IR275K: A Benchmark for Infrared Multi-Frame Super-Resolution Toward Efficient Remote Sensing

2026-07-24

AI Summary

This study addresses the absence of a unified public benchmark for infrared multi-frame super-resolution, which hinders progress in handling challenges such as low thermal contrast, strong noise, weak textures, and platform motion. To this end, the authors introduce IR275K, the first large-scale infrared video benchmark comprising 594 sequences and 275,196 frames, along with standardized train–test splits and a ×4 evaluation protocol. They also propose CGMamba, a lightweight state space model (10.90M parameters, 112.14G FLOPs) that integrates 2D rotary positional encoding (2D RoPE) and a center-guided cross-Mamba (CGCM) mechanism to enable efficient implicit multi-frame reconstruction. Experiments show that CGMamba achieves 33.19 dB PSNR on IR275K, outperforming existing single-image infrared methods by 0.35–0.52 dB at lower computational cost, while ablation studies confirm the critical role of 2D RoPE in suppressing grid-like artifacts and enhancing performance.

Abstract

Efficient processing is becoming increasingly important in infrared remote sensing, where satellite constellations produce large volumes of observations under constrained detector resolution, power, and downlink bandwidth. Multi-frame super-resolution (MFSR) offers a software-based route to spatial enhancement, but its evaluation in infrared sensing remains fragmented across private datasets and ad-hoc protocols. Existing benchmarks do not explicitly capture the thermal contrast, sensor noise, weak texture, and platform-induced frame-to-frame variation that characterize infrared video. We introduce IR275K, a curated benchmark containing 594 infrared video sequences and 275,196 frames. It provides sequence-level train/validation/test splits and a reproducible X4 evaluation protocol. As an initial architectural probe, we further evaluate CGMamba, a lightweight state-space model with 10.90M parameters and 112.14G FLOPs. CGMamba combines 2D rotary position encoding (2D~RoPE) with center-guided cross-Mamba (CGCM) fusion for implicit multi-frame reconstruction. It achieves 33.19dB PSNR, outperforming infrared single-image super-resolution references by 0.35--0.52~dB at substantially lower computational cost. Ablation results show that removing 2D~RoPE from CGCM causes a 1.53dB drop and severe grid-like artifacts. This indicates that explicit spatial anchoring is critical for stabilizing SSM-based cross-frame gating under infrared conditions. IR275K provides a reproducible foundation for accuracy--efficiency evaluation of infrared MFSR methods, while the architectural analysis offers a concrete starting point for spatially aware SSM design under resource-constrained infrared sensing. Dataset and evaluation resources are available at: https://github.com/InfraRecon7/IR275K.

Problem

• infrared remote sensing
• multi-frame super-resolution
• benchmark
• thermal contrast
• frame-to-frame variation

Innovation

• infrared multi-frame super-resolution
• IR275K benchmark
• state-space model
• 2D rotary position encoding
• efficient remote sensing