Using a neural network for industrial character recognition
The pattern classification abilities of neural networks make them suitable for practical image recognition tasks such as industrial character recognition. In this thesis, backpropagation trained multi-layer networks applied to recognition of IC characters are investigated with the aim of ascertainin...
Saved in:
Main Author: | |
---|---|
Format: | text |
Language: | English |
Published: |
Animo Repository
1993
|
Subjects: | |
Online Access: | https://animorepository.dlsu.edu.ph/etd_masteral/1519 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Institution: | De La Salle University |
Language: | English |
Summary: | The pattern classification abilities of neural networks make them suitable for practical image recognition tasks such as industrial character recognition. In this thesis, backpropagation trained multi-layer networks applied to recognition of IC characters are investigated with the aim of ascertaining the network sizes that are suitable for both rotated and unrotated characters, and the performance of these networks with untrained font types. To avoid a huge combinatorial explosion of possibilities to explore, a single method for preprocessing and representing character data was used for all the networks. Characters also consisted only of digits to limit training time. A significant feature in all the training sessions was the exclusion of actual IC character images in the training sets. This was to support the objective of determining the extent of font type invariance of backpropagation networks. Despite this, 100 percent recognition of the test ICs was still possible in one case. Lastly, it is emphasized that the results of the investigation are conclusive within the parameters of this investigation. |
---|