Modified SWIN-IR Based Super Resolution Approach for Thermal Imagery


Ekşi E., Taşkıran M.

2024 15th National Conference on Electrical and Electronics Engineering (ELECO), Bursa, Türkiye, 28 - 30 Kasım 2024, ss.1-5

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.1109/eleco64362.2024.10847221
  • Basıldığı Şehir: Bursa
  • Basıldığı Ülke: Türkiye
  • Sayfa Sayıları: ss.1-5
  • Yıldız Teknik Üniversitesi Adresli: Evet

Özet

While RGB cameras are inadequate in low light conditions, thermal cameras offer a solution. However, since the resolution of thermal images is low, super-resolution methods are used to improve these images. Due to the large size and high cost of highresolution thermal camera sensors, super-resolution methods are highly preferred especially in compact security camera solutions. In this study, a modified Swin image restoration (SwinIR) approach using a modified Swin transform is proposed for super resolution in thermal images. In the study, the modified SwinIR algorithm is compared with other solutions. The results show that the proposed modified SwinIR approach outperforms the other methods. The success of the model is evaluated by PSNR (Peak SignaltoNoise Ratio) and SSIM (Structural Similarity Index) metrics. The PSNR value of the proposed approach was calculated as 33.56 and the SSIM value was calculated as 0.9145, thus providing the best result.