Generative denoising improves infrared small target detection accuracy
Seeing through Denoising: Frequency-guided One-step Denoising for Infrared Small Target Detection
Computer Vision and Pattern Recognition
Summary
Detecting tiny objects in infrared images is very hard because these targets are very small and often hidden in noisy backgrounds. The authors present a new method that treats small targets as noise and uses a special model to clean the background first, then finds targets by spotting what was removed. This approach is faster thanks to a shortcut that reduces processing steps and better preserves important image details. Their method can work alongside existing detection tools without slowing them down and shows better performance on several test datasets.
What this means in practice
- •For security monitoring teams: Automatically identify small heat signatures more reliably in infrared surveillance footage using improved background separation.
- •For defense imaging specialists: Enhance detection of small infrared targets in cluttered battlefield imagery by integrating generative background reconstruction techniques.
Authors
Xuelin Qian, Zixuan Wang, Chang Xu, Dingwen Zhang, Junwei Han
Abstract
Infrared small target detection (IRSTD) is a challenging task due to the extremely small size of targets, low signal-to-noise ratios, and heavy background clutter in infrared imagery. While the progress of deep learning has advanced this field, existing methods primarily adopt discriminative paradigms that directly map images to target masks, overlooking the potential of generative modeling. In this paper, we propose DiffIRSTD, a novel generative framework that reformulates infrared small target detection as a background reconstruction problem. Instead of explicitly predicting target regions, we reinterpret small targets as noise-like signals and leverage a diffusion-based model to recover a clean, target-free background from a perturbed input image. The residual between the reconstructed background and the original image is then exploited to highlight candidate target regions, providing explicit cues for subsequent detection. To address the computational cost typically associated with diffusion sampling, we introduce a one-step denoising strategy that enables efficient background reconstruction while maintaining practical inference speed. Furthermore, we present a frequency-guided perception loss to preserve high-frequency background details and focus more on the reconstruction quality of target regions. More importantly, our DiffIRSTD operates as a plug-and-play framework, compatible with existing IRSTD models without increasing the computational complexity of detectors. Extensive experiments on IRSTD-1K, NUAA-SIRST, and NUDT-SIRST datasets demonstrate the superiority of our method, achieving state-of-the-art performance across three benchmarks.