KAO: Kernel-Adaptive Optimization in Diffusion for Satellite Image

TL;DR

KAO introduces a kernel-adaptive, diffusion-based framework with latent space conditioning and explicit propagation, achieving efficient and high-quality satellite image inpainting for very high-resolution datasets.

Contribution

The paper presents KAO, a novel satellite image inpainting method that combines kernel-adaptive optimization with latent space conditioning and explicit propagation, improving efficiency and accuracy.

Findings

Sets new benchmarks on DeepGlobe and Massachusetts Roads datasets.

Balances efficiency of preconditioned models with flexibility of postconditioned models.

Demonstrates superior inpainting quality and stability.

Abstract

Satellite image inpainting is a crucial task in remote sensing, where accurately restoring missing or occluded regions is essential for robust image analysis. In this paper, we propose KAO, a novel framework that utilizes Kernel-Adaptive Optimization within diffusion models for satellite image inpainting. KAO is specifically designed to address the challenges posed by very high-resolution (VHR) satellite datasets, such as DeepGlobe and the Massachusetts Roads Dataset. Unlike existing methods that rely on preconditioned models requiring extensive retraining or postconditioned models with significant computational overhead, KAO introduces a Latent Space Conditioning approach, optimizing a compact latent space to achieve efficient and accurate inpainting. Furthermore, we incorporate Explicit Propagation into the diffusion process, facilitating forward-backward fusion, which improves the stability and precision of the method. Experimental results demonstrate that KAO sets a new benchmark for VHR satellite image restoration, providing a scalable, high-performance solution that balances the efficiency of preconditioned models with the flexibility of postconditioned models.