LG-DBNet: Local and Global Dual-Branch Network for SAR Image Denoising

Synthetic aperture radar (SAR) tends to be seriously affected by speckle noise due to its inherent imaging characteristics, which brings great challenges to the high-level visualization task of SAR images. Speckle suppression, therefore, plays a crucial role in remote sensing image processing. Attention-based SAR image denoising algorithms frequently struggle to capture rich feature information and face challenges in balancing the trade-off between denoising and preserving texture details. To solve the above problems, this article constructs a local and global dual-branch network (LG-DBNet) for SAR image denoising. This network can effectively suppress speckle noise while fully retaining the detailed information of the original image. First, the shallow features are extracted through simple convolution. Then, a dual-branch structure constructed using different attention modules is used to extract deep features from SAR images. Specifically, one branch performs local deep feature extraction of an image through a hybrid attention module built by a convolutional neural network (CNN), while the other branch uses a superposition of self-attention mechanisms for global deep feature extraction of the image. Finally, the final denoised image is generated through global residual learning. LG-DBNet can effectively extract the local and global image information through the dual-branch structure and further focus on the noise information, which can better retain the texture information of the image while effectively denoising. The experimental results show that compared with the state-of-the-art SAR image denoising algorithms, the proposed algorithm not only improves on various objective indexes but also shows great advantages in the visual effect after denoising.

I. Introduction

Synthetic aperture radar (SAR) is an active observation system applied in the field of remote sensing. With the development of electronic information technology, SAR can now obtain long-range, high-resolution images. Since SAR can be imaged all day and in all weather, it is widely used in civil and military fields, such as environmental monitoring and ocean monitoring [1]. Because of the limitation of the imaging mechanism, the quality of SAR images can be seriously affected by the speckle noise generated by the interference of backward-scattered microwave signals, which brings challenges to the subsequent visual interpretation of SAR images (e.g., target detection, etc.). The suppression of speckle noise in SAR images is, therefore, a crucial task. To obtain high-quality SAR images, scholars from various countries have conducted various research on the task of SAR image denoising. In general, SAR denoising algorithms can be divided into three major categories: denoising algorithms based on spatial domain filtering, denoising algorithms based on frequency-domain filtering, and denoising algorithms based on deep learning. They are simply introduced in the following.