Design of a prototype in line inspection system for apples

194 Pages·2000·18.9 MB·English

by  Aneshansley, D. J. (Daniel J.)

Historic, Archive Document
Do not assume content reflects current
scientific knowledge, policies, or practices.

aSB363
ES
-A54
2000
58-193 1-5-019
DESIGN OF A PROTOTYPE IN
LINE INSPECTION SYSTEM FOR
APPLES
USDA-AFRS
PREPARED: DJ ANESHANSLEY
Uri ectei edc h
oye oFT a ton of
Xela Uitte
ee
—————————
Teste Caiuiicel Bl cclay
U.S.D.A., NAL
FINAL REPORT
Agreement Number 58-1931-5-019
Oct. 1, 1995 — June 30, 2000 Cataloging Prep
Design of a Prototype In-Line Inspection System For Apples
Prepared by:
Daniel J. Aneshansley, Assoc. Prof.
Cornell University
The agreement continued the collaborative work between the Appalachian Fruit Research Station
and Cornell University with techniques to segment fresh and stored apple bruising on the same
apple (Throop et. al. 1995). Studies for either positioning the stem/calyx ends in a known position
by orientation (Throop et. al. 1995, Throop et. al. 1997, Throop et. al. 1999) or identifying the stem
and calyx (Campins et. al. 1997) to avoid detection as a defect were pursued. Spectra to determine
the reflectance properties of the many different defects that can occur on the different cultivars that
grow in Eastern United States were recorded. Statistical and neural network analysis was used to
find the best wavelengths for identifying 22 different defects on 10 cultivars (Aneshansley et. al.
1997, Miller et. al. 1998). Methods to segment defects were developed for each of the identified
wavelengths or combination of wavelengths (Throop and Aneshansley, 1997a and 1997b).
Orientation, reflectance spectra, and methods of segmentation were incorporated into an inspection
station to find defects on apples at a rate of 5 apples per second (Throop and Aneshansley, 1999a
and 1999b). With the possibility of orientation, investigations using body transmittance to detect
internal browning were conducted (Upchurch et. al. 1996). Further work needs to be done to bring
this work on line.
Papers and patents associated with this project are listed below:
Throop, J. A., D. J. Aneshansley, and B. L. Upchurch. 1995. An image processing algorithm to find
new and stored apple bruises. Applied Engineering in Agriculture of the American Society of
Agricultural Engineers. 11(5):751-757.
Throop, J. A., D. J. Aneshansley, and B. L. Upchurch. 1995. Apple orientation on automatic sorting
equipment. ASAE Paper 956176. ASAE, St. Joseph, MI 49085-9659
Throop, J. A., D. J. Aneshansley, B. L. Upchurch. 1997. Apple orientation on automatic sorting
equipment. Proceedings from the Sensors for Nondestructive Testing International Conference,
Orlando Florida. pp324-342.
Aneshansley, D. J., J. A. Throop, and B. L. Upchurch. 1997. Reflectance spectra of surface defects
on apples. Proceedings from the Sensors for Nondestructive Testing International Conference,
Orlando Florida. pp143-160.
Campins, J., J. A. Throop, and D. J. Aneshansley. 1997. Apple stem and calyx identification for
automatic sorting. ASAE Paper 973079. ASAE, St. Joseph, MI 49085-9659
gy oo ¢ . cbih cha | a pele ePye aleeee
S see raaneh | ‘ 4
i Copee Boro e a 7 si 7 tee
gor"
ity 88us
why Brak €
<: “aE “
sve UT tis nee
as woh
; 4
be
K
p}t y; ‘
€
j 4 i
“4 fg Sh) ; ' -
ar
’ j j a |
' ~ a ts. es oa ve a
4 i
¢ ‘ } f% oats ; 4 4 rc f
si ive ry a f dansttrg 0 | ri ‘es art Le 4
i 76 ¢ Ai sie al # 1ouriae. an fT ‘y ear “Fa es}
: 7 a 7 pe ase i Or
yy De Hoatafigt a
“~ as
(wind ae ‘ eaaere %
' 7 ¥ ai. i ip oY
uiinabi cy ied. bee meee See Fae. ai
2890b RALA es wet1 03 Ui2h®
ee =P me in a }
Throop, J. A., D. J. Aneshansley. 1997a. Inspection of processed fruit before and after peeling.
ASAE Paper 976044. ASAE, St. Joseph, MI 49085-9659
Throop, J. A., D. J. Aneshansley. 1997b. Apple Damage Segmentation utilizing reflectance spectra
of the defect. ASAE Paper 973078. ASAE, St. Joseph, MI 49085-9659
Upchurch, B. L., J. A. Throop, and D. J. Aneshansley, 1997. Detecting internal breakdown in apples
using interactance measurements. Postharvest Biology and Technology 10(1997)15-19.
Miller, W. M., J. A. Throop, and B. L. Upchurch. 1998. Pattern recognition models for spectral
reflectance evaluation of apple blemishes. Postharvest Biology and Technology 14(1998)1 1-20.
Throop, J. A., D. J. Aneshansley, and B. L. Upchurch. 1999a. Fruit orienting device. U. S. Patent
No. 5,855,270. 19p. Commissioner of Patents and Trade Marks, Washington, D.C.
Throop, J. A., D. J. Aneshansley, and B. Anger. 1999b. Inspection station detects defects on apples
in real time. ASAE Paper 993205. ASAE, St. Joseph, MI 49085-9659
Silce) +s tes dio’ hod. tomes ‘be nol aenl avec unten €
Ciye-n3008 IM riqetol 37
THe soteeol" sith ity ete patpead susragtl tn a ave! eal
UPRC-22008 IEA stageol 32 “TABA Papas a
AWMOSS nt: oohoigen isaiwedte giro ot) Toy | Vuln WA Chew. soc i
G).8 TOC. yoolumieT bre yqoiat & srerbeot <memeusg
ann. } FMM Asters marie eee; soewida J # ett noowlt Pf a
tae" ‘ awit iP .t- vai il 3S . oA Labb ie 7 OUGk ‘ wera" * cusrre told stein 10 a OnevA.
lem eso Une Wht. ter) Soto
10 cofpnndea pa1 ieth shard bas
zaliun NO sletleh teed alter-nonequil 4400! so_nA & bee. weleieaC
WAG ZK0Ph 4 siqenct 12 AAA EOCERS ie
AN IMAGE PROCESSING ALGORITHM TO
FIND NEw AND OLp BRUISES
J. A. Throop, D. J. Aneshansley, B. L. Upchurch
BtadAaAd((Arireri pfsNuBetefospbiIaSeafuuislRsesceTren)e . tr R,a ee AwtrnwdTroiCyBice h et Tremrifbhsu .aisleioiie gmxnbtscp eaheteeSa sgwti ttvte ewoohotadneee lrele wikdeertoure sndoen afr)a c fit eb nbtn e toirdb‘gbdhuouRr rel nneiuunso, sadidieae wleass nw tdereIDd ie/sdiemp,r m. leb aae uairssgdtngbuc sA iebeiia sn trssosws hfeuu uheaeisesineda la ’asrl tbn leeegwerdyudxio ruas pct reipiNheoadsipst Ipne t,litsR hesrt t smsesoIfaau shm reonpar eacrdwsepfae gacawhelnmltersaae a od spcrcv hpettgdaearofa (eeu ov nod fsvcceeura2ieeels ncm5snr lmbt oesigoaorlhnid tlrpud nao, leit gd s dhs3b.a pe e5 aesnh nd isudetoat mero rtlh df etgenma )iiroce dcs ruhwese(irp nuad 0etoetni.ph bincgs9rrfm htm8euye oa i iads rs tset baboeosnuo sc ecstet.dt o1eidh ie.rmdn sr 3pagPci8loor rdo srlekeeu tuditevfninn ericit.cbl aono) sry eunui ut ddsdnnhii f t(esartenlmsnemohede,tedenrm wuw tte d ieasitKcsdrhei(ebe a sso2yrls rl4bosuwukd r)liheoe hud sarr e)ilv wdaes sbeevsn rbet er.udordleii s .u ns hil AenasioaSsptoreglw.hphehsrs nealat om sepe aT.arseshll t ,nsaeoih dt meswzT hA ee eohap lnpdfenpiea. bNgltas rhIheuarsuteR n-lA ie ibsgfbrqbienro surilsfaurotium lriileniaiastr hutsgreehayoteeded,mlded..d d
ibmhnruasulvpiteesi cevtdao brrei fear(bnoGMatll marhtgete ato rfperui rimrlnxeetpsoi tbthtn cr meaeceulasoadi.srf ,m s s eeett1rbsdoh9dtay e 8a 1sta)Nieoi.psIpnwm ptalRiailry tcgea ah etil sea mt aiCafsgoblnseefra uw use is aoasipfrealfp es ilga seeoneir w ra neiraNfltcptIlohph hrRalemm tera a s srt .chei lafauoa lvsnsDehei sic unitgmffgbaiira neteoceandemln a tocCrtfeoehi mesurippNTncreehtIu creeRltasd ath rauatra-bropes soeef hv tl bheaerebcf puiotyierraas d lenp fsgc2ei.eo4nr c r diliihmuwt.neshh gttim Nee crrIht sRwh ae ( sws(r herseareofhplelnaae elpt cyeibt fo raanuacnsifptchsaopeirecl )pdti. oeh rdaoa Osfn n t0dlob. t y7ehh aeeeipnerltpl odll iesap rh1ate.toiia 5cwm a)rnal ow -goi aetrmtsteho o
artdr ehieqeqmfu uAelaipend rircsidxatiinetggaolii nntcc aga el ra stentylidae itm lnielsiie-etsmmivccaiceaoagnlnln ea s tnuccibolmananamigsucens nkirgigfab fir oeoodtrrcua hmatn tiapdo tt suyonpr (paoeTetrdarcimy aaaaltlulioe sirm.ez, add agi tes1its9doo8 an5rt o)atf. ipv oraanHorp ocipewalisbeensi v leNoirfInto,Ryer SuiUlctnenphosbccasrrr hnauegugtiaerehs.sc ea ednhsw ‘ T iuhtteneitthbs o rs auuplee.bir, xes ot oe1bevd9elen 9qerd2u me)ta .ditls i somNufeSeIa t RnSo(d o2ro r4anrt geeei efhnx l tgceo ac( eftRtetfawreednuroh ci kteb m utrohgwuoenlhif tes ihibrcrnsr ehg)uf ,i el stced eobcudauurt ltli.a dt,nn i gscs l1eshuo9c ae7wov 6lloed;iyf s
NIRA rneofnlsetcattainsctiec alw asa lgdoervietlhomp edfo r (lTihnreosocpan eti maalg.,e s1 9o8f8 )a papnlde rbeotmho vaolld f(rsotmo resdt)o rabgreu,i sehsa s annodt nbeewen bardudirseess,s eda.c quired during
tested on experimentally machine-harvested apples
(Rehkugler et al., 1989). This algorithm creates a new OBJECTIVES
tiiaisshdm rjsaeauigscbneehtvno retlar dfctitrpeneiodgdxm .e lf rtaTvhonhaemdel uoetbhsrcle io agoncaifkvvn eeatrlrhpa teig xeebedsdly a meicat lmvouae srgrtoeaewa. .rg si bTnThihagenre ae rrayeo esr vpiuagllilituniumranaaalgtlg eiediti ymm aafggbooeeyrf odusfees pmiaouTnrsnhgaese trt e rN fasiItoruRsleptd p loaarbinneje defd ecl txenipvceveatwerra iinmaibcsbere lnut.etio sa s,led T de hvmmeetie litnoshspsiuo emedc au onfm nrt adoo l mseg shoOtrasbitipjmoteraeh etcmde tf aiwcavtthpheoipe rc l,he vi sas al cnubatdenyo
threshold factor required in the algorithm for optimal
Se bruised tissue classification.
AInSfAoAErrm taiPtcailpoeen r haNanosd. b9Ee3le-enc6 t5r3ri4ec.av li ewTeedc hnaonldo giavpepsr ovDievd. ofofr ApSuAbEl.i caPtrieosne ntbeyd thaes MATERIALS AND METHODS
Angorti cMucelontntuisroianld eRroefes dse paercacinhf iceS enprdrvooidcruesc,e tmso er nitst hf eo ra butythh eo rCiso.nr fnoerlmla tioUnn ivofe rtshie tyre,a deUr SaDnAd -i s LC1A9 05112F AdNi-o0de1 1a,r rayS unlinnye vaslcaen, cCaamleirfa. ) (wEiGth& G a R2e5t imcomn
AEaRnnSedgsA ieETnaBhe rieecorM hl,e oa gmuSiAtbucshepasoprloro c sri EatEtn naeggr Sieip nnePeecrJeioearafrmlei,isne ssgsto,A,r g ,r ADiI.Ctac hounalTmicteehaullr,r l oa olJN p..U Y,nEA.i n;nvg eeAirsSasnhnieAatdeEny r,s, Bl reDMuUyeecSp,mteD . Ab A-eLSAor.fgA rEAEUig npcrgcuiMcihleunutlmeurtbrecuraerhalr,,l wfl(ra0oos.nn9 gt8p paoeisnfrs. f )t ohflrieeml nteseldr e nus(s bWiyrncaa gat p ttaeunlnr i neaedp 8 e9srBtcth,aue nr eKi opmdrsaeaogtckete,sis .nsg RooIroc mfha e(gfsD2et.T e02 rd8,ia5 gn6idNt, .i Yzw.aiDt)tai hto ianna
Research Service, Appalachian Fruit Research Station, Keamesville, W.V. Translation, Marlboro, Md.) set for externally triggered
| bC1i4ol8ro5gr3ie:c salep -omEnanidgli:in nge<e jraiatnu1gt2,h@ ocrCo:om renlJella.lm eedsUu n>i.Av. e rsTihtryo,o pR,i leDeyp-tR.o bbo f HAaglrli, cuIltthaucrae, aNnYd aS1ni5dn6 .g2la5e toktlHailzn ,el i5npe2e 5r c ocutonrutin gtgsoe fr p 3e6ra0 c.lq iunTieh,se i at iilnoitnnee nstmirtaoynd sefve,ar l uveca llofuocer k eoedfv e2r0ay,t
Applied Engineering in Agriculture
VOL. 11(5):751-757 © 1995 American Society of Agricultural Engineers 0883-8542 / 95 / 1105-0751 751
fmpCmiiiiovtxtceye or,lr co ocSuio.naDmmtn.aspd) gu etptp eehrrreo o vfllii (indGnetee had etw eesrcwaaea ap npyp u lalvecse3 aer8m ase6gturSoera Xdf./ ta 2rcsi5eoSg ugetbmhwrra aHotszu b to,ati hfn3 oNe6ro 0trmhVtfee hud nx.s c tSte1iip0Toop5hnueHesxr (f2ssf0ouci0buag.snlst .idi n Ttg1uha aten e dfada nacdvdit nieo vrr1tia bhd)ge,eae dn d ides il irnnniettoneaor nc mssha2ip 0tlhy0ipe .zir exeode Tf lh i eemt aoav cgapahelr uoged(r3 fue6iyc0gi t. n lp1etodivhf)xea .elt lt h iecnc toorellneususiummtlnnyt inoiigfss
(RplOarheanDidmodTIo pgm_l)sroLal annui mpdmn(lmi,imeani nocsanndeWtcegdeaira lso nnh blw, .eah)4 tnic 0wgtoAuewen,aeeDs grnaiee tas4 cta0are u(nyWdsM l ,eiio dTccou r rftGoaf eedosnreniro si fefglfarrthauela tfstgl leoCi errso c itntE6tai).l[hvn0e 2em,dc5 at4rrMdc adiiehnc cuva,dremm i ol nbsoCceopl(atrfme1nht ev0ed ne estClusmo.racir“f nlepaC)dn.c]”,t,e. LmpluereonnoiaatEcfnDiaeoc cserhlmsAfs Ni lDatbw vpeaaripLc sslikoe cbg wolrtien oo mv Puoaiasnlgnsdmue s ot t ioh(aeFotfv InihegL i rsTm.aiE agtsRehg1 e aep AecaPriabnnfnPda toL cerEbkn1m esbge i)Irdtf Mo.ioiuA seu nGsnlTd Edoh aeodf fseo odrn t ehtbeeahad an ctdb k rfgaaoi r rg5om h outtxrnhe iedrs5
Apples were rotated by a stepper motor (ABS 3008-006, normalization.
12V, 1.0°, Hurst, Princeton, Ind.) driven spindle on the
oasa pppphpgoeAlosre oie cdt tu ei(s sGstssrtouieeamda n etd ( taTouLrfhrda rn ktoeheodetsp o a7cPre.lrte6ay p2slar tlie.icc,sc msed1 ,ni 9tf(9 Bf33uu.)suf0.enf rabi.lrn o.u),i sdNei.daY m.e)t‘ eRires d kwnhDioetwleni c idetoolu rsib’ne NurctopoOo pwlR0TesuM.hr mA eTnwL h h1Ieo8s Zss0uuE aemmv p eisBroxauafAergm leCe tK hrineGspoo RiwrexcsOmqee auUlnla tNiilesiDz r ne tfodero.c nOus Fogni rdltTe.Auyha Pm tePneAf Lr lo lrEr pe teihmcIxaaaoecnMillh Anu Gi8minE01nng8 %st 0e npsopoiififxt x eitele1hlsse8 0cc aooerflpen iturxtmsaehnellet
to be normalized is found and divided into 200. Each pixel
ALGORITHM DESCRIPTION intensity for that column is multiplied by the resulting scale
LOAD AND PROCESS IMAGE OF DELRIN SPHERE factor (fig. 2a). A similar operation is performed on rows
The delrin sphere image (fig. 1c), captured for the same 181 to 360 resulting in the apple image background
lighting and camera setup as for capturing the apple images
24
SNAN<NI LS
ARAASN
SSRPENY S NS
RNAAN N
SAAS *rA eSeR S IESE
eens
A2pT3a2SS2n T aE N MAES
Aas esS CaSOnAORnS e
SSNS SeSeN eS ber
RBy ee
PH
~
vee
ASN!SWS S S, SA SeSAUT NS
ASSN
os
SSAA
ss
eeSA NS
Se PES
PSGS :RS SASI ARS ESNe ana
SARSP ASNS Y
RwewA eM AN SayA SSY
PPIRRISSIS
(c) (d)
(a) (b)
Figure 1-Composite NIR grey tone digital linescan image: (a) apple with 24-h-old bruise, (b) apple with 2-month-old bruise, (c) delrin sphere
unnormalized, and (d) delrin sphere normalized to grey level 200.
APPLIED ENGINEERING IN AGRICULTURE