TIRMamba：一种面向亚百万参数红外图像超分辨率的 Thermal-Prior-Modulated 状态空间网络
TIRMamba: A Thermal-Prior-Modulated State-Space Network for Sub-Million-Parameter Infrared Image Super-Resolution
- National Chung Hsing University（国立中兴大学）
- Providence University（静宜大学）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
TIRMamba 提出一种亚百万参数（896K-910K）的红外图像超分辨率状态空间网络，通过热先验调制与重放训练，在减少参数和延迟的同时匹配或超越现有方法。
AI中文摘要：
红外图像超分辨率目前由基于 Mamba 的网络主导，其参数量在 2600 万至 3700 万之间，难以部署在热成像最需要的机载和手持平台上。本文提出 TIRMamba，一个具有 896K 至 910K 参数的单通道热图像网络。一个 Thermal Prior Highway 在输入时一次性计算梯度、局部对比度和光谱线索，并通过每个残差组一个适配器，调制一个权重共享的双向状态空间主干，并门控其双尺度细节分支；一个三路径重建将学习到的残差添加到双三次辐射度基线。由于标准基准仅提供 265 张红外训练图像，并在完整图像上评估融合产品，我们采用重放策略进行训练：灰度 DIV2K 预训练，随后在从红外和自然语料库中以等概率抽取的 64 像素块上进行微调。在尺度因子 4 下，TIRMamba 在两个官方测试集上匹配最强的协议训练方法，参数减少 29 至 40 倍，延迟降低 2.8 至 9.4 倍；在尺度因子 2 下，它在两个测试集上均给出最高的 SSIM。一个具有先验条件选择性的变体 TIRMamba-Rad，纠正了中间尺寸不变设计的 3 dB 原始热失败，并在原始热、无人机和独立传感器测试集上在尺度因子 4 下给出最佳结果。代码和训练模型将在录用后于此 https URL 发布。
英文摘要：
Infrared image super-resolution is currently led by Mamba-based networks with 26 to 37 million parameters, which are difficult to deploy on the airborne and handheld platforms where thermal imaging is most needed. This paper presents TIRMamba, a network with 896K to 910K parameters for single-channel thermal imagery. A Thermal Prior Highway computes gradient, local-contrast and spectral cues once at the input and, through one adapter per residual group, modulates a weight-tied bidirectional state-space trunk and gates its dual-scale detail branch; a tri-path reconstruction adds the learned residual to a bicubic radiometric baseline. Because the standard benchmark provides only 265 infrared training images and evaluates fusion products on full images, we train with a replay strategy: grayscale DIV2K pre-training followed by fine-tuning on 64-pixel patches drawn with equal probability from the infrared and natural corpora. At scale factor 4, TIRMamba matches the strongest protocol-trained methods on both official test sets with 29 to 40 times fewer parameters and 2.8 to 9.4 times lower latency; at scale factor 2 it gives the highest SSIM on both. A variant with prior-conditioned selectivity, TIRMamba-Rad, corrects a 3 dB raw-thermal failure of an intermediate size-invariant design and gives the best results at scale factor 4 on raw-thermal, unmanned-aerial-vehicle and independent-sensor test sets. Code and trained models will be released at https://github.com/julian135707/TIRMamba upon acceptance.
