Engineering, 2
http://dx.doi.or
g
Copyright © 2
0
Desi
g
ABSTRA
C
The human-c
o
automatic do
c
movement is
r
signal is reco
r
order to extra
c
mand. The d
e
assistant tool
f
Keywords: S
u
1. Introdu
c
N
owadays, P
C
HCI method
i
with a mouse
nient for dis
a
themselves.
T
b
etween the
e
change with t
h
difference ti
m
face electrom
y
movement c
a
sEMG has be
e
engineering.
A
the sEMG h
a
with other bo
d
is easier to
c
strength than
for experime
n
thod need s lo
t
experimenter
s
fortable. The
EMG experi
m
*
This work is su
p
61074113; Scien
t
Chinese Schola
r
Project by Shan
g
12DZ1940903; S
013, 5, 166-17
0
g
/10.4236/eng.
2
0
13 SciRes.
g
n of R
e
and
1
D
e
E
a
2
Ke
y
(East China U
n
C
T
o
mputer inter
a
c
ument contro
l
r
ealized accor
d
r
ded and con
v
c
t the charact
e
e
veloped hum
a
f
or disabled
pe
u
rface Electro
m
c
tion
C
is widely
u
i
s usually ac
h
or a touch sc
r
a
bled people
T
he research
i
e
lectrodes stuc
k
h
e movement
m
e and differe
n
y
ogram (sEM
a
n be detecte
d
e
n widely use
d
A
s a member
o
a
s its unique
d
y signals in
a
c
ollect. Besid
other signals
a
n
ters to use. T
t
s of complex
e
s
which alway
individuals c
a
m
ent always c
o
p
ported by Nation
t
ific Research F
o
r
s, State Educati
g
hai Municipal S
c
hanghai Leading
A
0
2
013.510B036
P
e
al-Ti
m
Surfac
e
Zhen
W
e
partment of A
u
a
st China Unive
r
y
Laboratory of
A
n
iversity of Sc i
e
a
ction (HCI) i
l
technique w
h
d
ing to the su
r
v
eyed to a P
C
e
ristics of sE
M
a
n-computer i
n
e
rson.
m
yography;
H
u
sed everywh
e
h
ieved with a
r
een. But it is
to use these
i
ndicated that
,
k
to the skin
o
of muscle. T
h
n
ce movemen
t
G). All the si
g
d
and recorde
d
d
in the field
o
o
f the family
o
advantages
w
a
pplication. Fi
r
es that, sEM
G
a
nd sEMG ha
s
he normal sE
M
e
lectrodes lin
k
s make the us
a
n’t move for
l
o
nsumes muc
h
Nature Science
F
o
undation for the
on Ministry; M
e
c
ience and Tech
n
A
cademic Discip
l
P
ublished Onlin
e
m
e Docu
e
Elect
r
W
ang
1,2
, Be
i
u
tomation, Sch
o
r
sity of Science
A
dvanced Cont
r
e
nce and Techn
o
w
Rece
i
s
now playin
g
h
ich is based
o
r
face electrom
y
C
terminal by
u
M
G, recognize
n
teraction tec
h
H
uman-Comp
u
e
re. The nor
m
keyboard alo
n
really inconv
equipmen
t
s
b
,
the differen
c
o
f muscle wou
l
h
e differences
t
s make the s
u
g
nals of musc
d
[1]. Recentl
y
o
f rehabilitati
o
o
f body signa
l
w
hen compar
e
r
st of all, sE
M
G
has strong
s
more chann
e
M
G collect m
k
ed with PC a
n
ers feel unco
m
l
ong time as t
h
h
time, and t
h
F
oundation of Chi
Returned Overs
e
e
dical Cooperati
o
n
ology Commissi
o
l
ine Project B504
.
e
October 2013
ment
C
r
omyo
g
i
Wang
1,2
, X
i
o
ol of Informati
o
and Technolog
y
r
ol and Optimi
z
o
logy), Ministr
y
w
d
i
ve
d
May 2013
g
a great role
i
o
n the human
h
y
ograph y (sE
M
u
sing wireles
s
the waving
m
h
nique can be
u
ter Interactio
n
m
al
n
g
e-
b
y
c
e
l
d
at
ur
-
c
le
y,
o
n
l
s,
e
d
M
G
er
e
ls
e-
n
d
m
-
h
e
h
at
usuall
y
This
sEMG
link li
n
are col
rearm.
individ
u
p
roces
s
The st
a
wirele
s
conve
y
nated
b
p
otenti
a
compo
n
the im
p
vented
dispos
e
The sy
p
ractic
a
2. Da
t
The su
b
gradua
t
age ag
hand a
s
are tot
a
menter
na
e
as
o
n
o
n
.
(http://www.sc
i
C
ontrol
g
raphy
(
i
ngyu Wang
1
o
n Scie nce and
E
y
, Shanghai, 20
0
z
ation for Chem
i
y
of Education,
S
d
u.cn
i
n computer t
e
h
and waving
m
M
G). A collec
s
Zigbee. An
a
m
ovements, an
d
used as a ne
w
n
; Zigbee; Doc
u
y
makes them
v
paper propo
to PC wirele
s
n
es across res
e
lected by the
The features
o
u
al’s hand w
s
ed first by a
H
a
r topology o
f
s
s transmit net
w
y
ed to PC rap
i
b
y the ECG, i
t
a
lly yields
m
n
ent analysis
p
act of ECG.
autonomousl
y
e
the data and
stem finally
c
a
l use.
t
a Acquisit
i
b
jects of sEM
G
t
ed students (
f
e is twenty-t
h
s
the strong h
a
a
l four series
o
’s forearm. T
h
rp.org/journal/
e
Based
o
(
sEM
G
1
,2
E
ngineering,
0
237, China
i
cal Processes,
S
hanghai, 2002
3
e
chnology. T
h
m
ovements. T
h
tor is s et on t
h
a
utomatic alg
o
d
transmit to
d
w
gallery for t
u
ment Contro
l
v
ery tired.
ses a new
m
s
sly, so there
e
archers and
u
instruments
s
o
f sEMG will
ave up or d
o
H
PT in order t
o
f
Zigbee was
w
o
r
k. The sE
M
i
dly. Since E
M
t
usually ham
p
m
isinterpretati
o
method (ICA
)
A real-time
y
which can r
e
transmit the
d
c
omes true wi
t
i
on
G
pick up exp
e
f
ive male, fiv
e
h
ree. Eight st
u
a
nd while two
o
f sEMG data
h
e sEMG sig
n
e
ng)
o
n Zig
b
G
)
*
3
7, China
h
is study intro
d
h
e recognitio
n
h
e forearm. T
h
o
rithm is dev
e
d
ocument cont
r
eaching, as w
l
m
ethod to co
n
are no misc
e
u
sers. And all
s
et on individ
u
be detected
w
o
wn. The dat
a
o
eliminate th
e
settled to b
u
M
G signals c
a
M
G is often
c
p
ers data ana
l
o
ns. The ind
e
)
is used to g
control syste
m
e
ceive the sE
M
d
ata to contro
l
t
h stable, sec
u
e
riment are te
n
e
female) wh
o
u
dents use t
h
use the left o
n
obtained fro
m
n
als were rec
ENG
b
ee
d
uces an
n
of hand
h
e sEMG
e
loped in
r
ol co
m
-
ell as an
n
vey the
e
llaneous
the data
u
al’s fo-
w
hen the
a
is pre-
e
clutters.
u
ild up a
a
n easily
c
ontami-
l
ysis and
e
pendent
et rid of
m
is in-
M
G data,
l
system.
u
rity and
n
healthy
o
se ave
r
-
h
eir right
n
e. Th ere
m
experi-
orded at

Z. WANG ET AL. 167
extensor carpi radialis muscle, flexor carpi radialis and
extensor carpi ulnaris musculus according to anatomy.
The actual electro position is shown in Figure 1.
The experiment uses an amplifier with a sampling fre-
quency of 500 Hz, a sensitiv ity of 100 uV. The high-pass
filter (HPF) frequency is 100 Hz while the notch filter is
50 Hz.
3. Wireless Zigbee
The Zigbee technology is famous for its advantage on
low power consumption, concise, short distance and so
on. Zigbee is becoming popular in our daily life. It has a
particular advantage in the field of short distance and low
consumption equipment transforms digital data wireless,
especially during periodic and intermittent applications.
It can work at the frequency of 2.4 Ghz (world wild
scope), 868 Mhz (Europe scope) and 919 Mhz (America
scope) allocate with the speed of 250 Kbit/s, 20 Kbit/s
and 40 Kbit/s. The transport distance can range from 10
to 75 meters.
In this study, the Zigbee module works at the fre-
quency of 2.4 GHz and the channel sets to 20 through
MAC_RADIO_SET_CHANNEL(x) control command.
The data will be transited in a periodicity method. Once
the receiver Zigbee get into receiving status, the net-layer
can receive sEMG signals through MAC service. Before
the application layer receive the data, the hardware part
has accomplished receiving the data and storing them in
the buffer which the software can read data from buffer
to achieve corresponding function [2]. The wireless
transaction module performs its three functions in this
system. First, it takes charge of the transformation be-
tween SEMG and PC while making sure that the infor-
mation is real-timing. Second, primary dispose sEMG
signal is collected from forearm. Including amplify the
feeble signal and filter for the first time. At last, trans-
form sEMG signal to PC and store them in the buffer for
display online.
4. System Structure
The flow char of the system is shown in Figure 2. First
of all, detect the sEMG plus through the electrodes fixed
in forearm. Then pre-process the sEMG signal such as
filtering, A/D transforming and amplifying with an am-
Figure 1. The actual electro position.
Figure 2. The flow chart of the system.
plifier. Third, transfor m them to PC with wireless Zigbee
part, store them in the buffer memory and display them
in an invented window. Fou rth, deal with the digital data
collected from forearm. An independent component
analysis (ICA) arithmetic is put into use to digital filter-
ing preliminary. Fifth, extract the hand wave movement,
take clear of the exact movements and the number of
them, output them as control sign al. At last, stop and exit
the system.
4.1. Independent Component Analysis
EMG recordings are often incorporated by the electro-
cardiogram (ECG), which can disturb the classifications
of hand movement and result in misinterpretations [3].
Independent component analysis (ICA) is widely used
for a situation involving two signal sources [4]. The dia-
gram to the operation of ICA is illustrated in Figure 3.
The mixtures X are gene rat ed by the operati on
ASX

(1)
In this case,
1
2
s
S
s






, 1 2
x
X
x



And the mixing matrix A is given by
11 12
21 22
aa
Aaa







The aim is to estimate an unmixing matrix W that ena-
ble the signal sources U to be ob t ained by
UWX

(2)
where
11 12
21 22
ww
Www







, 1
2
u
Uu



The ICA algorithm is performed by each iteration, the
unmixing matrix W is updated until convergence is
achieved. The algorithm stops training when the rate of
change falls below a predefined small value.
The description of ICA can be referred to the book by
LEE [5].
4.2. Hand Wave and Number of Clench Fist
There will be a pulse when the individual wave their

Z. WANG ET AL. 168
Figure 3. Diagram to illustrate the operation of ICA.
hands while the potential difference is much lower when
there are no wave movements. A threshold will be de-
fined to distinguish the hand waving movements. The
data will set to zero when they are lower than the thre-
shold while set to one when they are higher than the
threshold. Then the hand wave movement and number of
clench fist could be calculated exactly [6].
4.3. Control Flow Chart
The system normally operate flow the process which
consist of six steps such as entering the system, selecting
file list, confirming to play, documenting page up, do-
cumenting page down and exiting or replaying the sys-
tem. The detailed control flow is shown in Figure 4,
1) Enter the system: The experimenter can clench their
fist three times to enter the system when he or she enters
for the first time or after stop. The other clench has no
means.
2) Select file list: After enter the system, the individual
could wave his or her hand from up to down or reverse to
choose which file to play. Citing wave down three times
means choose the third item file to play. Of course, the
file list could be updated manually.
3) Confirm to play: The individual could clench fist
twice to confirm the chosen file to play.
4) Document page up: When the performer needs to
control the document to page up he or she could wave his
or her hand from right to left.
5) Document page down: When the performer needs to
control the document to page down he or she could wave
his or her hand from left to right.
6) Exit or replay: The individual could replay the
whole system by clench fist twice or exit the system by
disconnect the wireless Zigbee part.
5. Result
5.1. Data Analysis
There are totally four series of sEMG data which indicate
wave up, wave down, wave right, wave left and clench
fist movement. According to anatomy, extensor carpi
radialis muscle, flexor carpi radialis muscle and extensor
carpi ulnaris muscle can generate electronic signals when
individuals wave up or wave down and sEMG signal
may change when they rotate their arm to wave right and
Figure 4. The detailed control flow chart.
wave left. It is also indicated that the sEMG signals are
much stronger when individuals clench their fist than just
wave their hand [7]. Depend on the above information,
the correspond sEMG waveform is shown in Figures 5-
8 as follows.
In Figure 5, the sEMG data is normally under 0.3
when there is no wave movement produced. However,
the sEMG data suddenly arrive to 0.5 - 1.5 when the in-
dividuals move their hand. According to the threshold we
set, we can distinguish the data to zero and one for con-
venience of the output of control command. As in figure5,
we can get that there are totally four times of wave up
movement or wave right movement happened. Along
with the rotate arm signals shown in Figure 7, there is
rotate arm movement happened on the same time which
indicate that the actual wave movement is wave right.
The sEMG data is entirely rolling-over when individuals
wave their hand down rather than wave up. The same as
wave right and wave left. In Figure 8, we can see that
the clench fist sEMG signal actually up to 2.0 - 2.5 which
higher than other signals. So we can easily distinguish
clench signals and wave hand signals.
5.2. Real-Time Control System
After extract the wave hand movement and the number
of clench fist, the output signal can be used in the docu-

Z. WANG ET AL. 169
Figure 5. Wave up sEMG signals threshold analysis.
Figure 6. Wave down sEMG signals threshold analysis.
Figure 7. Rotate arm sEMG signals threshold analysis.
Figure 8. Clench fist sEMG signals threshold analysis.
ment control system. First, we could choose the file to
play with wave hand. Second, after select the file we
could control a document to display, such as page up and
page down or close the document. The software is de-
signed and executed successful on VB6.0 [8]. The
MSCOM controller is used to connect with serial data
from Zigbee. The HCI system is working as shown in
Figure 9 [9].
6. Result
The gap of convenience is very common between differ-
ent HCI methods. In order to get more information about
the proposed HCI, a survey is made in different groups.
Totally twenty students whose average age is 24 are
asked to make this survey. The survey information is
listed as follow. Take use of an online survey system, a
valuable report is shown in table1 detailed.
Question 1:
Which is the most comfortable HCI pattern?
A, mouse down; B, sEMG; C, voice
Question 2:
Which action will you tak e to control a PPT?
A, clench fist; B, wave hand; C, shake figure
Question 3:
Which hand waving actio ns will you chose to generate
useful signal?
a, left to right; b, right to left; c, front to back
As the result shown in Ta b l e 1, about 60% people tend
to take use of sEMG as control signal; 50% people fa-
miliar with wave hand to achieve document control; 60%
people expect to enforce document by wave their hand
Figure 9. Document control system.
Table 1. Survey result report.
question
option
Survey result
No.1 No. 2 No. 3
A 5 6 2
B 12 10 12
C 3 4 6

Z. WANG ET AL. 170
from right to left.
According to this survey result, it is convenient for
many people to select hand wave movement as document
control signal.
7. Conclusions
In this paper, we propose a new HCI system which has
many innovations. First of all, the control command is
produced through the analysis of sEMG signals. Besides
that, the wireless data transition system along with a data
transit network built by Zigbee. EOG signals are elimi-
nated from sEMG with th e ICA algorith m which enh ances
the accuracy of control command. A computer platform
is built to deal with the wireless data transit module, the
recognit ion of sEMG, th e output of c ontrol command and
so on.
This paper proposes a new application of sEMG along
with Zigbee technology. An online surv ey system is used
to make sure that the most acceptable movement of
document control is wave hand. In addition, the direct
body movement in a speech with PPT is very natural and
comfortable because body movement is the instinct of
oneself. The ICA arithmetic can combine with other
arithmetic such as JADE and PCA in order to get more
instinct control sign al. In the future, the recognition algo-
rithm can be improved with the development of technol-
ogy. This system can be used in many aspects of the so-
ciety such as teaching and disabled treatment.
REFERENCES
[1] K. Ando, K. Nagata, D. Kitagawa, N. Shibata, M. Ya-
mada and K. Magatani, “Development of the Input Equip-
ment for a Computer Using Surface EMG,” 28th Annual
International Conference of the IEEE Engineering in Me-
dicine an d Biolog y, New York City, 2006, pp. 1331-1334.
[2] X. Chen and Z. Jane Wang, “Pattern Recognition of Num-
ber Gestures Based on a Wireless Surfa ce EMG System,”
Biomedical Signal Processing and Control, Vol. 8, 2013,
pp. 184-192.
http://dx.doi.org/10.1016/j.bspc.2012.08.005
[3] G. R Naik, D. K Kumar, S. P Arjunan, H. Weghorn and
M. Palaniswami, “Limitations and Applications of ICA in
Facial sEMG and Hand Gesture sEMG for Human Com-
puter Interaction. Digital Image Computing Techniques
and Applications,” 9th Biennial Conference of the Aus-
tralian Pattern Recognition Society, Vol. 58, 2007, pp.
15-22.
[4] L. Vigon, M. R. Saatchi, J. E. W. Mayhew and R. Fer-
nandes, “Quantitative Evaluation of Techniques for Ocu-
lar Artefact Filtering of EEG Waveforms,” IEE Proceed-
ings of Science, Measurement and Technology, Vol. 147,
No. 5, 2000, pp. 219-228.
http://dx.doi.org/10.1049/ip-smt:20000475
[5] T. W. Lee, “Independent Component Analysis, Theory
and Application,” Kluwer Academic Publishers, Boston,
1998.
[6] N. W. Willigenburg, A. Daffertshofer, I. Kingma and J. H.
Dieen, “Removing ECG Contamination from EMG Re-
cordings: A Comparison of ICA-Based and Other Filter-
ing Procedures,” Journal of Electromyography and Kine-
siology, Vol. 22, 2012, pp. 485-493.
http://dx.doi.org/10.1016/j.jelekin.2012.01.001
[7] K. Kiguchi and Y. Hayashi, “An EMG-Based Control for
an Upper-Limb Power-Assist Exoskeleton Robot,” IEEE
Transactions on System, Man, and Cybernetics-Part B:
Cybernetics, Vol. 42, No. 4, 2012, pp. 1064-1071.
[8] J. U. Chu, I. Monn and M. S. Mun, “A Real-Time EMG
Pattern Recognization System Based on Linear-Nonlinear
Feature Projection for a Multifunction Myoelectric Hand,”
IEEE Transactions on Biomedical Engineering, Vol. 53,
No. 11, 2006, pp. 2232-2239.
http://dx.doi.org/10.1109/TBME.2006.883695
[9] Y. Yu, C. Xiang, T. Youqiang, Z. Xu and Y. jihai, “Hu-
man-Machine Interaction System Based on Surface EMG
Signals,” Journal of System Simulation, Vol. 22, No. 3,
2010, pp. 651-655.