This paper proposes using computationally efficient Consistency Models (CMs) for super-resolution of low-quality satellite images, achieving significant speed improvements and enhanced image quality compared to traditional Denoising Diffusion Probabilistic Models (DDPMs).
불규칙적으로 샘플링된 Sentinel-2 시계열 데이터를 활용하여 고해상도 영상을 복원하는 방법을 제안한다.