Bayesian classification (AutoClass): theory and results
Advances in knowledge discovery and data mining
A Database for Handwritten Text Recognition Research
IEEE Transactions on Pattern Analysis and Machine Intelligence
ICDAR '01 Proceedings of the Sixth International Conference on Document Analysis and Recognition
A bayesian approach to temporal data clustering using the hidden markov model methodology
A bayesian approach to temporal data clustering using the hidden markov model methodology
Model Length Adaptation of an HMM based Cursive Word Recognition System
ICDAR '03 Proceedings of the Seventh International Conference on Document Analysis and Recognition - Volume 1
Traffic analysis attacks on Skype VoIP calls
Computer Communications
Hi-index | 0.00 |
An important parameter for building a cursive scriptmodel is the number of different, relevant letter writing variants.An algorithm performing this task automatically byoptimizing the number of letter models in an HMM-basedscript recognition system is presented.The algorithm iterativelymodifies selected letter models; for selection, qualitymeasures like HMM distance and emission weight entropyare developed, and their correlation with recognitionperformance is shown.Theoretical measures for the selectionof overall model complexity are presented, but best resultsare obtained by direct selection criteria: likelihoodand recognition rate of training data.With the optimizedmodels, an average improvement in recognition rate of upto 5.8 percent could be achieved.