Abstract | ||
---|---|---|
The recognition of handwritten characters, words, and text arouses great interest today. To develop the best working system is subject of many papers published. With this paper, methods to improve the performance of existing word recognition systems are discussed. The availability of a sufficient data sets for training and testing the system assumed, optimization algorithms are presented. The usage of different feature sets and the combination of different recognizers are proposed. Tests with Arabic handwriting recognition systems using the reference IfN/ENIT-database show the usefulness of the proposed methods. An improvement of the recognition rate of up to 28% of the best single system is achieved. |
Year | DOI | Venue |
---|---|---|
2009 | 10.1109/ICDAR.2009.11 | ICDAR-1 |
Keywords | Field | DocType |
word recognition,text analysis,pixel,hidden markov models,database system,handwriting recognition,system testing,availability,feature extraction,noise reduction,communications technology | Data set,Computer science,Handwriting recognition,Artificial intelligence,Natural language processing,Intelligent word recognition,Intelligent character recognition,Pattern recognition,System testing,Word recognition,Feature extraction,Speech recognition,Hidden Markov model | Conference |
ISSN | ISBN | Citations |
1520-5363 E-ISBN : 978-0-7695-3725-2 | 978-0-7695-3725-2 | 3 |
PageRank | References | Authors |
0.39 | 12 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Haikal El-Abed | 1 | 436 | 29.39 |
Volker Märgner | 2 | 295 | 29.02 |
Haikal El-Abed | 3 | 256 | 12.91 |