Practical wave optical measurements for material and biophysical appearance analysis
File(s)
Author(s)
Nogué, Emilie
Type
Thesis or dissertation
Abstract
Since the 1990s, computer graphics has advanced dramatically in step with exponential growth in computing power, evolving from basic 3D scenes to visual effects nearly indistinguishable from reality. A core reason for this progress is the creative adaptation of established physical theories—especially ray optics—into efficient rendering systems. In this ray-based paradigm, light is treated as incoherent energy traveling along straight paths, which is a powerful simplification for many macroscopic visual phenomena.
However, certain aspects of material appearance and biological tissue reflectance depend critically on the wave nature of light—its polarization, coherence, and wavelength-specific interactions—which conventional ray models overlook. This thesis revisits this overlooked dimension by demonstrating how the wave properties of light, in particular polarization, can be practically measured and exploited to recover physically meaningful parameters from real surfaces and skin.
The work develops compact, physics-driven frameworks for estimating spatially varying BRDFs of planar materials and biophysical maps of facial skin using polarization cues and narrowband RGB imaging. These methods show that direct measurement, guided by wave-optical principles and closed-form analytical solutions, can yield robust, interpretable results with minimal hardware and no reliance on learned priors. Moreover, they provide a foundation for hybrid pipelines that combine fast measurement with differentiable rendering or machine learning for fine-tuning and real-world generalization.
Together, these contributions demonstrate that reintroducing wave optics into material and skin modeling bridges the gap between theoretical light transport and practical, accessible acquisition—enabling more faithful rendering, more transparent inverse models, and new applications in graphics, vision, and biomedical imaging.
However, certain aspects of material appearance and biological tissue reflectance depend critically on the wave nature of light—its polarization, coherence, and wavelength-specific interactions—which conventional ray models overlook. This thesis revisits this overlooked dimension by demonstrating how the wave properties of light, in particular polarization, can be practically measured and exploited to recover physically meaningful parameters from real surfaces and skin.
The work develops compact, physics-driven frameworks for estimating spatially varying BRDFs of planar materials and biophysical maps of facial skin using polarization cues and narrowband RGB imaging. These methods show that direct measurement, guided by wave-optical principles and closed-form analytical solutions, can yield robust, interpretable results with minimal hardware and no reliance on learned priors. Moreover, they provide a foundation for hybrid pipelines that combine fast measurement with differentiable rendering or machine learning for fine-tuning and real-world generalization.
Together, these contributions demonstrate that reintroducing wave optics into material and skin modeling bridges the gap between theoretical light transport and practical, accessible acquisition—enabling more faithful rendering, more transparent inverse models, and new applications in graphics, vision, and biomedical imaging.
Version
Open Access
Date Issued
2025-06-18
Date Awarded
2025-10-01
Copyright Statement
Attribution-NonCommercial 4.0 International Licence (CC BY-NC)
License URL
Advisor
Ghosh, Abhijeet
Sponsor
European Commission
Grant Number
956585
Publisher Department
Department of Computing
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)