A Single Image Motion Deblurring Model of a Dynamic Scene

In recent years, convolutional neural network has achieved great success in the application of image deblurring, which has attracted more and more researchers' attention. At present, most of the image deblurring methods do not make full use of the hierarchical features of the original blurring image, so the performance is relatively low. This paper make full use of all level characteristics of convolution layer, introducing residual dense block, through intensive connection layer convolution extract rich in local features, puts forward a dynamic scenes based on residual dense blocks to motion blur model of multi-scale network, hereinafter referred to as residual multiple scales dense network (MSRDN) model, in the full extraction of partial feature of image at the same time, as far as possible to restore the higher quality of image information. The experimental results show that the proposed model performs better than similar algorithms in terms of peak signal to noise ratio (PSNR) and structural similarity (SSIM).