An HMM-based approach for off-line unconstrained handwritten word modeling and recognition

Yacoubi, Mounîm El; Gilloux, M.; Sabourin, Robert; Suen, Ching Y.

doi:10.1109/34.784288

Cited by 195 publications

(71 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Hidden Markov models (HMMs) are able to do this, which is one reason for their popularity in unconstrained handwriting recognition [14], [15], [16], [17], [18], [19]. The idea of applying HMMs to handwriting recognition was originally motivated by their success in speech recognition [20], where a similar conflict exists between recognition and segmentation.…”

Section: Introductionmentioning

confidence: 99%

A Novel Connectionist System for Unconstrained Handwriting Recognition

Graves¹,

Liwicki

Fernández

et al. 2009

IEEE Trans. Pattern Anal. Mach. Intell.

1,740

887

View full text Add to dashboard Cite

Abstract-Recognising lines of unconstrained handwritten text is a challenging task. The difficulty of segmenting cursive or overlapping characters, combined with the need to exploit surrounding context, has led to low recognition rates for even the best current recognisers. Most recent progress in the field has been made either through improved preprocessing, or through advances in language modelling. Relatively little work has been done on the basic recognition algorithms. Indeed, most systems rely on the same hidden Markov models that have been used for decades in speech and handwriting recognition, despite their well-known shortcomings. This paper proposes an alternative approach based on a novel type of recurrent neural network, specifically designed for sequence labelling tasks where the data is hard to segment and contains long range, bidirectional interdependencies. In experiments on two large unconstrained handwriting databases, our approach achieves word recognition accuracies of 79.7% on online data and 74.1% on offline data, significantly outperforming a state-of-the-art HMM-based system. In addition, we demonstrate the network's robustness to lexicon size, measure the individual influence of its hidden layers, and analyse its use of context. Lastly we provide an in depth discussion of the differences between the network and HMMs, suggesting reasons for the network's superior performance.

show abstract

Section: Introductionmentioning

confidence: 99%

A Novel Connectionist System for Unconstrained Handwriting Recognition

Graves¹,

Liwicki

Fernández

et al. 2009

IEEE Trans. Pattern Anal. Mach. Intell.

1,740

887

View full text Add to dashboard Cite

show abstract

“…Ibrahim and Odejobi [15] developed a system to recognize handwritten character of Yoruba Upper case letters. The work presented an approach of Bayesian and decision tree.…”

Section: Related Workmentioning

confidence: 99%

Yoruba Handwriting Word Recognition Quality Evaluation of Preprocessing Attributes using Information Theory Approach

F.Ajao¹,

O.Olabiyisi²,

O.Omidiora³

et al. 2015

IJAIS

View full text Add to dashboard Cite

This paper presents an approach to evaluate the quality of handwritten words using the set of features in the preprocessing stages. This is to determine the effect of various stages of preprocessing of the recognition of Yoruba handwritten characters. We demonstrate our methods using handwritten words samples of the domain of Yoruba medical terminology collected from indigenous Yoruba literate writers. Samples were digitized at 300dpi to facilitate much detail representation of the image dataset. We study the impact of entropy measure for an intuitive interpretation of the analysis of the handwritten words. From the experiment carried out, it was observed that the entropy measure of handwritten word is higher than the typewritten word. This implies that the Information content of the handwritten word is affected by perturbations which need to be removed using appropriate image preprocessing tools to obtain low entropy measure which implies same information content as the original. General TermsPattern Recognition, Handwritten word recognition, Image preprocessing and Information Theory.

show abstract

“…In particular, the classical HMC widely used is obtained by putting in (1), which gives (4) In the following, we will call an HMC the classical HMC, verifying (4).…”

Section: Pairwise Markov Chainsmentioning

confidence: 99%

“…The range of applications of the latter is rather vast, covering signal and image processing, economical prediction, and health sciences, among others. Specific applications of HMCs in image processing include image segmentation [2], [3], handwritten word recognition [4], document image analysis, tumor classification, vehicle detection [5], acoustic signal recognition [6], or even gesture recognition [7]. Multisensor or multiresolution images can likewise be segmented using HMCs [3], [8].…”

Section: Introductionmentioning

confidence: 99%