Fake currency detection using image processing

It is approved project on fake currency detection

1. JA\VAHARLAL NEHRU TECHNOLOGICAL UNIVERSITY ANANTA
PUR
ANANTA PURAMU -5 15002, AP,INDIA
Project On
Recognition of fake currency using image
processing
Presented by:
8.Anivinder
Reddy
M.K.Harik:I
B.V.Suma1hl
•18001A0<03
-18001A0410
•1900SA0<03
Undertheguidanceof
Sri.M.Sreedhar
Asslst<intPioressorAdhoc
oertmentofECE

2. Contents:
I.Airn orThc Project
2.lntroduction
3.Scopc of \\'ork nnd ldcn
4.0vcrvic\v of "fhe
Project
S.f lo'v chan 6.Steps
involved
7.fe111urcs involved
S.Block diagrnnt
9.RcsultS
I O.Novchy of
project I
I.Advnnu1g<.'-S
I 2.Lin1i1ation..
13.Application.s
14.Conclusions
I S.Fu1urc scope
16.Jtcfercnccs

3. Aimoftl1eproject:
,,.l11eICChnologyofcurrencyrL'tOgniti()nb11sicully'1i1nsroridentifying11ndCXln1c1ing
visiblenndinvisibletCaturesofcu1TCncynotes.Untilno,v,n1any1echniqucshavebeen
1>roposed10iden1ify1hccurrencynolc.Buithebest\vnyis10usethevisiblefeatures
ofthenolcforcxantplc,colorandsize.

4. Introduction:
,.F11kcCurrencyNoteisn1crnllhatrefersto1hccounterfeitcurrencynotesthnl
rapidlycitcula1cdintheeconon1y.1l1escdaystechnologyisbeengro\vingveryfas1.
Conscquenllythebankingsectorisalsogcningn1odcmdnybydny.Currencyduplic.ot
ionnlsokno\vnascoun1crfcitcurrencyisavulnerable1hrca1oncconon1y.
,.In 1hc proposed n1odcl • acquired image of currency note is checked \Vhcthcr it is fake
or
real on the basis of couJtting the nuntl>cr or intcrn1ptions in the security thread. The
can1cra
pic1urcs nrc noted and nrudysctlby MATLAB progn1minstalle<Ion computer.

5. Problemstatement:
, ·rhc coun1crfcitcrs no\vadays, c1n evade Lhe chen1ical propeny & physical t'Ca1ure
based
counterfeit paper currency dc1ec1ion system due 10 technological advancc1nent.
,.Thecirculu1ionofulru'};ct1111ounloffakecum:ncyi11cn.-:1se.s1hc:1.111ounlof
moneyincircula1ion.\vhichntaylead10highdc1nondforgoodsandco1nn1oditics.
l11eriseindemandintumCrt'.Ul<'.Sascarci1yofgoods,leadingto1.1risein1hcpriceof1hc
goods.This
leads 10currency devaluation

6. Scopeofworka11didea:
,. This projccl proposes an app«>•lch that \\Jill dc1ec1 f;ake currency note being
circulatc<I in our cou111ry by using Iheir irnage.Our project \viii provide required
mobili1y and con,pn1ibi li1y 10 n1os1of the people ond provides credible accurocy for
the fnke currency dc1cc1ion. \Ve arc using 111nchinc learning to nmkc it portable
:ind efficient.

7. Overviewoftl1eproject:
, The fnkc currency de1ec1ion using lllachinc lcan1ing '''as in1plenlen1ed on MATLAB.
Fc11urcs of currency no1c like serial nu111bcr. security 1hld. ldc111ific-a1ion mnrk.
Mnha1n1u Gaodhi ponrJit were CXlntcted. The process suui.;.; rro1n i1nagc ac:quisilion
10 cnlcula1ioll or intensity of each extracted fcuturc.

8. Flowchart:
lmoge Ac<1uisi1ion
l
Initial Scgmcn1:uion
Gray Scale
Con\•ersion
f(. llurc ex111u::1ion
Resuh F•kc"' Real

9. Stepsinvolved:
ln111geAcquisition:Itistheactionofretrieving<lJlirnagefron1source.IIgoesthroughLhein1agc
\\•bichi..givenasopalhas;1in1>ul.IIalsochecks1hcoverrillimage\vhichisgiven11sinput.II
selects1hcn.".<lUin.:dfcnturcs10proceedforthefurlhcrprocessing.
PreProcessing:Preprocingiufon1ili11rnttntcforOpt.:n11ions\\'ithin1oges:titheIO\\'Ctlevelof
11bstmction-bothinputandoutputarcintensityiinagcs.111caimofPre·processingisan
iinprovemcntofthei1nageduta1ha1supprcssun,,11n1cddis1ortionsorenhancessonici1nage
fea1urcs:cssentiRIforfurtherprocessing.
, Initial Segn1en1n1ion :In 1his, s1cp \VC divide an ilttagc into \'31'iou.s p-irts thru have si1nilar
auributcs \vhich
urccalledosinuagcobjc.-.cts.Itisthefirststc-1>forinn1gcanalysis

10. Continuation..
ros1 Proccs'iing :r\djusting c:<po!-urc. contr.Li,I and brighlnt.'$:.. and also udjus1ing colors. hues.
loncl),
S3turation and light levels. II alsochecks the true and fake pixels of the image.
...GrayScaleConversion:Thisi.tepisused10enhance1hcgrayi1nilg,C10e1nphasizedi.lrklin1.
illliglucrbackgroundandalsohelpsincheckinthebl11ckstripsofthercnlnote•Itnlsodc1cc1s
cx11c1ft.-.nturesofthenoteaflerconvcrtingtheitna,gciruograyscaling.
rfeatureE'1r:icrion;Extrncti,.thefeature1h:itnrcnt-.cdcxl10becon1p.;1rcdand10conclude
\\•hcthcr1hcnoteisf{lkcorreal.Jntheoverallprocessingeachstepisha\'ingtheunique\\·ay
ofcxtmctingthefca1urcsoftherc-111andfukcnote.

11. Featuresinvolved:
,.Coni rust ; The difference in brighlncss bct\\'c<:"n light and dark 11rcas of irnngc. Conlrast dctcnnincs
the
nun1ber of shade$ in 1he i1n11gc.
,. Energy : It is the dist1.1nccs of so111c qu.ality bCl\Vt.-.en the pixels of so1ne locality
,..l-lu1nogcnci1·: It expresses ho"•' similar ccr1nin clen1cnts(pixcls) of 1h1.; imngc: nrc. Generally on irnngc is
hon1ogcnous if each pixel in the in1age hos 1he snn1ccolor .
,lenn :Mean value is the sun1of pixel values divided by the total number of 1>ixel values.
,.En1ropy:Entropyisarncasurcofinu1geinfom1;11ioncon1en1.\Vhichisin1erpre1ednsthe
avemgeunccnainityofinfom1a1ionsource.Itisdefinedascorrespondingstalesofintcnsilylevel
\Vhichindividualpixelsc-annd:tJ>L

12. BlockDiagra1n:
.... ,_
EXTRACTIN
O
FEATURE
S
ln1ngc User
DISPlAYRESULTS
Oata
SCI
SVM
INPUT
Pn.
Processing
Grey Scale
Conversion
Edge
Dc1cc1ion
Scg.111cnlntio
n

13. Co11tinuation:
, Rl\IS : To gc1 an cs1in1n1c of the sin1ilarity bct\vccn source in1agc nnd the scgmcnlcd i1nagc, \VC use
rool n1can square error.Using 1his the da1a can be divided by best lil 10 find ou1 ho\v conccrunucd nn
image is.
, S1:i11durd d('\!i:ltion : Standard deviation or the irnngc implies a l_;T'OS!; lllCJISUrc or 1hc irnprt'Cision
or
vari111ion :1bout the target value of liglu in1cnsi1y ut c.1tch such data point
,.\f:1rinncl' :The variance gi ves an idea hO\Y 1111:pixel volues arc spread.
, Srnoocbncss:Smoo1hncss 111casurcs the relative !'n1001h11ess of intensity in a region. II is high for a
region of co11s:u1n1iruen!'ity and IO\\' for regions \\'ilh lorgcexcursions in the \>alucs of i1.s in1cns:i1y
levels.
, I OJ\1: Inverse Difference Mon1cnt is usually called hon1ogcnci1y lhat n1casurcs the local
hon1ogcnchy ornn
i1nage

14. Techniqueused:
Sup1>or1 V ctor Machine
, Support \f<".ClOr Machine or SVM is one or1he n1ost popul11r Supervised Le.urning
algori1hn1s. 'vhich is t•S<'.<I for Classificotion os \VCll as Regression problcn,is.
,. S\lpport ve<:tor nu1chinC$ (SVf\1s) urc a set of supervised le.urni ngn1cthods
us(.-.d
for cl11ssifica1ion,regressio11aod outliers: det(."(;t ioo.
-The advantag<'.S of support vcc lor machines arc:6ffcc1ivc in high dimensional SJN CC$.
S1ill
cff<. tivc in cases ''here nun1bcr ordintcnsions ls greater thnn the number of sru11plcs.

15. Results:
Genuinenote:
- I
r
..
.
I
"
I
-
•
•
•
figure.OriginiLIin1ag.c
Figure.IISVimage
--"
'
.
.
./"
r. I
' ,'
Figure.13\\'in1:1g._,
Figure.Genuinein1:1ge

16. Fnkc note:
Figu.1-ISVimage
....
Ftgun:.fakeim.igc

17. Noveltyofproject:
...De1ec1ion of ahc f.1ke currency note i.s done by couoting the nun1ber of in1crru1nioni.>
in the
thrc.ad line.
, Predicts \\
1hcther the note isreal or f'.lkc on the basb of number of in1crru1>tions.
,II'thenumberofintcm.iplionis1.cro.1hcnitisrealnoteothcn\'iscitisfakenote.And
:alsowecalculatethecn1ro1>Yofthecurrencyno1esrorthecnicicn1detectionof
Cakecurrencynote.
,.MATLAB sofh\•nrcis usc'.d to detect the fake currcnc}'note.

18. Advantages:
,.Simplicity
, Rcmov;i1Iorunncocss.uryda1nin
imngc

19. Limitations:
, lnpullirniltuion
... Accuracy
...Lightinglimi1ation

20. Applications:
,Fakecurrencydc1cc1onsystcrncanbeutlizedinshops.b:1nkcou1ltcrsundinco1npu1crized
1cllcr1nachinc.auton1crchan1111nchinesandsoon.
, ll1csystenisnrccreatedutilizingdiversetechniquesandulgorithrns.

21. Conclusion:
, l11e survival of 1he financial sy111111etry n1ay be affected \Yit h its value, rapidity.ou 1pul and
\\•cllbcing by
counterfeiting of b;1nk notes.
,. \Vhh improvcmcn1 of rcccn1 banking scl'•iccs. au1on1n1ic methods for paper
currency recognition
bl!COnH!vital in nu1ny upplica1ions such a.,.:; in ATI and tn1101nn 1ic goods seller 111t1chines.
,. TI1c sys-tc111 hos a best pcrfon1u•ncc for both agreeing valid banknotes and deleting
invalid duta. h nlso
shO\VS 1he techniques lOr currency rccog.ni1ion us ng in1agc processing.
,111eIndianc.urrencyno1eshavebeenidentifedandc-0un1erfeit1101csha...:;beenfound.ll1is\VUtkis
donebyusingvariousfilters.Thisn1cthodisveryrosy10in1plcn1cn1inrcnl1in1c\Vorld.Allas1
\VChaveconcluded1hn1if\\'cproposeson1ceflicientpreprocessingandfeaturecx1mc1ionn1cthod
1hcn\\'ccaninlprovc1hcaccuracyofidcntificotionsystem..\Veconnlsodevelopap1>fordetcetionoff1.-k
ccurrency.

22. Futt1rescope:
Manydifl'crcnl:idnptations.1cstsandinnovationshavebeenkepiforthefuturedue
101helackofti11lc.Asfu1·ure'
1orkconcensdeepernnalysisorpanicularnlechaoisntS.
ne'vproposals101ry<lifTerc111methodsorsirnplccuriosity.
,.I . In fut ure \VC \VOu ld be including u 1n001.1lc for currency conversion.
,. 2.\\'ecan iinplc111en1the sys1e1n for foreign currencies.
,. 3.Tn:1cking of device's loc.ation through \Vh ich the currency is scanned and
maintaining the
san1c in 1hcd.atabasc.

23. References
,.. Megha·11uLkurandAn1ri1Kaur,"Vt1riousFakeC"11rro11t1,•/)ett!ttlo11lt!t·/111iq11(s",lrucma1onalJournalfor
TechnologicnlResearchInEngint::cring.vol.1.no.11,July2014..
, BRKnvy••ndBDc,<endnm,"IND/ANCURRENCYDETECTION1INDDENOMINATIONUSING
SIFT',ln1cma1ionalJoumnlofScienceEngineeringandTechnologyJtcscarch.vol.4,no.6.June2015.
,,.\V,K.ElSaid,*FakeEg,•1111'<111C'urrenc)'D(•1ec1io11SJ·ste11111si11gTerture<1111/SJ111pe
Clu1r11tterisrics",huema1ionalJournalofCoolputcl'1-\pplicatioos(0975-8887),vol.143.no.2,JuJle
2016••
,..BinodPrasadyadav.C.Spatil,R.ltKarhc,P.Hpatii-"HSI'Tecl111iq11cb)
1using1\1.111'1.AB'"Thefokc
cuJTencyisdetL-cted111ununlly.

24. THANKYOU!!!