Fusion of Images of Different Spectra Based on Generative Adversarial Networks
- Authors: Vizil’ter Y.V.1, Vygolov O.V.1, Komarov D.V.1, Lebedev M.A.1
-
Affiliations:
- State Scientific Research Institute of Aviation Systems (GosNIIAS)
- Issue: Vol 58, No 3 (2019)
- Pages: 441-453
- Section: Pattern Recognition and Image Processing
- URL: https://journals.rcsi.science/1064-2307/article/view/220389
- DOI: https://doi.org/10.1134/S1064230719030201
- ID: 220389
Cite item
Abstract
A method for fusing images of different spectra by using generative adversarial networks is proposed. An original architecture of a FusionNet neural network is developed based on pix2pix. It enables the synthesis of a complex (integrated) image that comprises the most informative fragments of different-spectra images, thus being more informative than any of these individual images. A technique for generating training and test sets, as well as the process of data augmentation, is described. The operation of the proposed image fusion method is demonstrated on some real-world infrared and visible images.
About the authors
Yu. V. Vizil’ter
State Scientific Research Institute of Aviation Systems (GosNIIAS)
Email: MLebedev@gosniias.ru
Russian Federation, Moscow, 125319
O. V. Vygolov
State Scientific Research Institute of Aviation Systems (GosNIIAS)
Email: MLebedev@gosniias.ru
Russian Federation, Moscow, 125319
D. V. Komarov
State Scientific Research Institute of Aviation Systems (GosNIIAS)
Email: MLebedev@gosniias.ru
Russian Federation, Moscow, 125319
M. A. Lebedev
State Scientific Research Institute of Aviation Systems (GosNIIAS)
Author for correspondence.
Email: MLebedev@gosniias.ru
Russian Federation, Moscow, 125319