🤖 AI Summary
To address insufficient global column-wise feature modeling and cross-level semantic discontinuities in infrared image stripe noise removal, this paper proposes the Asymmetric Sampling Correction Network (ASCNet). ASCNet integrates column-wise priors with data-driven features within a U-shaped encoder-decoder architecture, enabling— for the first time in single-image stripe removal—explicit modeling of long-range column dependencies while effectively decoupling stripe noise from authentic vertical structures. Key innovations include: (i) Residual Haar Discrete Wavelet Transform downsampling (RHDWT), (ii) Pixel-Shuffling unbiased upsampling (PS), and (iii) Column Non-Uniformity Correction Module (CNCM), collectively supporting multi-scale column-context modeling. Extensive experiments on synthetic and real-world infrared datasets, as well as downstream small-target detection tasks, demonstrate significant PSNR/SSIM improvements over state-of-the-art methods. The source code is publicly available.
📝 Abstract
In a real-world infrared imaging system, effectively learning a consistent stripe noise removal model is essential. Most existing destriping methods cannot precisely reconstruct images due to cross-level semantic gaps and insufficient characterization of the global column features. To tackle this problem, we propose a novel infrared image destriping method, called Asymmetric Sampling Correction Network (ASCNet), that can effectively capture global column relationships and embed them into a U-shaped framework, providing comprehensive discriminative representation and seamless semantic connectivity. Our ASCNet consists of three core elements: Residual Haar Discrete Wavelet Transform (RHDWT), Pixel Shuffle (PS), and Column Non-uniformity Correction Module (CNCM). Specifically, RHDWT is a novel downsampler that employs double-branch modeling to effectively integrate stripe-directional prior knowledge and data-driven semantic interaction to enrich the feature representation. Observing the semantic patterns crosstalk of stripe noise, PS is introduced as an upsampler to prevent excessive apriori decoding and performing semantic-bias-free image reconstruction. After each sampling, CNCM captures the column relationships in long-range dependencies. By incorporating column, spatial, and self-dependence information, CNCM well establishes a global context to distinguish stripes from the scene's vertical structures. Extensive experiments on synthetic data, real data, and infrared small target detection tasks demonstrate that the proposed method outperforms state-of-the-art single-image destriping methods both visually and quantitatively. Our code will be made publicly available at https://github.com/xdFai/ASCNet.