For the sake of removing the rain attached to rain images to restore the clarity of the images and due to the fact that existing rain removal methods cannot effectively remove rain from real-world images, a rain removal method for real-world images fusing deep learning and image enhancement is proposed. Firstly, a deep convolutional neural network based on supervised learning idea, multi-recursive LSTM and Spatial-Attention Mechanism is constructed to remove rain from real-world images. Then, the Rain Located and Filtered Algorithm is designed to further remove residual rain from derained images. Finally, the Visual Effect Improved Algorithm is proposed to improve the contrast of derained images and enhance the visual effect of derained images. The experimental results show that compared with the representative single image rain removal methods, the proposed method can not only effectively remove rain from real-world images, but also remove rain from synthetic images. As a result, the proposed method can make the processed images retain more detailed information and provide the better visual effect.

为了去除雨水图像中附着的雨水，恢复图像的清晰度，同时针对现有的雨水去除方法无法有效去除真实世界图像中的雨水，提出了一种融合深度学习的真实世界图像的雨水去除方法并提出图像增强。首先，构建了一个基于监督学习思想、多递归 LSTM 和空间注意机制的深度卷积神经网络来去除现实世界图像中的雨水。然后，雨水定位和过滤算法旨在进一步从去雨图像中去除残留的雨水。最后，提出了视觉效果改进算法来提高去雨图像的对比度，增强去雨图像的视觉效果。实验结果表明，与具有代表性的单幅图像去雨方法相比，所提出的方法不仅可以有效地去除真实世界图像中的雨水，还可以去除合成图像中的雨水。因此，所提出的方法可以使处理后的图像保留更详细的信息并提供更好的视觉效果。