DocumentCode
1616848
Title
Post code digit recognition employing low-level features
Author
Jones, Peter E.
Author_Institution
Dept. of Electr. & Electron. Eng., Western Australia Univ., Nedlands, WA, Australia
Volume
2
fYear
1997
Firstpage
523
Abstract
Australian postcodes use four digits. This paper describes the recognition of handwritten postcodes on Australia Post Office approved envelopes using low-level features extracted from the scanned image. World-wide efforts have steadily improved recognition rates over the years. It has now become more difficult to significantly improve them. We have therefore been examining low-level features in the digitised image for evidence left behind from the dynamics of the writing process and for evidence from people´s individual styles of forming digits. In many of our samples we find what we have named “tags” that provide useful clues as to what the character may be. For instance, our classifier used in earlier work would often confuse the digits “6” and “0”. This more recent work shows that often a tag is formed when writing a “6” and this can be used to choose the “6” over the “0”
Keywords
feature extraction; image classification; optical character recognition; postal services; Australia Post Office approved envelopes; Australian postcodes; digitised image; digits; handwritten postcodes; low-level features; post code digit recognition; scanned image; tags; writing process; Australia; Character recognition; Handwriting recognition; History; Image recognition; Neural networks; Optical character recognition software; Postal services; Shape control; Writing;
fLanguage
English
Publisher
ieee
Conference_Titel
TENCON '97. IEEE Region 10 Annual Conference. Speech and Image Technologies for Computing and Telecommunications., Proceedings of IEEE
Conference_Location
Brisbane, Qld.
Print_ISBN
0-7803-4365-4
Type
conf
DOI
10.1109/TENCON.1997.648259
Filename
648259
Link To Document