Bouma, S. and Buck, W.E.S.M. and Moelker, R.R. (2011) Vector-attribute character recognition. Bachelor's Thesis, Computing Science.
![[img]](https://fse.studenttheses.ub.rug.nl/style/images/fileicons/text.png) |
Text
Akkoord.PDF - Other Restricted to Registered users only Download (22kB) |
|
|
Text
INF-BA-2011-S.Bouma-W.Buck-R.Moelker.pdf - Published Version Download (3MB) | Preview |
Abstract
Traditional optical character recognition (OCR) uses thresholding and pre- defined characters to recognize text. This study makes the assumption that thresholding can cause loss of potentially valuable information. A more ver- satile approach to character recognition with minimal thresholding is offered which relies on a max-tree and image moments. The max-tree is a tree repre- sentation of the image where each level represents a grey-level and each node contains attributes. The image moments are scale and translational invariant, this allows recognition of a wide range of characters instances. Finally super- vised learning is used to classify the characters, and line and word segmentation is used to obtain a full textual reconstruction of an input image.
| Item Type: | Thesis (Bachelor's Thesis) |
|---|---|
| Degree programme: | Computing Science |
| Thesis type: | Bachelor's Thesis |
| Language: | English |
| Date Deposited: | 15 Feb 2018 07:46 |
| Last Modified: | 15 Feb 2018 07:46 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/9727 |
Actions (login required)
 |
View Item |