Seite - 146 - in Document Image Processing

J. Imaging 2018,4, 15 AppendixA. SomeRecognitionExamples ThisAppendixpresentssomerecognitionexamples. FiguresA1–A3present thebesthypothesis obtainedforseveral linesof theRodrigocorpus in theopenvocabularyexperiments,byusinga3-gram word-basedLM,a4-gramsub-word-basedLManda10-gramcharacter-basedLM. Text Image TextReference muerteepeormeresciaelporquantopassaraelmandami Word-based1-best me&peormataraelporquantopaganaelmanda Sub-word-based1-best mundo<SPACE>&<SPACE>por<SPACE>matara<SPACE>el<SPACE>por<SPACE>quanto<SPACE>pagana<SPACE>el<SPACE>manda<SPACE> mundo&pormataraelporquantopaganaelmanda Character-based1-best mucho<SPACE>&<SPACE>por<SPACE>meresc ia<SPACE>e l<SPACE> por<SPACE>quanto<SPACE>pagaua<SPACE>e l<SPACE>mandami mucho&pormeresciaelporquantopagauaelmandami FigureA1.Exampleof thebesthypothesesobtainedfor the12th lineofpage500ofRodrigo. Text Image TextReference portugal. Word-based1-best portugal.portugal. Sub-word-based1-best pe tugalzopetugalzo Character-based1-best por tugazportugaz FigureA2.Exampleof thebesthypothesesobtainedfor the9th lineofpage619ofRodrigo. Text Image TextReference maronlocaualleroeseyendoCaualleroenfermomuymal Word-based1-best nonloCauallero&seyendoCaualleroenfermomuydia Sub-word-based1-best naron<SPACE> la<SPACE>Caualle ro<SPACE>&<SPACE>seyendo<SPACE>Caualle ro<SPACE>enfermo<SPACE>muy<SPACE>dia<SPACE> naronlaCauallero&seyendoCaualleroenfermomuydia Character-based1-best maron<SPACE> l a<SPACE>caua l l e ro<SPACE>&<SPACE>seyendo <SPACE>Caual le ro<SPACE>enfermo<SPACE>muy<SPACE>mal maronlacauallero&seyendoCaualleroenfermomuymal FigureA3.Exampleof thebesthypothesesobtainedfor the4th lineofpage514ofRodrigo. References 1. España-Boquera,S.;Castro-Bleda,M.J.;Gorbe-Moya, J.;Zamora-Martinez,F. ImprovingOfflineHandwritten TextRecognitionwithHybridHMM/ANNModels. IEEETrans. PatternAnal.Mach. Intell. 2011,33, 767–779. 2. Al-Hajj-Mohamad,R.;Likforman-Sulem,L.;Mokbel,C. CombiningSlanted-FrameClassifiers for Improved HMM-BasedArabicHandwritingRecognition. IEEETrans. PatternAnal.Mach. Intell. 2009,31, 1165–1177. 3. Vinciarelli,A. Asurveyonoff-linecursivewordrecognition. PatternRecognit. 2002,35, 1433–1446. 4. Bianne-Bernard,A.L.;Menasri,F.;El-Hajj,R.;Mokbel,C.;Kermorvant,C.;Likforman-Sulem,L. Dynamic andContextual Information inHMMmodeling forHandwrittenWordRecognition. IEEETrans. Pattern Anal.Mach. Intell. 2011,99, 2066–2080. 5. Graves,A. SupervisedSequenceLabellingwithRecurrentNeuralNetworks. Ph.D.Thesis, Technische UniversitätMünchen,Munich,Germany,2008. 146