面向压缩感知的多尺度梯度引导展开架构与自适应 Mamba
Multi-Scale Gradient-Guided Unrolling Architecture with Adaptive Mamba for Compressive Sensing
把 Mamba 状态空间块嵌入 PGD 展开框架，用多尺度展开与高维梯度融合替代跨阶段同质化特征提取，为图像复原类逆问题提供兼顾全局感受野与效率的可解释网络骨架。
原文依据 we utilize our customized Adaptive State-Space Block (A-SSB) to unroll the well-known Proximal Gradient Descent (PGD) algorithm across multiple feature levels
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论在压缩感知（CS）领域，深度展开网络（DUN）通过将传统优化求解器与深度网络相结合，展现出卓越的性能与可解释性。然而，现有 DUN 存在跨阶段特征提取同质化以及梯度引导信息融合不足的问题。此外，特征提取模块难以在全局感受野与计算效率之间取得平衡，从而限制了图像重建细节的提升。为解决上述挑战，我们提出了一种面向压缩感知的多尺度梯度引导展开架构，并配备自适应 Mamba，命名为 MambaCS。具体而言，我们利用定制的自适应状态空间块（A-SSB）在多个特征层级上展开著名的近端梯度下降（PGD）算法，在保持计算效率的同时提取全面的图像特征。此外，我们设计了高维梯度融合（HDGF），确保梯度引导信息在不同尺度与维度上持续、稳定地注入，同时有效消除信息瓶颈。最后，我们开发了特征自适应近端算子（FAPO），以 A-SSB 作为与 PGD 近端算子相关联的稀疏基的扩展，增强了对多尺度特征的敏感性并改善了细节重建。大量实验表明，我们提出的 MambaCS 相较于当前最先进方法具有显著优势。我们的代码可在 https://github.com/nikou-arch/MambaCS 获取。
英文原文摘要 1580 字符
In the field of Compressive Sensing (CS), deep unrolling networks (DUNs) have demonstrated exceptional performance and interpretability by integrating traditional optimization solvers with deep networks. However, existing DUNs suffer from homogenization in cross-stage feature extraction and insufficient integration of gradient-guided information. Additionally, the feature extraction module struggles to balance the global receptive field and computational efficiency, which limits improvements in image reconstruction details. To address these challenges, we propose a multi-scale gradient-guided unrolling architecture with adaptive Mamba for CS, named MambaCS. Specifically, we utilize our customized Adaptive State-Space Block (A-SSB) to unroll the well-known Proximal Gradient Descent (PGD) algorithm across multiple feature levels to extract comprehensive image features while maintaining computational efficiency. Moreover, we design a High-Dimensional Gradient Fusion (HDGF) that ensures the persistent and stable injection of gradient-guided information across various scales and dimensions, while effectively eliminating information bottlenecks. Finally, we develop a Feature-Adaptive Proximal Operator (FAPO), using A-SSB as an extension of the sparse basis associated with the PGD proximal operator, which enhances sensitivity to multi-scale features and improves detail reconstruction. Extensive experiments demonstrate the significant advantages of our proposed MambaCS over the current SOTA methods. Our code is available at https://github.com/nikou-arch/MambaCS.