IEDANet: Image Enhanced Dual Attention Networks for SAR Image Denoising
Ruiqing Wang, Ruxian Wang, Mengting Zhang, Jing Fang · 2025
Due to the presence of speckle noise in SAR images, which can reduce image resolution, removing speckle noise is very important. Therefore, we propose an image enhancement dual attention network IEDANet, which first generates enhanced images through image enhancement blocks and combines them with noisy images, then extracts features, and finally adds two attention modules to guide the network to focus on important feature information and ignore unimportant noise information. To further improve denoising effectiveness, we design a joint loss function. This loss function not only considers pixel level errors, but also takes into account structural similarity, so as to guide the network to more accurately recover image features during training. The experimental results show that the network can effectively suppresses speckle noise and preserves image details. It is superior to current state-of-the-art methods.