RobustSCI: Beyond Reconstruction to Restoration for Snapshot Compressive Imaging under Real-World Degradations

2026-03-08

AI Summary

Existing video snapshot compressive imaging (SCI) methods suffer significant performance degradation under real-world degradations such as motion blur and low illumination. This work introduces, for the first time, an SCI recovery task tailored to realistic degradations, establishing a new benchmark dataset based on DAVIS 2017 that incorporates continuous real-world degradation patterns. We propose RobustSCI, a novel framework centered on the RobustCFormer module, which explicitly decouples and removes degradation factors through a dual-branch architecture combining multi-scale deblurring and frequency-domain enhancement. Additionally, a lightweight cascaded post-processing network, RobustSCI-C, is introduced to substantially improve reconstruction quality with negligible computational overhead. Extensive experiments demonstrate that our method consistently outperforms state-of-the-art approaches on both the newly curated test set and real SCI data, marking a pivotal shift from merely reconstructing observations to faithfully recovering real-world scenes.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Security and Privacy: Large-scale security measurements
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Deep learning algorithms for video Snapshot Compressive Imaging (SCI) have achieved great success, yet they predominantly focus on reconstructing from clean measurements. This overlooks a critical real-world challenge: the captured signal itself is often severely degraded by motion blur and low light. Consequently, existing models falter in practical applications. To break this limitation, we pioneer the first study on robust video SCI restoration, shifting the goal from"reconstruction"to"restoration"--recovering the underlying pristine scene from a degraded measurement. To facilitate this new task, we first construct a large-scale benchmark by simulating realistic, continuous degradations on the DAVIS 2017 dataset. Second, we propose RobustSCI, a network that enhances a strong encoder-decoder backbone with a novel RobustCFormer block. This block introduces two parallel branches--a multi-scale deblur branch and a frequency enhancement branch--to explicitly disentangle and remove degradations during the recovery process. Furthermore, we introduce RobustSCI-C (RobustSCI-Cascade), which integrates a pre-trained Lightweight Post-processing Deblurring Network to significantly boost restoration performance with minimal overhead. Extensive experiments demonstrate that our methods outperform all SOTA models on the new degraded testbeds, with additional validation on real-world degraded SCI data confirming their practical effectiveness, elevating SCI from merely reconstructing what is captured to restoring what truly happened.

Problem

• Snapshot Compressive Imaging
• motion blur
• low light
• image restoration
• real-world degradations

Innovation

• Snapshot Compressive Imaging
• Robust Restoration
• Motion Deblurring
• Degradation Modeling
• RobustCFormer