一种非对称的轻量级图像盲去模糊网络

An Asymmetric Lightweight Image Blind Deblurring Network

针对现有图像去模糊算法存在细节模糊不清、计算资源占用较大且图像处理速度较慢等问题,提出一种轻量级的图像盲去模糊网络.首先,网络主体使用多尺度架构,将不同分辨率的图像输入网络,通过循环处理逐步优化细节;其次,设计非对称结构以加强编码器的特征提取能力和解码器的特征融合能力.在编码器中,提出混合多尺度卷积层和残差金字塔模块,以强化特征提取并减少网络的参数量;在解码器阶段,使用跳跃连接引入深层语义,并提出多尺度联合结构损失函数进行优化.最后,在两个广泛使用的数据集GoPro和Kohler上使用两种评价指标,将该方法与其他经典方法进行性能对比.实验结果表明,该网络的去模糊效果优于传统方法以及其他经典深度学习方法,不仅在峰值信噪比(PSNR)和结构相似性(SSIM)上均有一定提升,且处理时间更短.

Aiming at the problems of blurred details,large computer resource occupation,and slow image processing for the existing image deblurring algorithms,we proposd a lightweight image blind deblurring network.Firstly,the main framework of the network used a multi-scale architecture to input images of different resolutions into the network,and gradually optimized the datails through cyclic processing.Secondly,the asymmetric structure was designed to enhance the feature extraction ability of the encoder and the feature fusion ability of decoder.In the encoder,the mixed multi-scale convolutional layer and residual pyramid module were proposed to enhance feature extraction and reduce the number of network parameters.In the decoder stage,deep semantics were introduced by using jump linkage,and the multi-scale joint structure loss function was proposed for optimization.Finally,we used two evaluation indicators to compare the performance of the method with the other classical methods on two widely used GoPro and Kohler datasets.The experimental results show that the effect of the network is better than that of the traditional methods and other classical deep learning mehtods.It not only improves the peak signal-to-noise ratio(PSNR)and structural similarity(SSIM),but also shortens the processing time.