Core Concepts
Blind image super-resolution is enhanced by integrating depth information and spatially variant blur kernels through the SSR framework.
Abstract
The content introduces the SSR framework for blind image super-resolution, emphasizing the importance of depth information and spatially variant blur kernels. It discusses the challenges faced by existing diffusion-based methodologies and presents the Adaptive Multi-Modal Fusion (AMF) module to align information from different modalities. The framework's effectiveness is validated through quantitative and qualitative experiments on various datasets.
Stats
Pre-trained diffusion models encapsulate a substantial reservoir of a priori knowledge.
SVKR estimates a Depth-Informed Kernel, enhancing depth information accuracy.
AMF aligns information from LR images, depth maps, and blur kernels.
Quotes
"Quantitative and qualitative experiments affirm the superiority of our approach."
"To address these challenges, we introduce the SSR framework."