A hybrid radial basis function network/hidden Markov model handwritten word recognition system

Gilloux, M.; Lemarié, Bernard; Leroux, Manuel

doi:10.1109/icdar.1995.599021

Cited by 13 publications

(9 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recognition accuracy Gilloux [3] 83.70% Olivier [8] 72.00% Saon [11] 90.10% Choisy [2] 86.20% Structural NSHP-HMM 87.52% Table 3. Different recognition scores obtained on the SRTP dataset fined as different weights can be assigned to the different features in function of their discriminating power.…”

Section: Systemmentioning

confidence: 99%

Structural information implant in a context based segmentation-free HMM handwritten word recognition system for Latin and Bangla script

Vajda

Belaïd

2005

Eighth International Conference on Document Analysis and Recognition (ICDAR'05)

View full text Add to dashboard Cite

In this paper, an improvement of a 2D stochastic model based handwritten entity recognition system is described. To model the handwriting considered as being a two dimensional signal, a context based, segmentation-free Hidden Markov Model (HMM) recognition system was used. The baseline approach combines a Markov Random Field (MRF) and a HMM so-called Non-Symmetric Half Plane Hidden Markov Model (NSHP-HMM). To improve the results performed by this baseline system operating just on low-level pixel information an extension of the NSHP-HMM is proposed. The mechanism allows to extend the observations of the NSHP-HMM by implanting structural information in the system. At present, the accuracy of the system on the SRTP1 French postal check database is 87.52% while for the handwritten Bangla city names is 86.80%. The gain using this structural information for the SRTP dataset is 1.57%.

show abstract

Section: Systemmentioning

confidence: 99%

Structural information implant in a context based segmentation-free HMM handwritten word recognition system for Latin and Bangla script

Vajda

Belaïd

2005

Eighth International Conference on Document Analysis and Recognition (ICDAR'05)

View full text Add to dashboard Cite

show abstract

“…There are also other segmentation-free approaches such as the holistic approach [5][6] [7] and HMM approaches [8][9] [10]. Holistic word matching treats the word as a single, indivisible entity and attempts to recognize it using features of the word as a whole [5] [6].…”

Section: Introductionmentioning

confidence: 99%

Top-Down Likelihood Word Image Generation Model for Holistic Word Recognition

Ishidera¹,

Lucas

Downton

2002

Lecture Notes in Computer Science

View full text Add to dashboard Cite

This paper describes a new top-down word image generation model for word recognition. This model can generate a word image with a likelihood based on linguistic knowledge, segmentation and character image. In the recognition process, first, the model generates the word image which approximates an input image best for each of a dictionary of possible words. Next, the model calculates the distance value between the input image and each generated word image. Thus, the proposed method is a type of holistic word recognition method. The effectiveness of the proposed method was evaluated in an experiment using typewritten museum archive card images. The difference between a non-holistic method and the proposed method is shown by the evaluation. The small errors accumulate in non-holistic methods during the process carried out, because the non-holistic methods can't cover the whole word image but only part images extracted by segmentation, and the non-holistic method can't eliminate the black pixels intruding in the recognition window from neighboring characters. In the proposed method, we can expect that no such errors will accumulate. Results show that a recognition rate of 99.8% was obtained, compared with only 89.4% for a recently published comparator algorithm.

show abstract

“…Finally, the emergence of neural networks (NNs) as universal approximators and their natural ability to introduce explicitly low-level context, led to their being coupled with HMMs in SR [7]. This resulted in robust "hybrid systems" for sequence recognition [8], then successfully used in handwriting [9][10][11][12][13][14]. In such systems, NNs are in fact used for classification, that is, to produce posterior probabilities of belonging to a given class (often letters, sometimes pseudo-letters), when a portion of the word is given as input.…”

Section: Introductionmentioning

confidence: 99%

“…Finally, unlike several hybrid systems, this approach permits the direct estimation of the emission probabilities in the HMM, that is, with neither extra estimation nor approximation. Indeed, the classificatory approach in hybrid systems leads naturally either to the estimation of prior state probabilities to compute approximative scaled likelihoods via Bayes' rule [10,11,13,14], or to introduce a specific renormalization method in order to improve the approximation of probabilities made by the NN's scores [12].…”

Section: Introductionmentioning

confidence: 99%

Maximum mutual information training for an online neural predictive handwritten word recognition system

Garcia-Salicetti

Dorizzi²,

Gallinari

et al. 2001

IJDAR

View full text Add to dashboard Cite

In this paper, we present a hybrid online handwriting recognition system based on hidden Markov models (HMMs). It is devoted to word recognition using large vocabularies. An adaptive segmentation of words into letters is integrated with recognition, and is at the heart of the training phase. A word-model is a left-right HMM in which each state is a predictive multilayer perceptron that performs local regression on the drawing (i.e., the written word) relying on a context of observations. A discriminative training paradigm related to maximum mutual information is used, and its potential is shown on a database of 9,781 words.

show abstract

A hybrid radial basis function network/hidden Markov model handwritten word recognition system

Cited by 13 publications

References 12 publications

Structural information implant in a context based segmentation-free HMM handwritten word recognition system for Latin and Bangla script

Structural information implant in a context based segmentation-free HMM handwritten word recognition system for Latin and Bangla script

Top-Down Likelihood Word Image Generation Model for Holistic Word Recognition

Maximum mutual information training for an online neural predictive handwritten word recognition system

Contact Info

Product

Resources

About