Research Article
Implementation of a CNN for Asterism Classification in Carte du Ciel Astrographic Maps
@INPROCEEDINGS{10.1007/978-3-031-44668-9_8, author={Lasko M. Laskov and Radoslav Radev}, title={Implementation of a CNN for Asterism Classification in Carte du Ciel Astrographic Maps}, proceedings={Computer Science and Education in Computer Science. 19th EAI International Conference, CSECS 2023, Boston, MA, USA, June 28--29, 2023, Proceedings}, proceedings_a={CSECS}, year={2023}, month={10}, keywords={Deep learning Convolutional neural network Astrographic maps}, doi={10.1007/978-3-031-44668-9_8} }- Lasko M. Laskov
Radoslav Radev
Year: 2023
Implementation of a CNN for Asterism Classification in Carte du Ciel Astrographic Maps
CSECS
Springer
DOI: 10.1007/978-3-031-44668-9_8
Abstract
Carte du Ciel together with Astrographic Catalogue form a 19th century huge international astronomical project whose goal was to map the stars in the visible sky as faint as 14th magnitude. The result, in the form of astrographic plates and their paper copies – astrographic maps, are stored and investigated in many astronomy institutes worldwide.
The goal of our study is to develop image processing and pattern recognition techniques for automatic extraction of astronomical data from the digitized copies of the astrographic maps. In this paper we present the design and implementation of a convolutional neural network (CNN) for automatic classification of stars images in scanned Carte du Ciel astrographic maps. We do not use any deep learning frameworks to build our model, and we focus on the low-level implementation of the CNN. Also, we provide comparison of our implementation with an implementation based on PyTorch.
