Thermal Polarimetric Multi-view Stereo

2025-10-23

AI Summary

Existing 3D reconstruction methods struggle with transparent, semi-transparent, and complex-material objects under varying illumination and material properties, yielding low geometric accuracy. To address this, we propose a multi-view stereo reconstruction framework based on long-wave infrared (LWIR) polarimetric imaging. By formulating a physically grounded thermal radiation polarization model, our approach resolves the inherent normal-albedo ambiguity prevalent in visible-light polarization, enabling robust, illumination- and material-invariant surface normal estimation. We further integrate multi-view geometry with LWIR polarimetric radiometric constraints into an end-to-end shape inference pipeline. Experiments demonstrate that our method significantly outperforms state-of-the-art techniques across diverse challenging materials—including glass, plastic, and translucent ceramics—achieving sub-millimeter geometric fidelity. This work establishes a novel paradigm for robust, textureless, low-reflectance 3D reconstruction.

Technology Category

• Computer Vision: 3D Computer Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Multi-instance/Multi-view Learning

Application Category

• Security and Privacy: Data transparency and provenance
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Sustainability and carbon-aware systems for Web, mobile, and WoT

Abstract

This paper introduces a novel method for detailed 3D shape reconstruction utilizing thermal polarization cues. Unlike state-of-the-art methods, the proposed approach is independent of illumination and material properties. In this paper, we formulate a general theory of polarization observation and show that long-wave infrared (LWIR) polarimetric imaging is free from the ambiguities that affect visible polarization analyses. Subsequently, we propose a method for recovering detailed 3D shapes using multi-view thermal polarimetric images. Experimental results demonstrate that our approach effectively reconstructs fine details in transparent, translucent, and heterogeneous objects, outperforming existing techniques.

Problem

• Reconstructing 3D shapes using thermal polarization cues
• Overcoming ambiguities in visible polarization analyses
• Reconstructing fine details in transparent and translucent objects

Innovation

• Uses thermal polarization cues for 3D reconstruction
• Independent of illumination and material properties
• Resolves ambiguities with long-wave infrared polarimetric imaging