HAL will be down for maintenance from Friday, June 10 at 4pm through Monday, June 13 at 9am. More information
Skip to Main content Skip to Navigation
Conference papers

Damaged Character Pattern Recognition on Wooden Tablets Excavated from The Heijyo Palace Site

Abstract : This paper describes damaged character pattern recognition on wooden tablets excavated from the Heijyo palace site (the ancient palace in the Nara period from AD. 710 to 794). Since most of excavated tablets have been stained, damaged, and sometimes broken into pieces, it is extremely difficult even for archaeologists to read characters from badly blurred or missing ink on tablets. The aim of the character recognition is to output candidates even for degraded or partially lost character patterns in order to give hints to human expert readers rather than to read handwritten characters at the maximum speed. We propose a method that applies nonlinear normalization and feature extraction for ternary character pattern images with gray missing area supplemented by readers. The proposed method realizes 60.2% as the 10th cumulative rate for 2,108 character patterns extracted from wooden tablets with partial damages applied artificially. This result is better than the result of the previous method.
Complete list of metadata

Cited literature [8 references]  Display  Hide  Download

https://hal.inria.fr/inria-00112657
Contributor : Anne Jaigu Connect in order to contact the contributor
Submitted on : Thursday, November 9, 2006 - 2:42:50 PM
Last modification on : Wednesday, February 10, 2021 - 1:12:01 PM
Long-term archiving on: : Tuesday, April 6, 2010 - 10:01:56 PM

Identifiers

  • HAL Id : inria-00112657, version 1

Collections

Citation

Masaki Nakagawa, Kei Saito, Akihito Kitadai, Junko Tokuno, Hajime Baba, et al.. Damaged Character Pattern Recognition on Wooden Tablets Excavated from The Heijyo Palace Site. Tenth International Workshop on Frontiers in Handwriting Recognition, Université de Rennes 1, Oct 2006, La Baule (France). ⟨inria-00112657⟩

Share

Metrics

Record views

95

Files downloads

62