A Handwritten Character Recognition System Using Directional Element Feature and Asymmetric Mahalanobis Distance
Issue No. 03 - March (1999 vol. 21)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/34.754617
<p><b>Abstract</b>—This paper presents a precise system for handwritten Chinese and Japanese character recognition. Before extracting <it>directional element feature</it> (DEF) from each character image, <it>transformation based on partial inclination detection</it> (TPID) is used to reduce undesired effects of degraded images. In the recognition process, <it>city block distance with deviation</it> (CBDD) and <it>asymmetric Mahalanobis distance</it> (AMD) are proposed for rough classification and fine classification. With this recognition system, the experimental result of the database ETL9B reaches to 99.42%.</p>
Handwritten Chinese and Japanese character recognition, directional element feature, city block distance with deviation, asymmetric Mahalanobis distance, ETL9B.
Nei Kato, Shin'ichiro Omachi, Yoshiaki Nemoto, Masato Suzuki, Hirotomo Aso, "A Handwritten Character Recognition System Using Directional Element Feature and Asymmetric Mahalanobis Distance", IEEE Transactions on Pattern Analysis & Machine Intelligence, vol. 21, no. , pp. 258-262, March 1999, doi:10.1109/34.754617