Reject Management in a Handwriting Recognition System
ICDAR '97 Proceedings of the 4th International Conference on Document Analysis and Recognition
Continuous Learning Systems Postal Address Readers with Built-In Learning Capability
ICDAR '99 Proceedings of the Fifth International Conference on Document Analysis and Recognition
Impact of Lexicon Completeness on City Name Recognition
IWFHR '02 Proceedings of the Eighth International Workshop on Frontiers in Handwriting Recognition (IWFHR'02)
Confidence Modeling for Verification Post-Processing for Handwriting Recognition
IWFHR '02 Proceedings of the Eighth International Workshop on Frontiers in Handwriting Recognition (IWFHR'02)
Rejection Measures for Handwriting Sentence Recognition
IWFHR '02 Proceedings of the Eighth International Workshop on Frontiers in Handwriting Recognition (IWFHR'02)
Adaptation of an Address Reading System to Local Mail Streams
ICDAR '01 Proceedings of the Sixth International Conference on Document Analysis and Recognition
Rejection Strategies Involving Classifier Combination for Handwriting Recognition
IbPRIA '07 Proceedings of the 3rd Iberian conference on Pattern Recognition and Image Analysis, Part II
Handwritten word-spotting using hidden Markov models and universal vocabularies
Pattern Recognition
Handwriting segmentation of Arabic text
SPPRA '08 Proceedings of the Fifth IASTED International Conference on Signal Processing, Pattern Recognition and Applications
Hi-index | 0.00 |
In this paper the performance of different confidencemeasures used for an address recognition system are evaluated.The recognition system for cursive handwritten Germanaddress words is based on Hidden Markov Models(HMMs). It is essential, that the structure of the address(name, street, city, country) is known, so that a specificsmall but complete dictionary can be selected. Choosinga wrong dictionary (OOV: out-of-vocabulary) or misrecognizethe word, the recognition result should be rejected bymeans of the confidence measure. This paper points out twoaspects: the comparison of four confidence measures forsingle words - based on the likelihood, a garbage-model,a two-best recognition or a character decoding - and thecomparison of using complete or wrong dictionaries. It isshown, that the best confidence measure - the two-best distance- has a quite different behavior using OOV.