0% found this document useful (0 votes)
78 views6 pages

Artificial Vision Based Computer Interface

This document describes an artificial vision-based computer interface to help people with motor disabilities use computers. The system uses a web camera to recognize patterns of movement in the user's face. It analyzes the images in real-time to emulate mouse functions like clicking and dragging. Specifically, it detects the user's face, locates a colored pattern on the forehead to track head movements, and uses mouth opening and closing to click. The application was developed in C++ to work on Windows platforms and has been successfully tested by different users.

Uploaded by

ASESPZO
Copyright
© © All Rights Reserved
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
Download as doc, pdf, or txt
0% found this document useful (0 votes)
78 views6 pages

Artificial Vision Based Computer Interface

This document describes an artificial vision-based computer interface to help people with motor disabilities use computers. The system uses a web camera to recognize patterns of movement in the user's face. It analyzes the images in real-time to emulate mouse functions like clicking and dragging. Specifically, it detects the user's face, locates a colored pattern on the forehead to track head movements, and uses mouth opening and closing to click. The application was developed in C++ to work on Windows platforms and has been successfully tested by different users.

Uploaded by

ASESPZO
Copyright
© © All Rights Reserved
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
Download as doc, pdf, or txt
Download as doc, pdf, or txt
You are on page 1/ 6

An Artificial Vision Based Computer Interface

Miguel Mrquez
1
, Erika Chacn
2
, ngelica Ramrez
3
, ngel Gil
4
Universidad Nacional Experimental del Tchira (UNET). Decanato de Investigacin. Laboratorio de
rototipos. !an "ristbal # $ene%&ela. ''() # *+,-.(.'/** Ext. /(+.
0
mmar1&e%2&net.ed&.ve3
*
eri4a.chacon2&net.ed&.ve3
.
angelica.ramire%2&net.ed&.ve3
/
agil2&net.ed&.ve
Abst ract - This research resen!s an ar!i"icial #isi$n % &ase' c$mu!er in!er"ace, (hich main $&)ec!i#e is
su$r!ing han'icae' e$le (i!h m$!$r 'isa&ili!* in !he use $" ers$nal c$mu!ers+ The s*s!em "acili!a!es !he
use $" s$"!(are alica!i$ns !$ in'i#i'uals !ha! ha#e null $r limi!e' m$&ili!* $" !heir uer lim&s+ ,n $r'er !$
acc$mlish !he $&)ec!i#es, !he s*s!em c$un!s $n a -e& camera !ha! all$(s rec$gnizing !he a!!erns $" m$#emen!
$" !he su&)ec!.s "ace, (hich is l$ca!e' in "r$n! $" !he camera &* means $" real !ime image r$cessing+ The
alica!i$n carries $u! an anal*sis $" !he images "$r !he e/ecu!i$n $" !he &asic "unc!i$ns $" !he m$use such as le"!
&u!!$n clicking an'0$r 'ragging $&)ec!s in !he c$mu!er screen+ The "ace 'e!ec!i$n r$cess is carrie' $u! !hr$ugh
!he imlemen!a!i$n $" an alg$ri!hm !ha! uses a casca'e s$r! ke* $" 1aar%2ike !*e+ The in#es!iga!i$n is &ase' $n
!he resen! kn$(le'ge $" ar!i"icial #isi$n, su&%"iel' $" !he ar!i"icial in!elligence (hich $""ers !$$ls $" anal*sis an'
33 image r$cessing in $r'er !$ in!erre! !he ca!ure' scenes+ The alica!i$n (as 'e#el$e' in C44 !$ &e use'
$n -in'$(s 56 la!"$rm+ Tr*$u! $" !he alica!i$n has &een er"$rme' &* 'i""eren! users, sh$(ing e/cellen!
acce!a!i$n an' sh$r! !ime !raining requiremen!s+
1. Introduction
7$(a'a*s !he ers$nal c$mu!ers ha#e &ec$me a !$$l $" 'ail* use in 'i""eren! ac!i#i!ies $" !he s$cie!*8 such !he*
ha#e 'i#erse 'e#ices !ha! all$( !he in!erac!i$n (i!h !he users, am$ng !hem aears !he eriheral use' "$r 'a!a
inu!, &eing !he ke*&$ar' an' !he m$use !he m$s! $ular+ The in!erac!i$n (i!h !hese eriheral requires $" a
minimum h*sical user caa&ili!* !$ $&!ain an e""icien! use8 ne#er!heless a 19: $" !he ($rl'%(i'e $ula!i$n
resen!0'isla* a! leas! s$me kin' $" m$!$r 'isa&ili!* ;1<, !his makes 'i""icul! !he in!erac!i$n &e!(een !hem an'
'e#ices such like !he m$use, hence !he use $" !he c$mu!ers8 i! is "$r !ha! reas$n !ha! arises !he necessi!* !$
'e#el$ing al!erna!i#e means !ha! "acili!a!e !he in!erac!i$n (i!h c$mu!ers+ =r!i"icial #isi$n $""ers !he necessar*
!$$ls "$r imlemen!ing $ne $" !hese al!erna!i#es &* means $" anal*zing ca!ure' images using a camera -e&+
Thr$ugh !he anal*sis $" !he user "ace ges!ures in "r$n! $" !he camera !he ac!i$n $" !he &asic $era!i$ns $" !he
m$use (ill &e emula!e' &* !he s*s!em+
2. General Description
2.1 General diagram of the System
>igure 1 resen!s a 'iagram sh$(ing !he general $era!i$n 'escri!i$n $" !he alica!i$n, c$nsi'ering !he main
r$cesses $" !he 'e#el$e' alica!i$n+
The alica!i$n uses a -e& camera !$ ca!ure !he scene l$ca!e' in "r$n! $" !he c$mu!er, (here, in "irs! ins!ance,
i! searches "$r a "ace+ ?nce a "ace is l$ca!e' in !he scene, !hen i! searches "$r a c$l$r a!!ern, !ha! mus! &e l$ca!e'
in !he "ace i'en!i"ie' in !he scene+ This research ($rk uses a green rec!angular a!!ern l$ca!e' in !he "$rehea' $"
!he in'i#i'ual+ This a!!ern (ill ser#e as !arge! !$ &e "$ll$(e' &* !he image r$cessing m$'ule $" !he s*s!em an'
all$(s !he user !$ l$ca!e !he curs$r in !he screen &* his0her hea' m$#emen!s+ =n$!her &asic "unc!i$n !$ emula!e
is !he click $" !he m$use, !he alica!i$n c$ns!an!l* re#ie(s !he s!a!e $" !he m$u!h $" !he user, !ha! is !$ sa*,
(he!her i! is $en $r cl$se'+ The s*s!em uses !he "requenc* (i!h (hich !he user $ens an' cl$ses !he m$u!h !$
emula!e !he ressing an' releasing $" !he le"! ush &u!!$n $" !he m$use+ ," !he $en m$u!h image "ul"ills cer!ain
am$un! $" h$!$gram is assume' !ha! !he user (ishes !$ main!ain resse' !he le"! ush &u!!$n, !$ c$n'uc! s$me
ac!i$n such as !$ 'rag a "$l'er+

>igure 1+ General 'iagram $" !he alica!i$n
2.2 Description of the Application
sed !ools" The E/!reme 6r$gramming $r 56 is use', (hich $""ers a se! $" !echniques !ha! c$n"$rms a simle
me!h$'$l$g* "$r s$"!(are 'e#el$men!
;2<
, als$ C44 2anguage (as use' al$ng (i!h !he li&raries $" $en c$'e "$r
!he !rea!men! $" ?enC@ images A?en B$urce C$mu!er @isi$nC
;3<
an' ,62 $" ,n!elD, &$!h $" !hem inclu'ing a
series $" #er* use"ul "unc!i$ns "$r !he 'e#el$men! $" !his kin' $" alica!i$ns+ The r$cesses $" !he alica!i$n
(ere m$'ele' using !he Eni"ie' M$'ele' 2anguage AEM2C+
ser #ace Detection" The alica!i$n c$ns!an!l* ca!ures !he images $" !he scene !ha! is in "r$n! $" !he
c$mu!er screen an' in each images i! search "$r a "ace+ This r$cess $" "ace 'e!ec!i$n is ma'e !hr$ugh !he
imlemen!a!i$n $" an alg$ri!hm !ha! uses a casca'e s$r! ke* $" 1aar%2ike !*e
;4<
, !his s$r! ke* l$$ks "$r !he
charac!eris!ics $" !he "ace in a h$riz$n!al (a*, reas$n (h* i!s single use '$es n$! guaran!ee !he 'e!ec!i$n $" !he
user "ace, 'ue !$ !he "ac! !ha! !he "ace can &e incline'+ There"$re, i" !he alg$ri!hm '$es n$! 'e!ec! a "ace in "irs!
ins!ance, au!$ma!icall* r$!a!es !he image 19F 'egrees cl$ck(ise an' search !he ne( image, un!il c$mle!ing G9F+
," !he "ace is n$! e#en 'e!ec!e', is c$me !$ !ake !he $riginal image, an' alies !he same r$ce'ure, &u! !his !ime
c$un!er cl$ck(ise+
The alg$ri!hm is a&le !$ 'e!ec! m$re !han $ne "ace8 reas$n (h* i! is rec$mmen'e' !ha! (i!hin !he #ie(ing angle
$" !he camera $nl* remains !he alica!i$n user+ ?nce !he "ace is 'e!ec!e', a circular &i! ma !rimme' image $"
!he z$ne (here !he "ace (as 'e!ec!e' is s!$re' "$r la!er anal*ses A>igure 2C+
>igure 2+ >ace 3e!ec!i$n
Detection of the pattern to $e follo%ed" ="!er !he r$ce'ure $" 'e!ec!i$n $" !he "ace is "inishe', !he s*s!em
searches in !he "ace image s!$re' !$ i'en!i"* !he c$l$re' a!!ern !ha! (ill ser#e as !arge! !$ m$#e !he curs$r
!hr$ugh !he c$mu!er screen+ = green rec!angular a!!ern is use', (hich is rec$mmen'e' !$ &e l$ca!e' in !he
user.s "$rehea'+ The green c$l$r (as selec!e' 'ue !$ i!s li!!le resence (i!hin !he range $" c$l$rs !ha! can &e
usuall* "$un' in !he human "ace A>igure 3C+
=n alg$ri!hm !$ anal*ze !he G c$m$nen! in !he RGH sace $" c$l$r (as 'e#el$e' !$ carr* $u! !he 'e!ec!i$n $"
!his a!!ern+ The alg$ri!hm cr$sses !he image i/el &* i/el e/!rac!ing $" each $ne i!s c$m$nen!s RGH (i!h !he
in!en!i$n $" #eri"*ing i" !he* "ul"ill !he "$ll$(ing c$n'i!i$nsI G J R, G J H, AG K RC L 29 an' AG K HC L 29+
,n case a i/el sa!is"ies !he men!i$ne' c$n'i!i$ns, !hen !ha! i/el is c$nsi'ere' as ar! $" !he !arge!, !his (a* i! is
e/!rac!e' all i/els &el$nging !$ !he rec!angular a!!ern, an' !hen !$ $&!ain i!s sace c$$r'ina!es A>igure 4C+

>igure 3+ 6a!!ern !$ "$ll$( >igure 4+ E/!rac!i$n $" !he a!!ern

7e/! s!e is !$ i'en!i"* !he c$$r'ina!es $" !he cen!ral $in! $" !he a!!ern, (hich (ill &e use' !$ !ransla!e !he
$si!i$n $" !he image !$ an* $si!i$n in !he c$mu!er screen+ ," !he in'i#i'ual.s "ace is r$!a!e' an angle
'uring !he r$cess $" "ace i'en!i"ica!i$n, !hen a r$!a!i$n a')us!men! mus! &e '$ne as "$ll$(s+ =ssuming !ha! !he
ini!ial c$$r'ina!es $" !he a!!ern cen!ral $in! are 6A51, M1C, an' !he r$!a!i$n angle $" !he $riginal image,
!hen !he $riginal $si!i$n $" !he cen!ral $in! is gi#en &*I
C c$sA N 1 r 5 =
A1C
C A N 1 sen r 6 =
A2C
&o'ement of the Cursor" The m$#emen! $" !he curs$r is gi#en &* !he !ransla!i$n $" !he cen!ral $in! $" !he
a!!ern using screen c$$r'ina!es+ 6re#i$us !$ !his calcula!i$n a mirr$r !rans"$rma!i$n is alie' !$ !he ca!ure'
image s$ !here is n$! c$n"usi$n &e!(een righ! an' le"! !$ !he user+
,n $r'er !$ manage an e""icien! m$#emen! $" !he curs$r, an' all$( !he user !$ $si!i$n elemen!s in !he screen, a
(eigh!e' a#erage $" !he las! "i#e AOC m$#emen!s carrie' $u! &* !he curs$r (as use'+ This r$ce'ure !akes !he las!
O l$ca!i$ns "r$m !he curs$r an' calcula!es a (eigh!e' $si!i$n+ The (eigh! assignmen! gi#es re"erence !$ !he
m$s! recen! $si!i$n+ The calcula!i$n $" !he curs$r (eigh!e' $si!i$n is &ase' $n !he use $" a #ec!$r $" O (eigh!s
genera!e' as "$ll$(s
;O<
I
>$r all 9 P i P 7, (here 7 is !he am$un! $" $si!i$ns !$ &e c$nsi'ere',

i
7eight
i
+
=
1
1
A3C
1+ ,! is necessar* s!an'ar'ize !he c$m$nen!s $" !he #ec!$r Q7eight iR !$ !he in!er#al ;9,1<

=
=
=
1
9
, 9
N 8
8
8
i
i
7eight
7eight
7eight N i
A4C
2+ The $si!i$ns are s!$re' in a #ec!$r, an' r$cee' !$ calcula!e !he ne/! $si!i$n (here !he curs$r
(ill &e l$ca!e'+

i
N i
i
9 x i 9 x
7eight movement osition :inal N S
1
9
C , A C , A
=
=
=
AOC
This (a* i! is manage' !$ $!imize !he m$#emen! $" !he curs$r gi#en a sm$$!h m$#emen! in !he screen+
()ecution of the left clic*" The ac!i$ns !$ ress an' !$ release !he le"! click are carrie' $u! &* !he user !hr$ugh
!he m$u!h $ening $r m$u!h cl$sing ac!i$n+ There"$re, an alg$ri!hm (as 'e#el$e' !ha! all$(s 'e!ec!ing !his
ac!i$n, as (ell as !he !ime !he m$u!h remains $en+ Then &* using =6, "unc!i$ns $" -in'$(sD is $ssi&le !$
e/ecu!e !he ac!i$ns require' &* !he user in !he s*s!em+ This r$cess is carrie' $u! &* means $" !he 'e!ec!i$n $"
!($ ellises A>igure OC, (hich reresen! !he $u!si'e &$r'er an' !he insi'e &$r'er $" !he lis, (hich are searche'
in in"eri$r hal" $" !he image re#i$usl* "il!ere' using a Gaussian sm$$!he' $era!i$n
;T<
+
>igure O+ 3e!ec!i$n $" !he $en m$u!h
The "$un' ellises mus! "ul"ill cer!ain c$n'i!i$ns as "ar as !heir 'imensi$ns, (i!h !he in!en!i$n $" n$! c$n"using
!he m$u!h (i!h an* $!her sha'e $r "$rm !ha! can &e (r$ngl* 'e!ec!e' &* !he alg$ri!hm+
-i!h !he in!en!i$n $" $!imizing !he e/ecu!i$n $" !he click $era!i$n, &e"$re making an* 'ecisi$n regar'ing !he
ac!i$n !$ make, !he alg$ri!hm all$(s #eri"*ing cer!ain s!a!es $" !he user.s m$u!h an' acc$r'ing !$ !his resul! i!
c$n'uc!s !he ac!i$n+ ATa&le 1C
+resent State +re'ious State &ouse Action
?en M$u!h Cl$se' M$u!h 6ress 2e"! Hu!!$n
?en M$u!h ?en M$u!h T$ $si!i$n Curs$r
?en M$u!h Cl$se' M$u!h T$ $si!i$n Curs$r
?en M$u!h ?en M$u!h Release 2e"! Hu!!$n
Ta&le 1+ @eri"ica!i$n $" s!a!es $" !he user.s m$u!h+
=''i!i$nall*, a !imer "$r !he ac!i$n $" !he click is use' !ha! ($rks $" !he "$ll$(ing (a*I
1+ -hen an $ene' m$u!h is 'e!ec!e', !he m$#emen! $" !he curs$r s!$s 'uring O h$!$gram, !his
(a* i! is $ssi&le !$ &e 'e!ermine' i" !he user (ishes !$ c$n'uc! !he clicking ac!i$n $r a''i!i$nall*
he (ishes !$ 'rag an $&)ec!+
2+ ," (hen assing !he O h$!$gram, !he m$u!h is cl$se', i! is c$me !$ release !he click, in !he
$riginal $si!i$n (here i! (as resse'+
3+ ," (hen "inalizing !he O h$!$gram !he m$u!h "$ll$(s $en, i! s!a*s resse' !he click, !$ e/ecu!e
!he 'ragging ac!i$n $" !he $&)ec! selec!e' &* !he user+
,. -esults
The 'e#el$e' alica!i$n is $" !*e -in'$(s GE,
;U<
, &ase' $n a "allbac4 !*e, !ha! su$r!s !he managemen!
$" messages sen! &* !he $era!ing s*s!em, (hich all$(s !$ !ake care $" !he requiremen!s $" !he user+ The
alica!i$n $""ers !he $ssi&ili!* $" m$#ing !he curs$r all $#er !he c$mu!er screen, !hr$ugh m$#emen!s $" !he
hea', as (ell as ressing $r releasing !he le"! click, (i!h a single m$u!h $ening $r cl$sing+ This (a* ac!i$n such
as $ening a r$gram $r 'ragging a "ile can &e ma'e+ The alica!i$n gi#es !he $ssi&ili!* $" !($ le#els $"
m$#emen! sensi!i#i!*, 'een'ing $n !he user.s caa&ili!ies+ =ls$ i! all$(s !he c$ns!an! m$ni!$ring $" !he ca!ure'
images an' r$cessing &* a (in'$( l$ca!e' in !he screen, (here !he ac!i#i!* $" !he user is sh$(n in real !ime
A>igure TC+
>igure T+ -in'$( $" m$ni!$ring
-in'$(s 56D $""ers 'i""eren! $!i$ns !$ imr$#e !he accessi&ili!* !$ !he s*s!em inclu'ing8 "$r e/amle, a
ke*&$ar' in screen+ These !$$ls al!$ge!her (i!h !he alica!i$n 'e#el$e' in !his ($rk all$( !he user !$ e/an'
!he lis! $" !asks !ha! can er"$rm, (i!h n$ nee' $" using !ra'i!i$nal inu! 'a!a eriherals+

.. Conclusions
=n al!erna!i#e !$ !ra'i!i$nal inu! 'a!a eriherals (as 'e#el$e' using ar!i"icial #isi$n+ The alica!i$n is
'irec!e' seciall* !$ in'i#i'uals !ha! resen!0'isla* s$me m$!$r 'isa&ili!* !ha! a""ec!s !he m$&ili!* $" !heir uer
mem&ers+
The alica!i$n c$un!s (i!h a simle in!er"ace, (hich '$es n$! 'eman' r$l$nge' !raining "$r i!s use+ =ls$, !he
s*s!em requires $nl* a camera -e& "$r i!s $era!i$n, reas$n (h* i! reresen!s an ec$n$mic al!erna!i#e+
The 'e#el$e' alg$ri!hm is a&le !$ r$cess u !$ 29 images er sec$n', (hich $""ers !$ !he user a see' $" use
near !$ a c$n#en!i$nal m$use+
The use $" !he me!h$' $" (eigh!e' $si!i$n a#erage, all$(e' a sm$$!h m$#emen! $" !he curs$r $n !he screen
making a m$re e/ac! m$#emen! $" !he curs$r, as (ell as $""ering !he $ssi&ili!* $" es!a&lishing !($ le#els $"
m$#emen! sensi!i#i!*, (hich "acili!a!es !he use $" !he alica!i$n+
-eferences
;1< =nz$la, R$sari$+ :&ndacin aso a aso; <Un m&ndo con acceso para todos=+ @enezuela+ 2999+
;2< Dapena P., Jos. Me!$'$l$gas ara 3esarr$ll$ en C$muni'a'+ ,@ Beminar* !he ?era!ing B*s!em 2inu/
, 2994+
;3< ,n!el+ >pen !o&rce "omp&ter $ision Librar9 ?e@erence Aan&al+ ,n!el C$r$ra!i$n+ E+B+=+ 2991+
;4< @i$la 6+ * V$nes M+ R$&us! Real%Time >ace 3e!ec!i$n. International Bo&rnal o@ "omp&ter $ision, + 13U%
1O4, 2993+
;O< C$ianiz T+ * T$rresani 2+ 23 3e"$rma&le M$'els "$r @isual Beech =nal*sis+ !peechreading b9 C&mans and
Aachines, + 3G1%3GW, 1GGT+
;T< 6a)ares G+, 'e la Cruz V+, M$lina V+, Cua'ra'$ V+ * 2ez =. Imgenes Digitales. rocesamiento prctico
con Bava+ =l"a$mega, MX/ic$+ 2993+
;U< Microsoft Corporation. 7indoDs EI+ E+B+=+ 299U+

You might also like