Abstract
With the development of convolutional neural network, video super-resolution algorithm has achieved remarkable success. Because the dependence between frames is complex, traditional methods lack the ability to model the complex dependence, and it is difficult to estimate and compensate the motion accurately in the process of video super-resolution reconstruction. Therefore, a reconstruction network based on optical flow residuals is proposed. In low resolution space, the dense residual network is used to obtain the complementary information of adjacent video frames, and then the optical flow of high-resolution video frames is predicted through the pyramid structure, and then the low resolution video frames are transformed into high-resolution video frames through the sub-pixel convolution layer, The high-resolution video frame is compensated with the predicted high-resolution optical flow. Finally, it is input into the super-resolution fusion network to get better effect. A new loss function training network is proposed to better constrain the network. Experimental results on public data sets show that the reconstruction effect is improved in PSNR, structural similarity and subjective visual effect.
Highlights
Super resolution reconstruction of a single image is a technique to recover high resolution image from low resolution image [1,2,3]
Kim et al [8] combined with residual network [9], proposed a deep convolution network for super resolution (VDSR), which solved this problem by accumulating characteristic graphs
The local information ignored by the video frames and the potential visual information of each frame in different sizes can be obtained, and the consistency and quality of the video can be enhanced by the fusion module, In order to build an end-to-end video super-resolution network framework, through the comparison with other algorithms, it is proved that our optical flow residual network can have advanced performance
Summary
Super resolution reconstruction of a single image is a technique to recover high resolution image from low resolution image [1,2,3]. Zhang et al [10] proposed a residual dense network (RDN), which can extract feature information more effectively and improve the quality of reconstruction. The method of channel splitting network (CSN) disperses the feature information in the sub network to reduce the learning burden of deep network and improve the training effect [12,13]. For super-resolution reconstruction, there is a problem of uncertainty, that is, for a low resolution image and video, it may be obtained by multiple high resolution. Xiaorong Zhang: Super resolution reconstruction algorithm of UAV image based on residual neural network images and video down sampling. (3) The example based multi-image super-resolution algorithm uses machine learning and deep learning methods to learn the nonlinear mapping between low resolution video frames and high resolution video frames. (4) This paper is based on the combination of reconstruction and example based method, through the deep learning method to learn the end-to-end nonlinear mapping relationship
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
