Recognition Method Of Xiaozhuan In Wazhou Based On Neural Network

With the rapid development of printed character recognition in recent years, the entire field of character recognition has made a lot of progress. Online handwriting recognition has achieved higher recognition accuracy. And the off-line handwriting recognition system is still facing limitations caused by unique identification samples. With the research topic of the Design and Development of Recognition System on Characters of Seal Script and Classification of Eaves Tile from Qin and Han Dynasty undertaking by applications of virtual technology. This article aimed at the research of Seal Script characters on ancient eaves tiles and using the computers to study the characters recognition problem to increase the accuracy. This article planned to produce positive effect for the digitized protection of eaves tiles culture hermitage.This article studied and actualized the whole process of character recognition of Seal Script characters on archaic eaves tiles basing on the automatic recognition of archaic script in digital images of eaves tiles and a combination of special structures of Chinese characters.Mainly including three parts:image pre-treatment, feature extraction, and classifier design1. Considering the stationary of the structure of eaves tiles, this article proposed a program of eaves tiles area elimination algorithm. Experiments proved that this algorithm could reach comparatively quick and accurate results.2. In addition, considering the synonymy abnormality of character scripts, this article proposed a program of neighbor mean sample grouping algorithm. Problem can be solved through the analysis of sample data.3. With the focus on experimental recognition of Seal Script characters, it was proved that the neural network combining with neighbor mean sample grouping algorithm could be useful to the identification of archaic script in eaves tiles. In the case of limited sample data, the recognition rate could reach86.9%.