Compressive sensing inspired self-supervised single-pixel imaging

2026-03-31

AI Summary

This work addresses the challenge of single-pixel imaging under strong interference, where existing methods often suffer from structural distortion and blurred details due to insufficient integration of local and global features, stemming from a lack of physically grounded sparsity constraints. To overcome this limitation, we propose SISTA-Net, which uniquely combines physics-driven sparse priors with self-supervised learning by unfolding the Iterative Shrinkage-Thresholding Algorithm (ISTA) into an interpretable deep network. The architecture incorporates a data fidelity module and a proximal mapping module, enhanced with learnable soft-thresholding operators and deep nonlinear sparse transforms. Leveraging a hybrid CNN–Visual State Space Model (VSSM) design, SISTA-Net achieves markedly improved robustness at extremely low sampling rates. Experiments demonstrate a 2.6 dB PSNR gain in simulations and an average 3.4 dB improvement in real-world long-range underwater scenarios, substantially outperforming current state-of-the-art approaches.

Technology Category

Computer Vision: Other Foundations of Computer VisionIntelligent Robots: State EstimationCognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for rankingGraph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphsSystems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Single-pixel imaging (SPI) is a promising imaging modality with distinctive advantages in strongly perturbed environments. Existing SPI methods lack physical sparsity constraints and overlook the integration of local and global features, leading to severe noise vulnerability, structural distortions and blurred details. To address these limitations, we propose SISTA-Net, a compressive sensing-inspired self-supervised method for single-pixel imaging. SISTA-Net unfolds the Iterative Shrinkage-Thresholding Algorithm (ISTA) into an interpretable network consisting of a data fidelity module and a proximal mapping module. The fidelity module adopts a hybrid CNN-Visual State Space Model (VSSM) architecture to integrate local and global feature modeling, enhancing reconstruction integrity and fidelity. We leverage deep nonlinear networks as adaptive sparse transforms combined with a learnable soft-thresholding operator to impose explicit physical sparsity in the latent domain, enabling noise suppression and robustness to interference even at extremely low sampling rates. Extensive experiments on multiple simulation scenarios demonstrate that SISTA-Net outperforms state-of-the-art methods by 2.6 dB in PSNR. Real-world far-field underwater tests yield a 3.4 dB average PSNR improvement, validating its robust anti-interference capability.

Problem

• single-pixel imaging
• compressive sensing
• sparsity constraint
• noise vulnerability
• feature integration

Innovation

• single-pixel imaging
• compressive sensing
• self-supervised learning
• sparsity constraint
• Visual State Space Model