Abstract:

[Objective] Remote sensing images have become an important data source in fields such as surface observation, environmental monitoring, and natural disaster prediction. However, the acquisition of remote sensing images is often affected by weather phenomena such as fog and clouds, which reduces the image quality and poses challenges to subsequent analysis and processing tasks. In recent years, the introduction of attention mechanisms has enabled models to better capture and utilize important features in images, thereby significantly improving defogging performance. However, traditional channel attention mechanisms usually rely on global average pooling to summarize feature information. Although this method simplifies the complexity of calculations, it is not satisfactory when dealing with images with significant local changes and sensitivity to outliers. In addition, remote sensing images usually cover a wide area, and the diverse terrain makes the fog pattern more complex. Therefore, to address this issue, a hybrid attention-based generative adversarial network hybrid attention-based generative adversarial network (HAB-GAN) was proposed in this research, which integrates an efficient channel attention (ECA) module and a spatial attention block (SAB). [Method] By merging feature extraction from both channel and spatial dimensions, the model effectively enhanced its ability to identify and recover hazy areas in remote sensing images. In HAB-GAN, the ECA module captured local cross-channel interactions, addressing the shortcomings of traditional global averaged pooling in terms of insufficient sensitivity to local detail information. The ECA module used a global average pooling strategy without dimensionality reduction, automatically adapting to the characteristics of each channel without introducing extra parameters, thereby enhancing the inter-channel dependencies. ECA emploied a one-dimensional convolution operation, which used a learnable kernel size to adaptively determine the range of channel interactions. This design effectively avoided the over-smoothing of global features common in traditional pooling layers, allowing the model to more precisely extract local detailed while maintaining low computational complexity. The SAB module introduced a weighted mechanism on the spatial dimension by constructing a spatial attention map to enhance the model's ability to identify hazy areas in the image. This module extracted feature maps through convolution operations and applies attention weighting in both horizontal and vertical directions, highlighting regions with severe haze, allowing the model to better capture spatial information in the image, thereby enhancing dehazing performance. The generator of HAB-GAN combined residual network structures with hybrid attention modules. It first extracted initial features from input images through convolutional layers and then passed these features through several residual blocks. The residual blocks effectively mitigated the vanishing gradient problem in deep neural networks and maintain feature consistency and continuity by passing input features directly to deeper network layers through skip connections. Each residual block incorporated ECA and SAB modules, enabling precise feature learning through weighted processing in both channel and spatial dimensions. After extracting effective features, the generator generated dehazed images through convolution operations. The discriminator adopted a standard convolutional neural network architecture, focusing on extracting local detail features from the images generated by the generator. It consisted of multiple convolutional layers, batch normalization layers, and Leaky ReLU activation functions. By extracting local features layer by layer and down-sampling, the discriminator progressively reduced the spatial resolution of the images, evaluating their realism at both global and local levels. The generator and discriminator were jointly optimized through adversarial training, where the generator aimed to produce increasingly realistic dehazed images, and the discriminator continually improved its ability to distinguish between real and generated images, thereby enhancing the learning effectiveness and image quality of the generator. [Results and Discussions] To validate the effectiveness of HAB-GAN, experiments were conducted on the remote sensing image scene classification 45 (RESISC45) dataset. The experimental results demonstrated that compared to existing dehazing models, HAB-GAN excels in key evaluation metrics such as peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM). Specifically, compared to SpA GAN, HAB-GAN improved PSNR by 2.642 5 dB and SSIM by 0.012 2; Compared to HyA-GAN, PSNR improved by 1.138 dB and SSIM by 0.001 9. Additionally, to assess the generalization capability of HAB-GAN, further experiments were conducted on the RICE2 dataset to verify its performance in cloud removal tasks. The results showed that HAB-GAN also performs exceptionally well in cloud removal tasks, with PSNR improving by 3.593 2 dB and SSIM improving by 0.040 2. Compared to HyA-GAN, PSNR and SSIM increased by 1.854 dB and 0.012 4, respectively. To further explored the impact of different modules on the model's performance, ablation experiments were designed, gradually removing the ECA module, the SAB module, and the entire hybrid attention module. The experimental results showed that removing the ECA module reduced PSNR by 2.642 5 dB and SSIM by 0.012 2; Removing the SAB module reduced PSNR by 2.955 dB and SSIM by 0.008 7, and removing the entire hybrid attention module reduced PSNR and SSIM by 3.866 1 dB and 0.033 4, respectively. [Conclusions] The proposed HAB-GAN model not only performs excellently in dehazing and beclouding tasks but also significantly enhances the clarity and detail recovery of dehazed images through the synergistic effect of the ECA module and the SAB module. Additionally, its strong performance across different remote sensing datasets further validates its effectiveness and generalization ability, showcasing broad application potential particularly in fields such as agriculture, environmental monitoring, and disaster prediction, where high-quality remote sensing data is crucial. HAB-GAN is poised to become a valuable tool for improving data reliability and supporting more accurate decision-making and analysis.

Key words: remote sensing image, deep learning, generative adversarial network, efficient channel attention module, spatial attention module, defogging