基于深度学习的非局部注意力增强网络图像去雨算法研究

单幅图像去雨技术的瓶颈问题是在缺少帧与帧时间序列信息的情况下，如何能够在有效去除多密度雨条纹的同时保留图像背景的细节结构信息.针对该问题，本文提出了一种新的基于编码解码器结构的单幅图像去雨算法.首先利用非局部操作获得不同像素点间的位置关系信息，从而获得图像全局信息表征.其次，采用空间注意力机制对全局信息在空间维度位置上进行权值重标定，即在通道维度上对特征进行非线性建模，从而达到聚集相似特征和有用信息的目的.最后，利用反卷积与长距离残差连接逐层恢复去雨图像的大小.分析和实验结果表明，本文提出算法雨痕去除效果明显，有效解决了去除具有不同雨密度大小雨条纹的现实困难，同时较好地保留图像的细节和边缘信息.

Image Raindrop Algorithm Research Using Nonlocal Attention Enhanced Network Based on Deep Learning

The bottleneck problem of single image de-raining is how to preserve the detailed structure information of image background with a lack of time series data between frames when removing the multi-density rain fringe.The new single image de-raining algorithm based on the coder and decoder structure is proposed in this paper.Firstly,the positional relationship information of various pixels between the points is obtained by non-local operation which can obtain the global information of the image representation.Secondly,the spatial attention is applied to recalibrate the global information in the spatial dimension position,and the channel features are nonlinearly modeled in the channel dimension to aggregate similar characteristics and useful information.Finally,the original size of rainy image is obtained by utilizing the de-convolution and long distance residual connection.The experimental results and analysis show that the proposed algorithm can obtain significant de-raindrop effect.The proposed algorithm can also resolve the difficulties of removing raindrops with various densities while maintaining the details of the image and edge information.