Powerpoint Te mplates Page 1 <<Recognition of basic hand movements using Electromyography>> Ilaria Pacifico Bionics Engineering University of Pisa Data Mining project:
8/15/2019 Recognition of Basic Hand Movements Using Electromyography
http://slidepdf.com/reader/full/recognition-of-basic-hand-movements-using-electromyography 1/13
8/15/2019 Recognition of Basic Hand Movements Using Electromyography
http://slidepdf.com/reader/full/recognition-of-basic-hand-movements-using-electromyography 2/13
Powerpoint Templates
Page 2
Introduction
Application using EMG signalThe use of the EMG signal is quite prevalent in a wide range of applications
!ne of the practical applications of the EMG signal is the use for control
e"os#eleton limps of $od%& such as hand It is more comforta$le for a hand
amputee to wear a glove that includes the EMG electrodes than using the
rather promising Electroencephalograph% 'EEG( electrodes in the area of thehead This issue is of significant importance for using such a s%stem in a dail%
$asis
The hand controlled from a single& highl% intelligent control s%stem that
uses two sensors to open and close the hand
8/15/2019 Recognition of Basic Hand Movements Using Electromyography
http://slidepdf.com/reader/full/recognition-of-basic-hand-movements-using-electromyography 3/13
Powerpoint Templates
Page 3
E"perimental Procedure
AI: to identif% si" $asic movements of the hand!pherical ')(:
holding an elastic $all
"ylindrical '*(:
holding a glass of water#ip 'T(:
the touch of the thum$ to the inde"
Palmar 'P(:
holding a pen
$ateral '+(:
holding a credit card
%oo& ',(:
holding the handle of -+ water $ottle
EA!'REE(#!:Five health% su$jects 'two males and three females( of the same age appro"imatel% './
to ..0%ear0old( were as#ed to repeat the si" $asic movements 3) times* using t+o
sensors for acquiring the data
)ampling rate: 1// ,2& 3utterworth 3and Pass filter 4101// ,2& 5otch 6ilter 1/ ,2
The total time for each measurement was 5 seconds
8/15/2019 Recognition of Basic Hand Movements Using Electromyography
http://slidepdf.com/reader/full/recognition-of-basic-hand-movements-using-electromyography 4/13
Powerpoint Templates
Page ,
E-. Empirical ode -ecomposition
•
EMD provides a novel adaptive method for anal%2ing non0linearand7or nonstationar% signals It decompose the signal into a num$er
of IM6s
8/15/2019 Recognition of Basic Hand Movements Using Electromyography
http://slidepdf.com/reader/full/recognition-of-basic-hand-movements-using-electromyography 5/13
Powerpoint Templates
Page /
6eatures E"tractionMEA58ch4
MEA58ch.
mode8ch4
mode8ch.
var8ch4 '9ariance(
var8ch.
2erocross8ch4
2erocross8ch.
))*8ch4 ')lope )ign *hanges(
))*8ch.
+8ch4 'aveform +ength(
+8ch.
AMP8ch4'illison Amplitude(
AMP8ch.
s#ew8ch4
s#ew8ch.
#urtosis8ch4
#urtosis8ch.
MA948ch4 'Mean A$solute value4(
MA948ch.
MA9.8ch4 'e"tension of MA94(MA9.8ch.
))I8ch4 ')imple )quare Integral(
))I8ch.
MD68ch4 'Median 6requenc%(
MD68ch.
;M)8ch4 ';oot Mean square(;M)8ch.
A;98ch4 'Average ;ectified 9alue(
A;98ch.
8/15/2019 Recognition of Basic Hand Movements Using Electromyography
http://slidepdf.com/reader/full/recognition-of-basic-hand-movements-using-electromyography 6/13
Powerpoint Templates
Page 0
4< E"perimental )etup5o filters0 1*90 Total Data)et
! Rip , Ib0 Adaboost Rand4orest
Accurac% ==11> ?@> .=> 1=> @444> @.11>
I$B: for several values of #: *onfusion matri" ;and6orest:
8/15/2019 Recognition of Basic Hand Movements Using Electromyography
http://slidepdf.com/reader/full/recognition-of-basic-hand-movements-using-electromyography 7/13
Powerpoint Templates Page 5
C-: )everal +evels of pruning
*onfidence 6actor/4 *onfidence 6actor4
;elativel% high level of accurac% in
correspondence of *6/.
'standard value in e#a(
Moreover& an higher value of *6
causes an increase of
computational time
8/15/2019 Recognition of Basic Hand Movements Using Electromyography
http://slidepdf.com/reader/full/recognition-of-basic-hand-movements-using-electromyography 8/13
Powerpoint Templates Page
.< E"perimental )etup5o filters0 Manual 1*9:4 6old per )u$ject
Ib0 Random 4orest
Accurac% ..--> .@>
ale1 4---> 111>
ale2 > 4=44>4emale1 .==?> ==?>
4emale2 -/> ?..>
4emale3 ....> 1>
9er% low levels of accurac%F
8/15/2019 Recognition of Basic Hand Movements Using Electromyography
http://slidepdf.com/reader/full/recognition-of-basic-hand-movements-using-electromyography 9/13
Powerpoint Templates Page 6
< E"perimental )etup
5o filters0 4.*9 for each )u$ject/ trials for each movement for each su$ject '= movements(: 4. folds
This time the $est classifier
was found to $e I$= '$etter
than ;andom 6orest(
Paper results:
8/15/2019 Recognition of Basic Hand Movements Using Electromyography
http://slidepdf.com/reader/full/recognition-of-basic-hand-movements-using-electromyography 10/13
Powerpoint Templates Page 1)
Attri$ute )election: 4/ most significant 6eatures:
"fs!ubsetEval Info7ain P"A
;esults:
Attribute!elector
"lassification Results
I80 Rand4orest rip , ! Adaboost
P*A 80.67% 88.11% 62.11% 65.44% 57.33% 76.55%
*fs
88.56% 90.78% 79.44% 79.44% 64.66% 89.89%
InfoGain 87.66% 91.66% 80.77% 84.57% 62.11% 92%
8/15/2019 Recognition of Basic Hand Movements Using Electromyography
http://slidepdf.com/reader/full/recognition-of-basic-hand-movements-using-electromyography 11/13
Powerpoint Templates Page 11
*onclusions Attri$ute )election vs 5o Attri$ute selection
8/15/2019 Recognition of Basic Hand Movements Using Electromyography
http://slidepdf.com/reader/full/recognition-of-basic-hand-movements-using-electromyography 12/13
Powerpoint Templates Page 12
)catter Plot
8/15/2019 Recognition of Basic Hand Movements Using Electromyography
http://slidepdf.com/reader/full/recognition-of-basic-hand-movements-using-electromyography 13/13
Powerpoint Templates Page 13