CLEAR-IR: Clarity-Enhanced Active Reconstruction of Infrared Imagery

2025-10-06

AI Summary

In low-light conditions, strong structured interference introduced by active infrared (IR) emitters severely degrades high-level vision tasks such as object detection, tracking, and localization. To address this, we propose a U-Net-based end-to-end IR image denoising and reconstruction method that directly recovers high-fidelity, interference-free IR images from raw IR streams corrupted by intense active emitter patterns—marking the first such approach. Our method significantly enhances feature quality critical for downstream perception tasks. Extensive evaluation across diverse illumination conditions demonstrates superior performance over state-of-the-art methods in both reconstruction fidelity (PSNR/SSIM) and robotic perception metrics (detection mAP, tracking IDF1, and localization accuracy). Key contributions include: (i) the first dedicated reconstruction framework explicitly designed for active IR interference suppression; and (ii) unified optimization of interference removal and high-level task performance, bridging low-level restoration with semantic-level perception gains.

Technology Category

• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Computer Vision: Low Level & Physics-based Vision
• Planning, Routing, and Scheduling: Activity and Plan Recognition

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

This paper presents a novel approach for enabling robust robotic perception in dark environments using infrared (IR) stream. IR stream is less susceptible to noise than RGB in low-light conditions. However, it is dominated by active emitter patterns that hinder high-level tasks such as object detection, tracking and localisation. To address this, a U-Net-based architecture is proposed that reconstructs clean IR images from emitter-populated input, improving both image quality and downstream robotic performance. This approach outperforms existing enhancement techniques and enables reliable operation of vision-driven robotic systems across illumination conditions from well-lit to extreme low-light scenes.

Problem

• Enhancing robotic perception in dark environments using infrared imagery
• Removing active emitter patterns hindering object detection and tracking
• Improving image quality for reliable vision-driven robotic systems

Innovation

• U-Net architecture reconstructs clean infrared images
• Removes active emitter patterns from IR stream
• Enhances robotic perception in low-light conditions