Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Dynamic High-frequency Convolution for Infrared Small Target Detection

Ruojing Li, Chao Xiao, Qian Yin, Wei An, Nuo Chen, Xinyi Ying, Miao Li, Yingqian Wang

TL;DR

This work tackles infrared small target detection (SIRST) by addressing the challenge that targets and background clutter both manifest as high-frequency components (HFCs). It introduces Dynamic High-frequency Convolution (DHiF), a region-adaptive, dynamically generated filter bank conditioned on local input features, with kernel parameters constrained to lie in a zero-centered interval to emphasize high-frequency content via the Fourier relation $\frac{d}{dt}f(t) \leftrightarrow j\omega F(\omega)$. DHiF acts as a drop-in replacement for standard convolutions and can be integrated into arbitrary SIRST encoders, enabling discriminative learning of HFC representations with negligible overhead. Extensive experiments on real-world datasets (IRSDT-1k and NUAA-SIRST) show consistent improvements across diverse networks and settings, demonstrating DHiF’s robustness and generality, with code released for public use.

Abstract

Infrared small targets are typically tiny and locally salient, which belong to high-frequency components (HFCs) in images. Single-frame infrared small target (SIRST) detection is challenging, since there are many HFCs along with targets, such as bright corners, broken clouds, and other clutters. Current learning-based methods rely on the powerful capabilities of deep networks, but neglect explicit modeling and discriminative representation learning of various HFCs, which is important to distinguish targets from other HFCs. To address the aforementioned issues, we propose a dynamic high-frequency convolution (DHiF) to translate the discriminative modeling process into the generation of a dynamic local filter bank. Especially, DHiF is sensitive to HFCs, owing to the dynamic parameters of its generated filters being symmetrically adjusted within a zero-centered range according to Fourier transformation properties. Combining with standard convolution operations, DHiF can adaptively and dynamically process different HFC regions and capture their distinctive grayscale variation characteristics for discriminative representation learning. DHiF functions as a drop-in replacement for standard convolution and can be used in arbitrary SIRST detection networks without significant decrease in computational efficiency. To validate the effectiveness of our DHiF, we conducted extensive experiments across different SIRST detection networks on real-scene datasets. Compared to other state-of-the-art convolution operations, DHiF exhibits superior detection performance with promising improvement. Codes are available at https://github.com/TinaLRJ/DHiF.

Dynamic High-frequency Convolution for Infrared Small Target Detection

TL;DR

This work tackles infrared small target detection (SIRST) by addressing the challenge that targets and background clutter both manifest as high-frequency components (HFCs). It introduces Dynamic High-frequency Convolution (DHiF), a region-adaptive, dynamically generated filter bank conditioned on local input features, with kernel parameters constrained to lie in a zero-centered interval to emphasize high-frequency content via the Fourier relation . DHiF acts as a drop-in replacement for standard convolutions and can be integrated into arbitrary SIRST encoders, enabling discriminative learning of HFC representations with negligible overhead. Extensive experiments on real-world datasets (IRSDT-1k and NUAA-SIRST) show consistent improvements across diverse networks and settings, demonstrating DHiF’s robustness and generality, with code released for public use.

Abstract

Infrared small targets are typically tiny and locally salient, which belong to high-frequency components (HFCs) in images. Single-frame infrared small target (SIRST) detection is challenging, since there are many HFCs along with targets, such as bright corners, broken clouds, and other clutters. Current learning-based methods rely on the powerful capabilities of deep networks, but neglect explicit modeling and discriminative representation learning of various HFCs, which is important to distinguish targets from other HFCs. To address the aforementioned issues, we propose a dynamic high-frequency convolution (DHiF) to translate the discriminative modeling process into the generation of a dynamic local filter bank. Especially, DHiF is sensitive to HFCs, owing to the dynamic parameters of its generated filters being symmetrically adjusted within a zero-centered range according to Fourier transformation properties. Combining with standard convolution operations, DHiF can adaptively and dynamically process different HFC regions and capture their distinctive grayscale variation characteristics for discriminative representation learning. DHiF functions as a drop-in replacement for standard convolution and can be used in arbitrary SIRST detection networks without significant decrease in computational efficiency. To validate the effectiveness of our DHiF, we conducted extensive experiments across different SIRST detection networks on real-scene datasets. Compared to other state-of-the-art convolution operations, DHiF exhibits superior detection performance with promising improvement. Codes are available at https://github.com/TinaLRJ/DHiF.
Paper Structure (16 sections, 8 equations, 3 figures, 2 tables)

This paper contains 16 sections, 8 equations, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. Importance of HFC Representation Learning
  4. Dynamic High-frequency Convolution
  5. Convolutions with predefined filters
  6. Sensitivity to HFCs
  7. Structure of DHiF
  8. Experiments
  9. Experimental Setup
  10. Datasets
  11. Metrics
  12. Comparison to the State-of-the-arts
  13. Ablation Study
  14. Position of drop-in replacement and efficiency
  15. Effectiveness on learning differential representations
  16. ...and 1 more sections

Figures (3)

  • Figure 1: Attribution visualizations of different methods for the predictions of target regions. The right three columns of subplots present the attribution maps generated by different methods, followed underneath by overlay visualizations with the input images. The red regions indicate contributions to the target predictions, with color intensity corresponding to the influence magnitude.
  • Figure 2: DHiF structure visualization. A dynamic local filter bank is generated via transformations based on local input features. The filters process local features to produce filtered high-frequency representations. Standard convolution then integrates filtered features with original input to yield output features.
  • Figure 3: Visualizations of features in the second levels of DNANet variants without and with DHiF. (a) The input image including (1) target, (2) building corner, and (3) bright clutter. (c) Local view of the output map generated by the original DNANet. (d, f) Local input features and (e, g) local output features of the first convolutions (standard convolution and DHiF). (h) Local filter banks of three HFCs.