Research Article

Generative Adversarial Network-Based Edge-Preserving Superresolution Reconstruction of Infrared Images

Table 1

Method comparison: quantitative results of the SRCNN, ESPCN, SRGAN, ESRGAN, and ESRGAN+ methods and the proposed method on the FLIR_ ADAS_1_3 validation dataset.

CCPSNR (dB)SSIMVIFUIQITIME (s)

SRCNN0.816018.39630.86550.79420.87702.7112
ESPCN0.862727.71710.93490.82690.93981.0206
SRGAN0.846328.04450.94970.81200.85030.9202
ESRGAN0.954728.54530.95060.79180.93820.1093
ESRGAN+0.980633.70180.98550.91190.98640.2713
Ours0.996437.98650.99800.90830.99650.1145