Handwritten Amharic text Recognition Applied to the Processing of Bank Cheques
No Thumbnail Available
Date
2000-05
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
The advent and wide application of Information Technology (IT) has brought changes on
the way we perform things. One area where IT particularly computers has showed
significant improvement in efficiency and effectiveness of activities is by handling routine,
repetitive, boring and error prone tasks. Among such tasks bank check reading is the one
that can be handled by computers in a stable manner using the OCR technology.
In this piece of work an attempt was made to adopt a recognition algorithm for
handwritten Amharic text. To achieve this target some preprocessing. segmentation and
normalization techniques implemented for other scripts were reviewed. The Feature oj
hand printed Amharic text in general and hand printed legal amounts in particular were
studied. Based on the identified Features some of the preprocessing. segmentation and
normalization algorithms were implemented and tested on a set of hand printed data
collected using sample checks. In addition, different feed forward neural networks with
back-propagation learning algorithm were implemented and the test results as well as the
procedure of the test are recorded. Finally recommendations are forwarded pointing out
some points for further consideration.
Description
Keywords
Information Science