Recognition of the cursive handwritten courtesy amounts of Malaysian bank cheques.

In this paper, we present techniques to recognise the cursive handwritten courtesy amount from Malaysian bank cheques and convert it into its corresponding ASCII codes. The image is pre- processed and segmented using the combination of crossed-point, curved-point and histogram techniques. Three t...

Full description

Bibliographic Details
Main Authors: Sulaiman, Md. Nasir, Khalid, Marzuki
Format: Conference or Workshop Item
Language:English
English
Published: 2001
Subjects:
Online Access:http://psasir.upm.edu.my/id/eprint/20832/
http://psasir.upm.edu.my/id/eprint/20832/1/ID%2020832.pdf
Description
Summary:In this paper, we present techniques to recognise the cursive handwritten courtesy amount from Malaysian bank cheques and convert it into its corresponding ASCII codes. The image is pre- processed and segmented using the combination of crossed-point, curved-point and histogram techniques. Three type of feature extraction methods which are transitions, shapes and moment invariant are used to extract local and global features of segmented handwritten digits. The main purpose of the feature extraction is to capture the most relevant and discriminant characteristics of the object. A three layer Neural Network architecture with the new error function of backpropagation learning algorithm is used as recogniser. This approach yields good recognition results with faster convergence rates.