Academic Journals Database
Disseminating quality controlled scientific knowledge

Comparison of Two Feature Extraction Method Based on the Raw Form and Skeleton for MNIST Database

Author(s): Kamal Moro | Mohammed Fakir | Badr Dine El Kessab | Belaid Bouikhalene | Cherki Daoui

Journal: International Journal of Computer Science Research and Application
ISSN 2012-9564

Volume: 02;
Issue: 03;
Start page: 13;
Date: 2012;
VIEW PDF   PDF DOWNLOAD PDF   Download PDF Original page

Keywords: Optical character recognition | neural network | feature extraction | MNIST database | skeletonization

In this paper, we present a comparison of two few extraction methods based on the raw form and the skeleton. The experimental results show the benefit of using these methods compared with other conventional approaches; the findings have been applied on the MNIST database. The paper presents the evaluation work for the comparison of the two methods and provides conclusions based on the study outcomes.
RPA Switzerland

RPA Switzerland

Robotic process automation


Tango Jona
Tangokurs Rapperswil-Jona