Enhancing Sentinel-2 Image Resolution: Evaluating Advanced Techniques based on Convolutional and Generative Neural Networks

TL;DR

This paper investigates the enhancement of spatial resolution in Sentinel-2 bands that contain spectral information using advanced super-resolution techniques by a factor of 2.5, and finds that while CNN-based approaches produce satisfactory outcomes, they tend to yield blurry images.

Abstract

This paper investigates the enhancement of spatial resolution in Sentinel-2 bands that contain spectral information using advanced super-resolution techniques by a factor of 2. State-of-the-art CNN models are compared with enhanced GAN approaches in terms of quality and feasibility. Therefore, a representative dataset comprising Sentinel-2 low-resolution images and corresponding high-resolution aerial orthophotos is required. Literature study reveals no feasible dataset for the land type of interest (forests), for which reason an adequate dataset had to be generated in addition, accounting for accurate alignment and image source optimization. The results reveal that while CNN-based approaches produce satisfactory outcomes, they tend to yield blurry images. In contrast, GAN-based models not only provide clear and detailed images, but also demonstrate superior performance in terms of quantitative assessment, underlying the potential of the framework beyond the specific land type investigated.

Figures (5)

• Figure 1: Aerial orthophotos of Carinthia as provided by KAGIS, highlighting the areas of interest.
• Figure 2: Visual representation of the histogram matching process.
• Figure 3: Generator architecture described in Wang2018 with $N_{\text{RRDB}} = 4$ and $N_{\text{UB}} = 1$.
• Figure 4: Discriminator Architectures: (a) ESRGAN proposed Wang2018 in with classic blocks, (b) Real-ESRGAN proposed Wang2021 with U-Net structure.
• Figure 5: Visual comparison of representative patches: Baseline vs. GAN models inclusive PSNR / SSIM / LPIPS metric values.