LLPR: Location-aware learning and physics-based reconstruction for raindrop removal from a single image

2026-09-24

AI Summary

This study addresses the challenges of occlusion-induced background distortion and high computational complexity in single-image raindrop removal. To this end, we propose an efficient deraining framework that integrates position-aware learning with a physical reconstruction model. Specifically, a removable position-aware branch is designed to be seamlessly embedded into CNN or Transformer architectures at zero additional inference cost. Concurrently, the latent background layer is inversely derived from a physical imaging model to achieve high-fidelity recovery of clear backgrounds. Extensive experiments demonstrate that the proposed method surpasses existing state-of-the-art approaches on real-world datasets, yielding significant improvements in both deraining quality and cross-scenario generalization capability.

Abstract

Raindrops can cause occlusion and distortion in the background scenes due to their adherence to windows or camera lenses. Existing raindrop removal methods concentrate on designing sophisticated CNN or Transformer architectures to recover distorted and missing texture. In this paper, we try to integrate location information and physical model into off-the-shelf CNN or Transformer architectures to help improve their performance. Specifically, we notice that existing methods deploy a preprocessing sub-network to generate a binary or soft mask to indicate the raindrop location, which will increase the network parameters and computational complexity. In contrast, a location-aware learning branch is embedded to teach the encoder in the training phase with the capability of perceiving the position of the raindrops. Note that this location-aware learning branch can be removed during the inference process (achieving performance improvements at no cost). Furthermore, instead of directly reconstructing the raindrop-free image (i.e., background scene), we devise a physics-based reconstruction scheme to first learn the transparency matrix and the raindrop layer. The latent background layer is then reversely derived based on the physical model. By combining the above-mentioned components, we propose our location-aware learning and physics-based reconstruction (LLPR) framework for this challenging ill-posed problem. We also collect a real-world raindrop-degraded image dataset, which is challenging for single-image raindrop removal (SIRR) methods. Extensive experimental results demonstrate the effectiveness and generality of our LLPR framework, achieving superior performance against state-of-the-art SIRR methods. The code will be made available upon acceptance.

Problem

• raindrop removal
• single image
• image restoration
• ill-posed problem
• real-world dataset

Innovation

• Location-aware learning
• Physics-based reconstruction
• Raindrop removal
• Single image restoration
• Transparency matrix