The Classification of Galaxy Images Using Neural Network Algorithm
Marjan Kuchaki Rafsanjani, Hashem Bagherinezhad, Ranbir Singh Batth · 2019 International Conference on Computing, Power and Communication Technologies (GUCON) · 2019
The observation of the sky objects helps the astronomers to understand how the world is shaped. Due to the large number of the objects observed by the modern telescopes, it is very difficult to analyze them manually. An important part of the galaxies research is the classification based on Hubble's design. The purpose of this paper is the classification of the galaxy images using the neural network. Due to Hubble's design, the galaxies are divided into the regular galaxies and the irregular galaxies. The regular galaxies are divided into two groups (the spiral and elliptical galaxies). The spiral galaxies can be regarded as the elliptical galaxies or the circular galaxies (mistakenly), thus the galaxies classification is important. The proposed method uses the sdss dataset which contains 570 images. In the first phase, the pre-processing is performed which it contains the noise removal. In the next phase, the feature extraction is performed. We extract 827 features using the sub-windows, the different color spaces moments and the features of the local configuration patterns. In the third phase, the classification is performed by using the neural network. The proposed method is implemented in MATLAB and its accuracy and execution time is evaluated. Also, in this study, the proposed method is compared with similar methods which the results show that the proposed method has the better performance.