信号处理2025，Vol.41Issue(12)：1926-1939,14.DOI:10.12466/xhcl.2025.12.006
SAR图像去斑的频率驱动异质模型
A Frequency-Driven Heterogeneous Model for SAR Image Despeckling
摘要
Abstract
Most of the existing Synthetic Aperture Radar(SAR)image despeckling models have been derived from re-lated generic vision fields and generally lack the capability to handle the spatially heterogeneous distribution characteris-tics of speckle noise.This paper proposes a Frequency-Driven Heterogeneous Model for SAR image despeckling,which explicitly separates high-and low-frequency features using pooling layers and processes them separately based on their distinct noise-distribution characteristics.First,a Dconv Cross-Covariance Attention mechanism was designed to model the spatial-channel global characteristics of high-frequency features,enhancing the representation of critical edge infor-mation and removing complex speckle effects.Then,a Large Kernel Convolutional Residual Module was introduced to effectively suppress the simple speckle interference in low-frequency features by leveraging the receptive field advan-tage.Finally,to strengthen the complementary interaction between frequency features and further improve the despeck-ling performance,a Selective Frequency Fusion Module was used to adaptively fuse the heterogeneously processed high-and low-frequency features with minimal parameters.The results of experiments on both simulated and real SAR image datasets demonstrated that the proposed model outperformed some existing despeckling methods in both quantita-tive metrics and visual quality.Compared to the suboptimal method,it achieved average PSNR and SSIM increases of 0.14 dB and 0.01,respectively,while the inference time was reduced by at least 99%.Compared to classical lightweight methods,it achieved average PSNR and SSIM increases of 0.78～1.12 dB and 0.02～0.06,respectively.关键词
SAR图像去斑/异质模型/卷积神经网络/注意力机制Key words
SAR image despeckling/heterogeneous model/convolutional neural network/attention mechanism分类
信息技术与安全科学引用本文复制引用
周彬,罗毅,张江波..SAR图像去斑的频率驱动异质模型[J].信号处理,2025,41(12):1926-1939,14.基金项目
四川省科技计划资助项目(2025ZNSFSC0496) Sichuan Science and Technology Program(2025ZNSFSC0496) （2025ZNSFSC0496）
