Unsupervised writer adaptation of whole-word HMMs with application to word-spotting

被引:8
|
作者
Rodriguez-Serrano, Jose A. [1 ,2 ]
Perronnin, Florent [1 ]
Sanchez, Gemma [2 ]
Llados, Josep [2 ]
机构
[1] XRCE, F-38240 Meylan, France
[2] Univ Autonoma Barcelona, CVC, Bellaterra 08193, Spain
关键词
Word-spotting; Handwriting recognition; Writer adaptation; Hidden Markov model; Document analysis;
D O I
10.1016/j.patrec.2010.01.007
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
In this paper we propose a novel approach for writer adaptation in a handwritten word-spotting task. The method exploits the fact that the semi-continuous hidden Markov model separates the word model parameters into (i) a codebook of shapes and (ii) a set of word-specific parameters. Our main contribution is to employ this property to derive writer-specific word models by statistically adapting an initial universal codebook to each document. This process is unsupervised and does not even require the appearance of the keyword(s) in the searched document. Experimental results show an increase in performance when this adaptation technique is applied. To the best of our knowledge, this is the first work dealing with adaptation for word-spotting. The preliminary version of this paper obtained an IBM Best Student Paper Award at the 19th International Conference on Pattern Recognition. (C) 2010 Elsevier B.V. All rights reserved.
引用
收藏
页码:742 / 749
页数:8
相关论文
共 50 条
  • [41] Effects of prompt style on user responses to an automated banking service using word-spotting
    McInnes, F.R.
    Nairn, I.A.
    Attwater, D.J.
    Jack, M.A.
    British Telecom technology journal, 1999, 17 (01): : 160 - 171
  • [42] WORD SPOTTING USING CONTEXT-DEPENDENT PHONEME-BASED HMMS
    MATSUOKA, T
    IEICE TRANSACTIONS ON COMMUNICATIONS ELECTRONICS INFORMATION AND SYSTEMS, 1991, 74 (07): : 1768 - 1772
  • [43] Effects of prompt style on user responses to an automated banking service using word-spotting
    McInnes, FR
    Nairn, IA
    Attwater, DJ
    Jack, MA
    BT TECHNOLOGY JOURNAL, 1999, 17 (01) : 160 - 171
  • [44] Does whole-word multimedia software support literacy acquisition?
    Karemaker, Arjette M.
    Pitchford, Nicola J.
    O'Malley, Claire
    READING AND WRITING, 2010, 23 (01) : 31 - 51
  • [45] Does whole-word multimedia software support literacy acquisition?
    Arjette M. Karemaker
    Nicola J. Pitchford
    Claire O’Malley
    Reading and Writing, 2010, 23 : 31 - 51
  • [46] Phonological Abilities of Children with Dyslexia in Jordan: A Whole-Word Approach
    Huneety, Anas
    Khashashneh, Nedaa
    Mashaqba, Bassil
    Abu Guba, Mohammed Nour
    Alshdaifat, Abdallah
    EURASIAN JOURNAL OF APPLIED LINGUISTICS, 2023, 9 (03): : 21 - 32
  • [47] Disfluent whole-word repetitions in cluttering: Durational patterns and functions
    Bona, Judit
    CLINICAL LINGUISTICS & PHONETICS, 2018, 32 (04) : 378 - 391
  • [48] Whole-word phonology and templates: Trap, bootstrap, or some of each?
    Velleman, SL
    Vihman, MM
    LANGUAGE SPEECH AND HEARING SERVICES IN SCHOOLS, 2002, 33 (01)
  • [50] MAP Estimation of Whole-Word Acoustic Models with Dictionary Priors
    Kintzley, Keith
    Jansen, Aren
    Hermansky, Hynek
    13TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION 2012 (INTERSPEECH 2012), VOLS 1-3, 2012, : 786 - 789