Seeing Through the Noise: Improving Infrared Small Target Detection and Segmentation from Noise Suppression Perspective

TL;DR

This paper proposes a novel noise-suppression feature pyramid network (NS-FPN), which integrates a low-frequency guided feature purification module and a spiral-aware feature sampling module into the original FPN structure and significantly reduces false alarms and achieves superior performance on IRSTDS task.

Abstract

Infrared small target detection and segmentation (IRSTDS) is a critical yet challenging task in defense and civilian applications, owing to the dim, shapeless appearance of targets and severe background clutter. Recent CNN-based methods have achieved promising target perception results, but they only focus on enhancing feature representation to offset the impact of noise, which results in the increased false alarm problem. In this paper, through analyzing the problem from the frequency domain, we pioneer in improving performance from noise suppression perspective and propose a novel noise-suppression feature pyramid network (NS-FPN), which integrates a low-frequency guided feature purification (LFP) module and a spiral-aware feature sampling (SFS) module into the original FPN structure. The LFP module suppresses the noise features by purifying high-frequency components to achieve feature enhancement devoid of noise interference, while the SFS module further adopts spiral sampling to fuse target-relevant features in feature fusion process. Our NS-FPN is designed to be lightweight yet effective and can be easily plugged into existing IRSTDS frameworks. Extensive experiments on the IRSTD-1k and NUAA-SIRST datasets demonstrate that our method significantly reduces false alarms and achieves superior performance on IRSTDS task.

Figures (7)

• Figure 1: Performance comparison of our method with other methods on the IRSTD-1k and NUAA-SIRST datasets. The outer regions represent superior performance.
• Figure 2: The discrete haar wavelet transform (DWT) is utilized to decompose the original image into low and high frequency components. There is a crossover between the red and blue lines.
• Figure 3: The overall structure of our NS-FPN and the details of its conponents. For each scale, the feature is first fed into the LFP module, which generates a target-relevant spatial attention map based on low-frequency components to guide two-stage feature purification of high-frequency features. Subsequently, the purified feature is fused with the upper-layer feature through the SFS module to achieve semantic complementation according with spatial distribution prior of infrared small targets. Note that, convolutions are omitted for better visualization and $Y_4$ does not contain SFS module.
• Figure 4: Visualization of DAT and SFS sampling process.
• Figure 5: Visualization of the features at the $X_2$ level after the gradual addition of LFP and SFS in NS-FPN.