Fluctuations-driven Light and Acoustic Scattering mitigation for High-resolution imaging in dynamic environments

Projektbeschreibung

Bildgebende Systeme für einen klaren Durchblick in unübersichtlichen Umgebungen

Die Erfassung klarer Bilder in komplexen Umgebungen wie etwa in Nebel oder im Inneren von lebendem Gewebe, ist eine enorme Herausforderung, da das Licht unvorhersehbar gestreut wird. Mit den derzeitigen Methoden können klare Bilder durch das Streulicht nur dann erzeugt werden, wenn während des Verfahrens alles stillsteht – was in der Praxis selten möglich ist. Anstatt des Versuchs, unter diesen chaotischen Bedingungen Klarheit zu erzwingen, wird das ERC-finanzierte Projekt FLASH das Chaos selbst als Hilfsmittel nutzen. Durch die Analyse der Wechselwirkung von Licht mit sich bewegenden Objekten oder sich verändernden Umgebungen können verborgene Muster extrahiert und in klare, detaillierte Bilder umgewandelt werden. Die theoretischen und experimentellen Aktivitäten des Projekts könnten dazu beitragen, Bereiche wie die medizinische Diagnostik, die Umweltüberwachung und die Weltraumforschung zu revolutionieren.

Ziel

Optical imaging systems are central to modern life, from smartphones and autonomous vehicles to laser-based microscopes in biomedical investigations and clinical diagnostics. Yet, even the most advanced imaging systems are severely limited by random scattering of light in dynamic environments, from live tissues to rapidly varying fog. Recent breakthroughs in computational scattering compensation, including my own works, have allowed an unprecedented leap in capabilities: it is now possible to reconstruct crystal-clear megapixel-scale images through visually opaque turbid samples by computationally processing a set of scattered light frames captured under controlled illumination. However, current techniques are fundamentally limited by their assumption of static scenes: both the scattering medium and the target must remain fixed during acquisition, severely limiting their impact on practical applications.
Here, I propose a new paradigm: to exploit scene dynamics as the source of information rather than trying to eliminate or outpace them. This is made possible by two unpublished insights: (1) that target dynamics can serve the same purpose as controlled time-varying illuminations; (2) that scattering dynamics can as well be mathematically equivalent to illumination or target variations. We translate these findings to two new formalisms: a “dynamic matrix” for dynamic targets, and a “dual matrix” for dynamic scattering. Merging these formalisms with the recent model-based computational correction techniques, it is possible to transform dynamic scattering from an obstacle to a source of information.
Project FLASH combines experimental, numerical, and theoretical efforts to develop and apply these techniques across different modalities: from holographic and fluorescence imaging to lensless endoscopy and photoacoustic tomography. The results will allow new capabilities for imaging under dynamic conditions, with implications in other domains, from geophysics to RADAR.