The 20th MAGDA Conference in Pacific Asia, November 14-16, 2011, Kaohsiung, Taiwan

2A32
Design and Modeling of Gripper and Cutting Tool System for Sweet Pepper
Harvesting Robot Hand
Shivaji BACHCHE1, a, Koichi OKA1, b, Hideto SAKAMOTO1, c

This paper presents a new design and modeling fundamentals of gripper and cutting system for 5 degree of freedom
robotic arm, designed to harvest sweet peppers in horticultural green house. The design consists of two parallel jaws
mounted on gears and operated with the help of servo motor. The same servo motor was used to operate the cutting
system which was composed of scissors. The complete system was designed to operate by using one servo motor on-
ly. The system model was developed in solidworks and tested for different kinematic and dynamic performances. The
performance of gripper and cutting tool system has been evaluated through simulation to determine the parameters to
design the practical prototype. Based on the design concept, the practical prototype of the gripper and cutting system
was developed by considering the results obtained by model developed in solidworks. The developed prototype was
tested to verify the feasibility and reliability of the model developed in solidworks.

Keywords： Gripper, cutting tool, gripper model, sweet pepper harvesting robot, gripper simulation

fruit already enclosed in the tube; the fruit then rolls

1. Introduction down the tube to a container. The use of vacuum grip-
pers in delicate fruit harvesting that avoids high air
The development in technology leads to wide appli- pressure and physical damage to fruit was tested by
cations of industrial robots in a large number of areas Sarig and et al. [5], Hayashi and et al. [6] and Monta and
such as assembly, material handling and machine et al. [7] with the help of vacuum technology. This type
tending, packing, picking, palletizing, gluing and of grippers shows good results for picking tightly
sealing, arc welding, spot welding, painting and coating, clustered fruits. At the same time, for vacuum grippers,
foundry applications and water jet cutting. Fruit harvest- small leakage in the system leads to failure of operation
ing is one of the important application in green house and higher construction and operating cost results in
horticulture that helps to save the labor cost and input expensive system. In addition with suction cup and
harvesting energy consumption. Robot gripper is the gripper, van Henten et al.[8] reported use of thermal
main functional part of robot arm that helps to grasp the cutting tool for cucumber harvesting in which cutting
fruit and then cut accordingly. It can also be used for was performed by a thermal cutting technique. Though
pick and place, packing or welding operations. In this thermal cutting operation has several advantages
agricultural harvesting many researchers are engaged to over bacterial transfer, it needs precise system control.
design robotic devices for efficient and delicate agricul- Thus, by considering the need of simple and eco-
tural product picking and handling. The agricultural nomical gripper, this research was carried out. Sweet
production rates are significantly influenced by utiliza- pepper is the 4th most important fruit vegetable in Japan
tion of robots and tools and techniques developed for grown on approximately 357 hector area of land which
decision support system [1]. The brief review of different needs not only high man power but also high input
types of the fruit harvesting robots and grippers is given energy consumption during harvesting operation leading
by Y. Sarig [2] and A. K. El-Kalay and et al. [3]. to increase in labor cost and production cost [9]. On the
To avoid the physical damage to fruits, manual har- other hand, as both, sweet pepper and leaves has almost
vesting is preferred which significantly increases the same color and due to that it is very difficult to recog-
total cost of fruit production. For the efficient mechani- nize them separately during automatic harvesting.
cal harvesting, the most important part is to design the Hence, by considering these issues, a sweet pepper was
proper gripper that can handle soft, delicate objects like selected for the study. Kitamura and Oka [10] developed
fruits, with respect to their various shapes and sizes. The sweet pepper picking robot in greenhouse to resolve
mechanism based on cutting device attached to a tubular these issues. This paper presents the further develop-
arm for picking soft fruits were designed by Harrell and ment in sweet pepper gripping and cutting system of old
et al. [4] which consists of a rotating lip that detaches the robotic arm. The old model was able to pick and cut
_______________________ sweet peppers by using two different servo motors for
Correspondence: 1Dept. of Intelligent Mechanical Systems gripping and cutting but in proposed design, both
Eng, Kochi University of Technology, 185, Miyanokuchi, operations intended to perform by using only single
Tosayamada, Kami City, Kochi, 782-8502, Japan
servo motor. This paper also focuses on modeling of the
TEL: 0887-57-2310, FAX: 0887-57-2320
a
146006r@gs.kochi-tech.ac.jp
prototype with various design parameters and different
aspects of kinematic and dynamic performances.

- 271 -
The 20th MAGDA Conference in Pacific Asia, November 14-16, 2011, Kaohsiung, Taiwan

2. Concept of Harvesting Robot

Fig. 2. Concept of picking tool.

Fig. 1. Concept of sweet pepper harvesting robot.
The gripper was connected to the big gears of same
size and cutter was attached on circular disc. The care
Figure 1 shows the overview of the concept of sweet was taken that the gripper links should not interfere with
pepper harvesting robot. The harvesting robot is com- small gear rotation. The gripper made from small
posed of three main units; first unit refers to recognition aluminum plates with internal sponge coating to avoid
system in which identification and location of the fruit any physical damage to the sweet pepper and scissor
was used as a cutting tool.
confirmed, second unit refers to picking system in
which grasping of fruit and then cutting operation
3.2 System Operating Mechanism
performed; and third unit refers to moving system in
which the programmed base sub-unit of the robot moves The rotating power from the servo motor was trans-
in the furrows during harvesting operation in green- ferred to the big size gears and then to small gear. The
house. In the recognition system, CCD cameras are used circular disc installed on the axis of small gear rotates
to capture the images; binarization of HIS color specifi- with the same speed of small gear while gripper con-
cation system is used to recognize the fruits and Halcon nected to the big gears rotates with the same speed of
big gear. The purpose behind using different size gears
software utilized for real time image processing. The was to obtain the speed variation in the harvesting
moving base system includes crawling tracks and operation; that means the gripping speed was slower
wheels controlled by line tracing program and carries than the cutting speed but both gripping and cutting
the robot arm manipulator on it. operations should perform simultaneously. This offers
good advantage of holding the sweet pepper slightly
earlier than cutting the stem.
3. Design of New Gripping and Cutting Tool
The gripping system and cutting system were con-
trolled by Kondo KRS 6003 HV servo motor through
3.1 System Design computer and whole system was programmed to operate
accordingly. The opening- closing of gripper and scissor
The mechanical system to which this paper repre- was controlled by program and developed in VB C++.
When servo motor rotates, the power from motor
sents is a picking system which composed of two parts;
shaft was transferred to the big gears which also rotate
the first part is aimed at grasping the sweet pepper and the gripper. At the same time, small gear meshed with
the second part is aimed at cutting the stem of sweet one of the big gear rotate the circular disc installed at
pepper from plant. These both parts have been specifi- other end of axis. Further, the scissor attached to the
cally designed with the purpose of imitating the manual circular disc starts operating with rotation of disc which
operations by labors. can be seen in figure 3. Due to variation in size of
holding area and cutting area, these big gears and small
Figure 2 shows the concept of the new gripping and gears offers speed variation that helps to hold and cut
cutting tool designed to harvest the sweet peppers. In sweet pepper at the same time.
this system, two same size gears and one small size gear
was utilized. The same size gears were meshed and
installed on supporting plate and power was provided to
one of the gear through servo motor. Further, the small
gear was meshed with one of the gear installed previ-
ously and a circular disc was attached to the top of small
gear axis. A sharp scissor was installed on the circular
disc by means of link mechanism so that it can operate
through the power provided by servo motor at base shaft. Fig. 3. Attached scissor on circular disc.

- 272 -
 The 20th MAGDA Conference in Pacific Asia, November 14-16, 2011, Kaohsiung, Taiwan

3.3 Speed Control of the System 5. Experimental Device and Experiments

Based on the results obtained from model, the grip-
Figure 4 represents the detailed block diagram of the
ping and cutting prototype was constructed. Figure 6
control system used to operate and control the gripping
illustrates built prototype with set of components used
and cutting system in which program was written in VB
in model.
C++ to interface between servo motor and computer.
The servo controlled KCB-1 was used to control the
Kondo servo motor. The motor rotates by 2700 without
restriction and the program was used to control the
rotation of the motor in two patterns, first pattern rotates
motor by 173.50 in slow speed while second pattern
rotates motor up to 202.50 in fast speed. The first pattern
assists in gripping with slow speed while second pattern
facilitates cutting with high speed. Therefore, for com-
plete and successful gripping and cutting of the sweet
pepper, motor was programmed to rotate by 202.50 with
Fig. 6. Experimental prototype.
two speed variations.
The plain carbon gears; two of module 2.5 with 36
teeth and one of same module and 18 teeth were used to
drive the system. A 50 mm circular disc was installed on
the top of small gear shaft along with scissor as shown
in figure 3. Further, Table 1 represents the characteris-
tics and properties of the experimental prototype.
Fig. 4. Block diagram of speed control of the system.
Table 1 Details of experimental prototype.
4. Modeling of New Gripping and Cutting Tool Property Value
Degree of Freedom 1/2
The designed gripping and cutting system was simu- Compliant, parallel
lated and validated in SolidWorks 2011 to observe the Grasp type
grasp, force closure
system behavior with real parameter environment and to Gripper
Maximum opening 150 mm
test the system against different kinematic and dynamic Grasping force 1.603 N
performances. The modeling of the system provides all Grasping surface 80 X 45 mm
the optimal parameters which were used to build the Slicing horizontal
Cutting type
prototype and improve the system performance. Figure cut, force closure
Cutter
5 shows different views of the gripping and cutting Maximum opening 45 mm
system model developed in SolidWorks. Cutting tool Scissor
DC servo, brushless,
Motor
5W
Motor torque 6.57 N-m
Torque
Required torque 1.32 N-m
2.5M 36T 20PA
Plain carbon (2)
10FW 10NSD
Gears
2.5M 18T 20PA
Plain carbon (1)
10FW 10NSD
Average time to
1.10 sec
grasp and cut

The performance of the developed experimental de-

Fig. 5. Gripping and cutting model developed in vice was tested by means of adopting the device to
SolidWorks with different views. grasp and cut the sweet peppers. For the experiments,
four different patterns were selected as;
The gripping and cutting model developed in soft-
ware was tested by means of high quality solid mesh 1. Sweet pepper without leaves
with 4 point Jacobian having 1.06 mm tolerance and 2. Sweet peppers with leaves
total 16626 nodes. After successful meshing of the 3. Partially overlapping sweet peppers
model, it was tested against stress, strain, factor of 4. Overlapping sweet peppers with leaves
safety, displacement of the components in model and
fatigue properties of the model.

- 273 -
The 20th MAGDA Conference in Pacific Asia, November 14-16, 2011, Kaohsiung, Taiwan

These four conditions were used during the experiments

so that the reliability and functionality can be deter-
mined easily. During each grasping and cutting, the time
required to perform the grasping and cutting operation
was recorded. The depth of stem cut from sweet pepper
was supposed to deep so the distance of cutter was
adjusted to 40 mm from the gripper. This distance can
be increased or decreased by adjusting the shaft of cutter
installed on circular disc. To check the practical feasibil-
ity of the tool, different size and shapes of sweet pep-
pers were used for experiments. Fig. 8. Stresses acting on the model.

6. Results and Discussion

6.1 Simulation Results of model
The gripping and cutting model developed in Solid-
works helps to determine the reliability, flexibility and
effectiveness of the system. The model was simulated to
obtain the data for static and buckling study; that means
to check the model for its physical properties and
performance.
Figure 7 represents the component contact and ap-
plied torque acting on the gripping system along with Fig. 9. Strains acting on the model.
center of gravity force acting centrally downward.

Fig. 10. Overall displacement of the model.

Fig. 7. Gripper components with acting torque.

The simulation results for stress on the system were

illustrated in figure 8 and indicates clearly that the
system support the stresses developed internally. This
also means that the model can tolerate the stresses
generated after application of torque and during the
operation. Only some part of gripper bar that connects
to the gears were observed more stressed than other
components. The minimum stress and maximum stress-
Fig. 11. Design insight view of the model.
es were recorded as 1.1749 N/m2 and 1.6821 N/m2
respectively.
The figure 10 shows the overall displacement of the
The simulation results for equivalent strain on the
model in which maximum displacement was found at
system were shown in figure 9 in which it was observed
gripper bars and then at scissor tips. This clearly indi-
that the gripper bar that connects to the gears were
cates that the gripper bars and scissor tips has supple-
slightly exposed to the strain. The minimum strain and
mentary movement in association with the gears. Also
maximum strain were observed as 5.62 and 5.92 respec-
the distinction in the displacement confirms that the
tively. It means overall system has almost inferior
speed variations of the large and small gears that re-
deformation when the external forces applied or during
sponsible for movement of the gripper bars and scissor
the operation.
tips at different speed.

- 274 -
 The 20th MAGDA Conference in Pacific Asia, November 14-16, 2011, Kaohsiung, Taiwan

Figure 12 represents the factor of safety plot for the

model in which the model was found significantly safe
against deformation and buckling effects. It also demon-
strates that the structural capacity of a system beyond
the expected loads was enormously higher.
The percentage of the life of the structure consumed
by the defined fatigue events can be seen in figure 13 in
which the percentage damage was found very less for
model. This property of the model was found signifi-
cantly affected by life cycle and can be seen in figure 14.
The biaxial ratio was found as 1 which clearly indicates
that the life cycle of the model affected moderately by
Fig. 12. Factor of safety plot. the performance of the system. The load factor of safety
for fatigue failure at each location was demonstrated by
figure 15 in which all the system was exposed to the
factor of safety as 3. The load factor less than 1 indi-
cates the failure of the system and in developed model it
was found as 5 which confirms the minimum fatigue
failure of the model.
The simulation results of the model validate the
physical characteristics and properties of the system and
confirm its reliability, flexibility and significant effec-
tiveness of the system.
6.2 Experimental Results
Fig. 13. Fatigue - damage plot. The constructed prototype was tested for the practi-
cal application for grasping and cutting the sweet
peppers. The performance of prototype was found
significantly reliable during the experiments. Figure 16
to 19 represents the performance of the prototype during
the experiments.

Fig. 14. Fatigue – life cycle plot.

Fig. 16. Performance of prototype with pattern 1.

Fig. 15. Fatigue – load factor plot.

A Design Insight plot shows the regions of the mod-
el that carry the loads most efficiently with a continuous
path between the various forces acting on model as
shown in figure 11. The model was found significantly
strong to carry all the forces acting on it without any Fig. 17. Performance of prototype with pattern 2.
modification.

- 275 -
The 20th MAGDA Conference in Pacific Asia, November 14-16, 2011, Kaohsiung, Taiwan

References
[1] Bertetto, A. M., Falchi, C., Pinna, R. and Ricciu, R., "An
integrated device for saffron flowers detaching and har-
vesting", Robotics in Alpe-Adria-Danube Region (RAAD),
2010 IEEE 19th International Workshop on, pp. 93-98,
24-26 June 2010.
[2] Y. Sarig, “Robotics of fruit harvesting: A state-of-the-art
review”, Journal of Agricultural Engineering Research, Vol.
54, pp. 265-280, 1992.
[3] A. K. El-Kalay, M. Akyurt, A. A. N. Aljawi and F. M.
Fig. 18. Performance of prototype with pattern 3. Dehlawi, “On gripping mechanisms for industrial robots”,
The Fourth Saudi Engineering Conference, Vol. IV, pp.
159-170, Nov. 1995.
[4] R. C. Harrell, P. D. Adsit, T. A. Pool and R. Hoffman,
“The Florida Robotic Grove-lab” ASAE Paper 88-1578,
St. Joseph, MI 49085, 1988.
[5] Y. Sarig, Y. Edan, N. Katz and T. Flash, “Some aspects
of robotics for fruit picking”, French-Israel Bi-National
Symposium on Advanced Robotics, Theory and Practice,
Tel-Aviv, May 30- 31, 1988.
[6] S. Hayashi, O. and Sakaue, “Tomato harvesting by
Fig. 19. Performance of prototype with pattern 4. robotic system”, ASAE Annual International Meeting,
Phoenix, Arizona, USA, Paper 963067, 1996.
[7] M. Monta, N. Kondo, and K. C. Ting, “End-Effectors for
In the 1st and 2nd patterns of experiments, the signifi- Tomato Harvesting Robot”, Artificial Intelligence Re-
cant performance of the prototype was observed. The view, Vol. 12, pp. 11-25, 1998.
grasping and cutting actions were smooth and easy. [8] E. J. van Henten, J. Hemming, B. A. J. van Tuijl, J. G.
There were no judgment errors found during the har- Kornet, J. Meuleman, J. Bontsema, and E. A. van Os,
vesting operation and all the samples were harvested “An Autonomous Robot for Harvesting Cucumbers in
successfully. While in the 3rd and 4th patterns, the slight Greenhouses”, Autonomous Robots, Vol. 13-3, pp. 241-
judgment error was observed but still the prototype was 258, 2002.
able to perform harvesting. In the 4th pattern, the obser- [9] S. Horiuchi, J. E. DeVay, J. J. Stapleton and C. L.
Elmore, “Solarization for greenhouse crops in Japan”, In-
vation were recorded that the two overlapped sweet
ternational Conference on Soil Solarization, pp. 16-27
peppers were grasped but only the first one was exposed Amman (Jordan), 19-25 Feb. 1990. from Ch04,
to cut by scissor. Also, even the stem was covered with www.fao.org/docrep/T0455E/T0455E04.htm
leaves; the cutter cut the stem through the leaves. The [10] S. Kitamura and K. Oka, "Recognition and cutting system
performance of the prototype was found satisfactory in of sweet pepper for picking robot in greenhouse horticul-
3rd and 4th pattern. As per the results obtained in simula- ture", Mechatronics and Automation, 2005 IEEE Interna-
tion, none of the system component exposed to defor- tional Conference, Vol. 4, pp. 1807- 1812, 29 July-1 Aug.
mation and fatigue. 2005.

7. Conclusions
The gripping and cutting tool was modeled, simulat-
ed and validated in Solidworks successfully. The per-
formance of the system was observed significantly
reliable and effective. The static and fatigue studies of
the system confirms the consistency and effectiveness
with maximum factor of functionality. Thus, the physi-
cal characteristics and properties of the system were
verified.
The system was constructed and tested successfully
based on the simulation results. The experiments with
four different patterns confirm the performance of the
system. The 1st and 2nd patterns were observed signifi-
cant successful rate of harvesting than the 3rd and 4th
patterns. Finally, the developed gripper and cutting
system tool for sweet pepper harvesting robot hand
confirmed its successful performance. Further, optimi-
zation of the physical parameters and properties with
different material can be considered as further research.

- 276 -
 The 20th MAGDA Conference in Pacific Asia, November 14-16, 2011, Kaohsiung, Taiwan

2A33
Evaluation of Equal Error Rate in Document Authentication System
Using Magnetically Labeled Pattern
Takayuki Yajima*, Syunji Sonoda, Tsutomu Yamada and Yasushi Takemura

Artifact-metric system using magnetic fibers can be applied for authentication system of stock certificate, bill,
passport, plastic cards and other documents. The security of this system is guaranteed by its feature of difficulty in
copy or reproduction. The security documents with randomly printed magnetic pattern based on artifact-metric
system using magnetic fibers is studied. In this study we focused on the equal error rate (EER). It is the error rate at
which false match rate is equal to false non-match rate. According to our evaluation for the accuracy of authentication
of the system, EER is 5.6 × 10-5 % by comparing the waveform calculated from the pattern with the pseudo
waveform obtained by MR sensor in the simulation. It is found that the authentication accuracy of the authentication
system using magnetically labeled pattern can be higher than that of artifact-metric system using magnetic fiber.

Keywords: authentication system, artifact-metric system, magnetic pattern, equal error rate

1. Introduction
It has been reported that the documents with
randomly dispersed magnetic fibers can be used for
authentication classified as artifact-metric system [1].
The advantages of this system can be utilized for the
authentication of stock certificate, bill, passport, plastic （a）Printed image of random magnetic pattern.
cards and other documents by supplying randomly
determined characteristics which are similarly to 0.3
biometrics using finger print, iris and others. The
0.15
security documents with randomly printed magnetic
Output [V]

pattern based on artifact-metric system using magnetic 0

fibers is studied.
-0.15
2. The documents with randomly printed magnetic
pattern -0.3
0 2 4 6 8 10
In the authentication system studied here, the Position [mm]
magnetic pattern consists of randomly dispersed
magnetic dots. Figure 1 shows an example of the （b）Waveform obtained by MR sensor.
document with randomly dispersed magnetic dots and
its waveform. A differential type magnetoresistance Fig. 1 Magnetically labelled pattern.
(MR) sensor is scanned along the magnetic pattern. A
waveform of the detected signal by the MR sensor is
complicated, it is difficult to determine the distribution 3. Evaluation of the authentication system
of the magnetic dots from the waveform. This is a
unique feature of the authenticating system. A degree of In the authentication system, two signal waveforms
similarity between a detected waveform and a waveform are used. A template pattern is defined as an output
of a genuine article is calculated in order to evaluate the waveform recorded from a genuine document. This
authentication accuracy. Different from artifact-metric pattern is used for authentication of documents by
system, the waveform of a genuine article can be comparing the detected waveform from other
calculated from the magnetic dots in our system. documents or the same document. These compared
Registration of the waveform of a genuine article is not waveforms are called sample patterns. At the point of
required, its low cost gives an advantage on our identification, a degree of the similarity between the
authentication system. template pattern and the sample pattern is calculated. A
_______________________ threshold value for the similarity is defined in order to
Correspondence: Yokohama National University, Yokohama, decide if two waveforms were obtained from the same
240-8501, Japan or an equivalent pattern.
*email:d10gd182@ynu.ac.jp

- 277 -
The 20th MAGDA Conference in Pacific Asia, November 14-16, 2011, Kaohsiung, Taiwan

template pattern 1.2

p1 p2

Output [arb. unit]

0.6
pn
0
q2
Output

-0.6
q2
-1.2
p3 qn
0 2 4 6 8 10
q1 Sample pattern Position [mm]

( a ) The calculated waveform.

Position
1.2
Fig. 2 Calculation of similarity.

Output [arb. unit]

0.6

3.1 Calculation of similarity 0

A degree of the similarity between the template -0.6

pattern and the sample pattern is calculated by the
correlation coefficient S. Then, the degree of similarity -1.2
between template (P) and sample pattern (Q) is defined 0 2 4 6 8 10
as S(P,Q), is calculated by the following equation: Position [mm]
n

( p i  p )　 ( qi  q ) ( b ) The pseudo waveform obtained by

S i 1
(1) MR sensor.
n n

( p
i 1
i  p )2  ( q
i 1
i  q )2
Fig. 4 The waveform in the simulation.

Where p and q are mean average of all the elements of

P and Q, respectively. The similarity S is in the range of 4. Analysis of equal error rate
 1  S  1 . The higher value of S is obtained when there The EER using the pseudo waveforms obtained by
is a similar waveform for P and Q. MR sensor as the template pattern is compared with the
3.2 Evaluation by equal error rate EER using the calculated waveform as the template
pattern. Figure 4 shows the calculated waveform and the
In this study we focus on the equal error rate (EER). pseudo waveform. In the simulation, the dimension of
It is the error rate at which false match rate is equal to the document is 7.0 mm wide. Scanning range is 12.8
false non-match rate. False match rate (FMR) is the mm by MR sensor. The density of dot is 60 %. The
probability that the authentication system incorrectly number of documents is hundred thousand copies.
accepts the false documents in an authentication process. Measurement times to each document are 5000 times.
And false non-match rate (FNMR) is the probability that
the authentication system incorrectly rejects the genuine 10-1
document during the authentication process.
10-2
EER = 5.3×10-4 % FMR-pse
Error rate [%]

10-3
Error Rate

FMR FNMR FNMR-pse

10-4 FMR-cal
FNMR-cal
10-5 EER = 5.6×10-5 %
EER
10-6
10-7
0.9 0.95 1
Threshold value [arb. unit]
Threshold value

Fig. 3 Evaluation by equal error rate. Fig. 5 Analysis of equal error rate.

- 278 -
 The 20th MAGDA Conference in Pacific Asia, November 14-16, 2011, Kaohsiung, Taiwan

Figure 5 shows a calculated result of the FMR and

FNMR. The EER using calculated waveform is
5.6×10-5 % and about ten times lower than the EER
using the pseudo measured waveform. In addition, the
EER by comparing the calculated waveform is lower
than the EER of artifact-metric system using magnetic
fibers [1][2].

5. Conclusions
In this paper, the accuracy of document
authentication using magnetically labelled pattern is
evaluated. In the authentication system studied here, the
magnetic pattern consists of randomly dispersed
magnetic dots. The authentication system using
magnetic pattern don’t require registration of the
waveform of a genuine article, its low cost gives an
advantage on our authentication system. It is found that
the accuracy of the authentication system can be higher
than that of the artifact-metric system using magnetic
fibers.

References
[1] H. Matsumoto, I. Takeuchi, H. Hoshino, T. Sugahara
and T. Matsumoto: An artifact-metric system which
utilizes inherent texture, IPSJ Journal, Vol.42, No.7, pp.1,
2001.
[2] T. Ikeda, S. Hiroe, T. Yamada, Y. Takemura and T.
Matsumoto: Document precision analysis of the
authentication system which uses the ferromagnetic fiber,
Proceedings of The 3rd International Conference on Anti-
counterfeiting, Security, and Identification in
Communication (2009ASID), pp. 382-385, Aug. 2009.

- 279 -