Object focused method improves satellite image resolution and detail
MOSAIC: Multi-Scale Object-Centric Slot Adaptation with Image-Guided Prototype Conditioning for Remote Sensing Image Super-Resolution
Computer Vision and Pattern Recognition
Summary
Improving blurry satellite images is hard because small objects and repeated patterns can get lost. The authors propose MOSAIC, a new way to break down images into meaningful parts that represent objects and layouts at different sizes. This helps a powerful image-enhancing AI understand and restore details better. When tested on several remote sensing image sets, MOSAIC showed clearer images and better object preservation than previous methods.
What this means in practice
- •For remote sensing analysts: Enhance low-resolution satellite images to better identify small and repeated structures for improved geographical analysis.
- •For urban planning teams: Create sharper high-resolution aerial maps that preserve buildings and layout details for accurate city planning and management.
Authors
Liuqian Wang, Hanyu Xie, Rongzhen Zhao, Qi Ming, Chunjie Ma, Wentao Chao, Yufei Gao, Chengming Liu, Lei Shi
Abstract
Pretrained diffusion Transformers provide powerful generative priors for image restoration, yet adapting them to structure-sensitive tasks remains challenging. Existing methods mainly rely on dense low-resolution features as conditions, which often entangle salient objects with dominant backgrounds and provide limited abstraction for recurring structures across scales and samples. We propose MOSAIC, an object-centric representation learning framework that learns compact structural conditions for generative model adaptation. MOSAIC organizes degraded observations into structural slots that capture complementary object-level and layout-level evidence, integrates them across resolutions, and retrieves recurring priors from a shared prototype memory to complement incomplete image-specific information. The resulting representations are injected into selected layers of a frozen diffusion Transformer through lightweight residual modulation, preserving its pretrained generative prior. We instantiate MOSAIC for remote sensing image super-resolution, where small objects, repetitive structures, and large-scale variation provide a challenging testbed. Experiments on UCMerced, AID, and RSSCN7 show consistent PSNR improvements across $\times 2$ and $\times 4$ settings and favorable LPIPS results at $\times 4$. Qualitative results further demonstrate improved preservation of small objects, structural boundaries, and repetitive spatial patterns.