Rethinking CNN-Based Pansharpening: Guided Colorization of Panchromatic Images via GANs

TL;DR

This paper introduces PanColorGAN, a novel self-supervised GAN-based framework that treats satellite image pansharpening as a colorization task, improving detail and reducing blur compared to existing CNN methods.

Contribution

It proposes a new perspective by framing pansharpening as colorization, with a self-supervised learning approach and noise injection for better generalization.

Findings

Outperforms previous CNN-based methods in quality.

Reduces spatial detail loss and blur artifacts.

Demonstrates robustness to varying downsampling ratios.

Abstract

Convolutional Neural Networks (CNN)-based approaches have shown promising results in pansharpening of satellite images in recent years. However, they still exhibit limitations in producing high-quality pansharpening outputs. To that end, we propose a new self-supervised learning framework, where we treat pansharpening as a colorization problem, which brings an entirely novel perspective and solution to the problem compared to existing methods that base their solution solely on producing a super-resolution version of the multispectral image. Whereas CNN-based methods provide a reduced resolution panchromatic image as input to their model along with reduced resolution multispectral images, hence learn to increase their resolution together, we instead provide the grayscale transformed multispectral image as input, and train our model to learn the colorization of the grayscale input. We further address the fixed downscale ratio assumption during training, which does not generalize well to the full-resolution scenario. We introduce a noise injection into the training by randomly varying the downsampling ratios. Those two critical changes, along with the addition of adversarial training in the proposed PanColorization Generative Adversarial Networks (PanColorGAN) framework, help overcome the spatial detail loss and blur problems that are observed in CNN-based pansharpening. The proposed approach outperforms the previous CNN-based and traditional methods as demonstrated in our experiments.