Physically-Guided Optical Inversion Enable Non-Contact Side-Channel Attack on Isolated Screens

TL;DR

This paper presents IR4Net, a novel optical inversion method that enables non-contact side-channel attacks on screens by combining physics-based regularization with deep learning to improve reconstruction fidelity and robustness.

Contribution

It introduces IR4Net, a new neural network architecture that integrates physical models and semantic constraints for effective optical inversion in side-channel attacks.

Findings

IR4Net outperforms existing neural methods in fidelity across four scene categories.

The approach maintains robustness against illumination perturbations.

Physically regularized inversion improves reconstruction accuracy.

Abstract

Noncontact exfiltration of electronic screen content poses a security challenge, with side-channel incursions as the principal vector. We introduce an optical projection side-channel paradigm that confronts two core instabilities: (i) the near-singular Jacobian spectrum of projection mapping breaches Hadamard stability, rendering inversion hypersensitive to perturbations; (ii) irreversible compression in light transport obliterates global semantic cues, magnifying reconstruction ambiguity. Exploiting passive speckle patterns formed by diffuse reflection, our Irradiance Robust Radiometric Inversion Network (IR4Net) fuses a Physically Regularized Irradiance Approximation (PRIrr-Approximation), which embeds the radiative transfer equation in a learnable optimizer, with a contour-to-detail cross-scale reconstruction mechanism that arrests noise propagation. Moreover, an Irreversibility Constrained Semantic Reprojection (ICSR) module reinstates lost global structure through context-driven semantic mapping. Evaluated across four scene categories, IR4Net achieves fidelity beyond competing neural approaches while retaining resilience to illumination perturbations.