Treść książki
Przejdź do opcji czytnikaPrzejdź do nawigacjiPrzejdź do informacjiPrzejdź do stopki
Red.S.Sędziwy,SchedaeInformaticae,Vol.17/18December2009
Kraków2009,ISSN0860-0295,©byUJ
11
C:X→RL–belongingfunction,
(7)
F:R
L→I∪{i0}–decision-making.
(8)
MappingBmeasuresthecharacteristicsofthetestobjectsandturnstheminto
itemsofthespacefeatures.
Theelementsofthisspacearen-elementvectors
x=(x1,x2,...,xn>∈X,whereX∈Rn.Thenthealgorithmgoestotheprocess
ofdetermininganobject’sd∈DmeasureofsimilaritytoclassesD
i,wherei∈I.
OnthebasisofthexvectorLfunctionsofbelongingarecalculatedandthenthe
valueofCmappingthatisanobject’smeasureofbelongingtooneoftheclasses
iscalculatedDi,C
i(ˆ
x),i=1,2,...,L.SincethereisLclasses,sowecanwriteRL.
Thefinalstageoftheimagerecognitionistodecidetowhichclasstoassignthetest
object.Themostcommonmethodisthegeneralrule:
∀ˆ
x∈X[[F(C
1(ˆ
x),C2(ˆ
x),...,CL(ˆ
x))=i]≡∀η∈I,η±i[C
η(ˆ
x)<Ci(ˆ
x)]].
(9)
Theobjectisincludedintheclassforwhichthevalueofbelongingwasthehighest.
3.Solution
Duringtheresearchintorecognizingroadsonphotographstwosystemshave
arisen.Inbothofthemtheengineusedtoidentifytheroadisatwo-layerneural
networktrainedbymeansoftheback-propagationmethod.Thefirstsystemconsists
ofasinglenetwork,whichisusedforeverytypeofimage.Thesecondisahybrid
systeminwhichtwo-layerPerceptrons(similartothefirstsystem)andtheKohonen
networkworktogether.
Fig.1.Howthefirstsystemworks
Fig.1illustrateshowthefirstsystemworks.Attheinputofthesystemaphotois
loadedandprocessed,thentheneuralnetworklocatesthepositionoftheroadon
thephoto.Ontheoutputthesystemreturnsabinarypictureoftheroad.During
