Super Resolution with Deep Learning in Thermal Images Termal Görüntülerde Derin Öğrenme ile Süper Çözünürlük

Öz M. C., NAVRUZ T. S.

32nd IEEE Conference on Signal Processing and Communications Applications, SIU 2024, Mersin, Turkey, 15 - 18 May 2024 identifier

  • Publication Type: Conference Paper / Full Text
  • Doi Number: 10.1109/siu61531.2024.10601089
  • City: Mersin
  • Country: Turkey
  • Keywords: Deep Learning, ESRGAN, GAN, Scaling, SRGAN, Super Resolution, Thermal Image
  • Gazi University Affiliated: Yes

Abstract

Although traditional methods and deep learning-based methods have been used for color images in super-resolution applications for a long time, these methods have only recently begun to be used for thermal images. The fact that thermal sensors are generally low resolution and high resolution thermal sensors are very expensive has led to the need to increase the resolution in thermal images by using deep learning. In this study, the effect of bicubic interpolation, one of the traditional approaches, and SRGAN and ESRGAN architectures, one of the deep learning-based methods, on super resolution on thermal images were examined.