Water body segmentation utilizing Synthetic Aperture Radar (SAR) images plays a crucial role in water resource and flood monitoring. However, existing SAR water segmentation methods often underutilize the information present in the multiple bands of SAR images. Moreover, U-shaped semantic segmentation algorithms based on convolutional neural networks (CNN) often suffer from the loss of valuable feature information and the neglect of global and local correlations caused by the inclusion of pooling layers. To address these challenges, this letter first generates two novel bands from the VV and VH bands of the SAR images and combines them with the VV band to create a new SAR image as the input for the model. Then, to enhance the model’s capacity for feature learning and minimize the risk of overlooking small target water bodies, the letter proposed a water body segmentation model for SAR images that improves upon the UNet model by integrating coordinate attention mechanisms, CBAM and PVTv2. Finally, to tackle the issue of high computational complexity associated with the model, traditional convolutions are substituted with depthwise separable convolutions. The experimental results demonstrate that the proposed method achieves a 1.68% increase in accuracy, a 4.21% increase in MIOU, and a 3.51% increase in F-score when compared to the widely adopted UNet algorithm. Consequently, the proposed model surpasses other algorithms in the domain of water segmentation.

For high-resolution remote sensing image retrieval tasks, single-scale features cannot fully express the complexity of the image information. Due to the large volume of remote sensing images, retrieval requires extensive memory and time. Hence, the problem of how to organically fuse multi-scale features and enhance retrieval efficiency is yet to be resolved. We propose an end-to-end deep hash remote sensing image retrieval model (PVTA_MSF) by fusing multi-scale features based on the Pyramid Vision Transformer network (PVTv2). We construct the multi-scale feature fusion module (MSF) by using a global attention mechanism and a multi-head self-attention mechanism to reduce background interference and enhance the representation capability of image features. Deformable convolution is introduced to address the challenge posed by varying target orientations. Moreover, an intra-class similarity (ICS) loss is proposed to enhance the discriminative capability of the hash feature by minimizing the distance among images of the same category. The experimental results show that, compared with other state-of-the-art methods, the proposed hash feature could yield an excellent representation of remote sensing images and improve remote sensing image retrieval accuracy. The proposed hash feature can gain an increase of 4.2% and 1.6% in terms of mAP on the UC Merced and NWPU-RESISC45 datasets, respectively, in comparison with other methods.