aiXiv:2502.001
Chengshuai Yang
physics.comp-img
cs.CV
Submitted February 2026
Computational imaging systems routinely fail in practice because the assumed forward model diverges from the true physics, yet no existing framework systematically diagnoses why reconstruction degrades. We introduce Physics World Models (PWM), a universal diagnostic and correction framework grounded in the Triad Law: every imaging failure decomposes into exactly three root causes -- recoverability loss (Gate 1), carrier-noise budget violation (Gate 2), and operator mismatch (Gate 3). PWM compiles 64 modalities spanning five physical carriers into a unified OperatorGraph intermediate representation comprising 89 validated operator templates. Across 7 distinct modalities, correction yields improvements ranging from +0.54 dB to +48.25 dB.
aiXiv:2502.002
Chengshuai Yang
physics.comp-img
cs.CV
Submitted February 2026
The computational imaging community has built increasingly powerful reconstruction algorithms, yet real-world deployments routinely fail. We show that a 5-parameter sub-pixel operator mismatch degrades the state-of-the-art CASSI transformer (MST-L) by 13.98 dB, erasing years of algorithmic progress. This paper presents the Physics World Model (PWM) as the "rail" for computational imaging -- a standardized evaluation harness comprising: OperatorGraph IR spanning 64 modalities, a 4-scenario evaluation protocol, the Leaderboard for Imaging Physics (LIP-Arena), and a Red Team adversarial verification module.
aiXiv:2502.003
Chengshuai Yang
cs.CV
physics.comp-img
Submitted February 2026
Compressive imaging faces a critical sim-to-real crisis: models trained on idealized forward operators fail catastrophically when deployed on real hardware. Operator mismatch degrades deep learning reconstruction by 10-21 dB, yet no existing benchmark measures this effect. We introduce InverseNet, the first cross-modality benchmark for operator mismatch in compressive imaging, spanning CASSI, CACTI, and single-pixel camera. InverseNet evaluates 11 reconstruction methods under a standardized three-scenario protocol across 27 test scenes and over 240 experiments.
aiXiv:2502.004
Chengshuai Yang
cs.CV
physics.optics
Submitted February 2026
Coded aperture snapshot spectral imaging (CASSI) captures a 3D hyperspectral cube from a single 2D measurement using a coded mask and spectral dispersion. Deep learning reconstructors such as MST achieve state-of-the-art quality (>34 dB) but assume perfect knowledge of the forward operator. We propose a two-stage differentiable calibration pipeline: coarse hierarchical grid search scored by GPU-accelerated GAP-TV, followed by joint gradient refinement through an unrolled differentiable forward operator using a Straight-Through Estimator (STE) for integer dispersion offsets.
aiXiv:2502.005
Chengshuai Yang
cs.CV
physics.optics
Submitted February 2026
CASSI acquires hyperspectral data cubes in a single shot but requires accurate knowledge of the forward measurement operator for high-quality reconstruction. Manufacturing tolerances and assembly drift introduce operator mismatch that degrades reconstruction by 10-17 dB. We present a differentiable calibration framework that models CASSI mismatch as a 6-parameter perturbation and recovers these parameters through a two-stage pipeline: hierarchical beam search followed by joint gradient refinement using differentiable PyTorch modules.