Publication Details
Využití PERO OCR při přepisu rukopisů
Hradiš Michal, Ing., Ph.D. (DCGM FIT BUT)
Žabička Petr, Ing. (MZK)
Kohút Jan, Ing. (DCGM FIT BUT)
Kišš Martin, Ing. (DCGM FIT BUT)
Beneš Karel, Ing. (DCGM FIT BUT)
Manuscripts, Kurrent, handwritten text recognition, PERO OCR, case study
The basic prerequisite for the automatic processing of handwritten documents is a high-quality transcription of their content. This paper presents the automatic document transcription tools developed in the PERO project, together with a web application for efficient manual corrections. A case study dealing with the use of these tools has shown significant time savings in creating transcriptions of older manuscripts without negatively affecting the quality of these transcriptions. At the same time, an analysis of the largest documents processed and corrected by the users of the PERO-OCR web application over the last two years shows that the error rate for well-legible Czech manuscripts ranges from 1 % to 4 %, which may be sufficient for numerous purposes.
@ARTICLE{FITPUB12914, author = "Martina Dvo\v{r}\'{a}kov\'{a} and Michal Hradi\v{s} and Petr \v{Z}abi\v{c}ka and Jan Koh\'{u}t and Martin Ki\v{s}\v{s} and Karel Bene\v{s}", title = "Vyu\v{z}it\'{i} PERO OCR p\v{r}i p\v{r}episu rukopis\r{u}", pages = "14--27", booktitle = "Archivn\'{i} \v{c}asopis", journal = "Archivn\'{i} \v{c}asopis", volume = 72, number = 1, year = 2022, location = "Praha, CZ", ISSN = "0004-0398", language = "czech", url = "https://www.fit.vut.cz/research/publication/12914" }