. The starting vertex of
each cluster and the reference vertex of X is found.
Then, each cluster is reconstructed with star-
topology elements and candidate watermarked
vertices are collected. In the same time, X is
translated such that the reference vertex is coincide
with the original point of the globe coordinate system,
and then the coordinates of the candidate embedded
vertices are transformed into spherical coordinate.
These candidate embedded vertices are then arranged
into a sequence {'
c
}. The same PN sequence used in
the embedding process is used to selected the
watermarked vertex '
m
from {'
c
} for the ASK
demodulation. In the ASK demodulation, the secret
bit m is extracted from the modulated feature
angle
"
of the selected feature triangle for each
selected star- topology. In the process of the ASK
demodulation the modulated feature angle
"
is
processed to recover the embedded secret bit m
according to the following formula

+ < +
+ <
=


) 3 4 ( ) 1 4 ( , 1
) 1 4 ( ) 1 4 ( , 0
k k if
k k if
m
w
w
(1)

After all the secret bits are extracted from the
ASK demodulation, they are rearranged into the two-
dimensional image H. By passing H through the
inverse toral automorphism (ITA), the recovered
watermark W is obtained.

7. The Quality Analysis Block of
Extracted Watermark

Several 3D models are conducted to simulate under
various conditions to explore the utility and
demonstrate the efficiency of the proposed scheme.
The 3D models Dragon (simple model, 1257 vertices,
and 2730 triangles) and dolphin (285 objects, 32828
vertices, and 61836 triangles) are used in simulation
for demonstrating the performance of the proposed
scheme. The logotype image of NCUT with sizes
8*8 and16*16 are used as watermarks for the host
3D model Dragon, the chess board with size
(8k)*(8k), k= 1, 2, . . ., 9 are used as watermarks for
the host 3D model dolphin. In order to demonstrate
host 3D model possesses excellent imperceptibility
without noticeable degradation, we compare it with
the test 3D models. Figure 8(a) shows the original 3D
model of Dragon. Figure 8(b) is the embedded 3D
model of Dragon with watermark of size16*16.
Figure 9(a) shows the original 3D model of dolphin.
Figure 9(b) is the embedded 3D model of dolphin
with watermark of size 80*80. Figure 8 and Figure 9
show that They are hard to identify the difference
between the watermarked 3D model and the original
3D model.
(a) (b)

Figure 8. The quality of the host 3D model: (a) the
Original 3D model Dragon, (b) the
Watermarked 3D model Dragon with SNR
70.3688.

(a) (b)

Figure 9. The quality of the host 3D model: (a) the
Original 3D model dolphin, (b) the
Watermarked 3D model dolphin with SNR
84.8747.

Imperceptibility is an important factor in
watermarking. We employ the SNR [28] to measure
the degree of transparency in this paper. To measure
the SNR of a watermarked 3D mesh object, the
following formula is used:

SNR = 1u log
10
_
(x

2
+y

2
+z

2
)
N-1
=0
((x

-x

)
2
+(y

-y

)
2
+(z

-z

)
2
)
N-1
=0
] (2)

Where
) , , (
i i i
z y x
and )
~
,
~
,
~
(
i i i
z y x are the coordinates of
vertex v
I
c
before and after the watermark embedding,
respectively. We calculate the SNR for embedded
Bunny and embedded dolphin. They are arranged in
table 1 to show the watermarks perceptibility, and
sketched SNR_ Watermark Size into figure 10 to
describe the change of SNR caused by the watermark
size. The SNR of the watermarked 3D model is
almost reciprocal to the size of watermark. The ratio
of ASKM to watermark size is around 60%, this
means that 60% watermarked vertices have to be
translated when the watermark is embedding into the
3D model.




Table.1: The performance of our scheme for the
test 3D models
3D Model Num. of
Objects
Mesh Vertices Candidate
Vertices
Bunny 1 2915 1494 391
Dolphin 285 61836 32828 13502
3D Model
Size of
Watermark
SNR
Num. of
ASKM
Ratio of
ASKM
Bunny
88 70.3688 39 60.93%
1616 66.2003 148 57.81%
Dolphin
88 93.9051 41 64.06%
1616 85.3686 158 61.74%
2424 85.3661 348 60.41%
3232 85.5171 592 57.81%
4040 85.0077 928 58.00%
4848 84.4162 1391 60.37%
5656 84.5377 1856 59.18%
6464 84.5998 2391 58.39%
7272 84.4468 3135 60.47%
8080 84.8747 3836 59.93%




Figure 10. The performance of our scheme for the
test 3D model dolphin

8. Conclusion

Digital watermarking is a promising method to
discourage unauthorized copying or to attest the
origin of digital data, including audio, video, images,
and 3D models.
MATLAB is a data visualization software
package and programming environment. It has
excellent numerical algorithms, very good graphical
capabilities, and a complete set of object oriented and
graphical user-interface programming tools such that
the problem in MATLAB is solved interactively and
the results displayed graphically. MATLAB also has
superior graphics handling and visualization
capabilities, and has a great GUI (graphical user
interface) development environment for creating
applications. In this paper, we combine several
toolboxes of MATLAB to develop a GUI of 3D
model watermarking system. The present GUI is a
graphical mouse-driven program that allows the user
to create, simulate and analyze 3D model
watermarking systems interactively.
In this proposed, we construct a MATLAB
GUI of 3D model watermarking system to develop a
cluster algorithm based on vertices connectivity and
the 3D model reconstruction with star- topology to
construct a robust watermarking scheme for 3D
model. In our scheme, the center of each star-
topology is selected to be a candidate vertex for
watermark embedding. There are four advantages
compare with other methods: (A) The proposed
MATLAB GUI of 3D model watermarking systemis
a graphical mouse-driven program that allows the
users to create, simulate and analyze the 3D model
watermarking algorithms interactively, easily, and
time saving . (B) The secret image has very high
security owing to using toral automorphism to scatter
the Watermark (C) The result host 3D model after
watermark embedding possesses excellent
imperceptibility without noticeable degradation (D)
All the extracted watermarks from the watermarked
simple or compound 3D models have the same
quality (100 % validity even under the combinational
attacks of rotations, translations, and scaling).


References

[1] L. M. Marvel, C. G. Boncelet, J.r., & C. T. Retter,
Spread Spectrum Image Steganography, IEEE
Trans. on Image Processing, 8 (8),1999,
1075-1083.
[2] W. Y. Chen & C. C. Liu, Multiple-Watermarking
Scheme of the European Article Number Barcode
Using Adaptive Phase Shift Keying Technique,
optical Engineering, 46 (6), 2007, 1- 12.
[3] J. Fridrich, M. Goljan, & R. Du, Deteting LSB
Steganography in Color and Gray-Scale Images,
IEEE trans.on Multimedia, 8 (4), 2001, 22-28.
[4] R. Chandramouli & N. Memon, Anlysis of LSB
Based Image Steganography Techniques,
Proceeding of Image processing, 3, 2001,
1019-1022.
[5] G. Mastronardi, M. Castellano, & F. Marino,
Steganography Effects in Various Formats of
Images, International Workshop on Intelligent
Data Acquistion and Advanced Computing
System: Technology and Applications, 2001,
116-119.
[6] C. C. Liu & W. Y. Chen, Multiple-watermarking
scheme for still images using the discrete cosine
transform and modified code division multiple-
access techniques, Optical Engineering,45 (7),
2006, 1- 12.
[7] W. Y. Chen & C. C. Liu, Robust watermarking
scheme for binary images using a slice-based
large-cluster algorithm with a Hamming Code,
Optical Engineering ,45 (1), 2006, 1-P10.
[8] W. Y. Chen & C. H. Chen, A robust watermarking
scheme using phase shift keying with the
combination of amplitude boost and low
amplitude block selection, Pattern Recognition,
38, 2005, 587-598.
[9] H. M. Chao, & Y. S. Tsai, A Bi-Polar
Multiple-Base Data Hiding Technique on
Information Security and Authentication, IPPR
Conference on Computer Vison, Graphics and
Image Processing, 2000, 307-313.
[10] Z. Zude, A. Qingsong, & Q. Liu, A SVD-based
Digital Watermarking Algorithm for 3D Mesh
Models, The 8th International Conference on
Signal Processing, 4, 2006 , 16-20.
[11] K.-R. Kwon, J. S. Sohn, Y. Huh, & S. H. Lee ,
The Watermarking for 3D CAD Drawing using
Line, ARC, 3D face Components, 2006 IEEE
International Conference on Multimedia and
Expo, 2006, 1361-1364.
[12] Yang, S. G. Li, C. X. Sun, S. H. Xu, & Y. Qun,
A Robust 3D Model Watermarking Scheme
Based on Feature Recognition, Eighth ACIS
International Conference on Software
Engineering, Artificial Intelligence, Networking,
and Parallel/Distributed Computing, 2007, 989
- 993.
[13] Benedens O, Geometry-based watermarking of
3D models, IEEE Computer Graphics and
Applications, 19 (1), 1999, 46-55.
[14] Y. K. Kang, P. Zhigeng, S. Jiaoying, & Z. David,
Robust mesh watermarking based on
multiresolution processing, Computer Graphics,
25, 2001, 409 420.
[15] T. Kanungo, D.M. Mount, N.S. Netanyahu, C.D.
Piatko, R. Silverman, & A.Y. Wu, An efficient
k-means clustering algorithm: analysis and
implementation, IEEE Transactions on Pattern
Analysis and Machine Intelligence, 24 (7), 2002, .
881-892.
[16] M. C. Su & C.H. Chou, A modified version of
the K-means algorithm with a distance based on
cluster symmetry, IEEE Transactions on
Pattern Analysis and Machine Intelligence, 23(6),
2001, 674-680.
[17] C. Dematte, N. Molinari, & J. P. Daurs,
Arbitrarily shaped multiple spatial cluster
detection for case event data, Computational
Statistics and Data Analysis, 51 (8), 2007,
3931-3945.


[18] A. Mattias, G. Joachim, & L. Christos,
Approximate distance oracles for graphs with
dense clusters, Computational Geometry: Theory
and Applications, 37 (3), 2007, 142-154.
[19] Y. Zhang , W. Wang, X. Zhang, & Y. Li, A
cluster validity index for fuzzy clustering,
Information Sciences, 178 (4), 2008, 1205-1218.
[20] W. Q. Xu, and S. W. Golomb, Optimal
interleaving schemes for correcting two
dimensional cluster errors, Discrete Applied
Mathematics, 155 (10), 2007, 1200-1212.
[21]M. H. Eres, G. E. Pound, Z. Jiao, J. L. Wason, F.
Xu, A. J. Keane, & S. J. Cox, Implementation
and utilization of a Grid-enabled problem
solving environment in Matlab, Future
Generation Computer Systems, 21, 2005,
920-929,.
[22] M. M. Teixeira, E. Assuncao, & M. R. Covacic,
Proportional Controllers: Direct Method for
Stability Analysis and MATLAB
Implementation, IEEE Transactions on
Education, 50 (10), 2007, 56-65.
[23] Y. J. Choi, K. McCarthy, & M. J. McCarthy, A
MATLAB graphical user interface program for
tomographic viscometer data processing,
Computers and Electronics in Agriculture, 47,
2005, 59-67.
[24] G. Voyatzis & I. Pitas, Application of Toral
automorphisms in Image Watermarking
Proceedings, International Conference on Image
Processing, 1996, 237-240.
[25] R. Pajarola, & J. Rossignac, Compressed
Progressive Meshes, IEEE Transactions on
Visulization and Computer Graphics, 6 (1),
2000, 79-93.
[26] J. H. Ahn, C.S. Kim, & Y. S. Ho, Predictive
Compression of Geometry, Color and Normal
Data of 3-D Mesh Models, IEEE Transactions
on Circuits and System for Video Technology,
16 (2), 2006, 291-299.
[27] S.P. Mudura, S. V. Babjib, & D. Shikhare,
Advancing fan-front: 3D triangle mesh
compression using fan based traversal, Image
and Vision Computing, 22, 2004, 11651173.
[28]S. Zafeiriou, A. Tefas, & I. Pitas, Blind robust
watermarking schemes for copyright protection
of 3D mesh objects, IEEE Transactions on
Visualization and Computer Graphics,
11(5), 2005,596 607.