Abstract
In recent years, the common algorithms for image super-resolution based on deep learning have been increasingly successful, but there is still a large gap between the results generated by each algorithm and the ground-truth. Even some algorithms that are dedicated to image perception produce more textures that do not exist in the original image, and these artefacts also affect the visual perceptual quality of the image. We believe that in the existing perceptual-based image super-resolution algorithm, it is necessary to consider Super-Resolution (SR) image quality, which can restore the important structural parts of the original picture. This paper mainly improves the Enhanced Super-Resolution Generative Adversarial Networks (ESRGAN) algorithm in the following aspects: adding a shallow network structure, adding the dual attention mechanism in the generator and the discriminator, including the second-order channel mechanism and spatial attention mechanism and optimizing perceptual loss by adding second-order covariance normalization at the end of feature extractor. The results of this paper ensure image perceptual quality while reducing image distortion and artefacts, improving the perceived similarity of images and making the images more in line with human visual perception.
Highlights
Image super-resolution reconstruction converts low-resolution images into high-resolution images to obtain images as close as possible to real images
Considering the dependencies between feature maps, we introduce a second-order channel attention mechanism and self-attention mechanism on the generator and the discriminator, so that the network focuses on more informative parts and improves the network’s expressive ability and discriminative ability, which more accurately restrain pictures generated by the generation network
RankSRGAN uses indicators such as perceptual index (PI), NIQE, and Ma’s score, that are more consistent with human visual perception as optimization goals
Summary
Image super-resolution reconstruction converts low-resolution images into high-resolution images to obtain images as close as possible to real images. After Dong et al pioneered SRCNN [1], image super-resolution reconstruction algorithms based on neural networks emerged in an endless stream and achieved remarkable results. These algorithms include the image distortion-driven algorithms (i.e., Peak-Signal-to-Noise Ratio (PSNR) value) [1,2,3,4,5,6], and perception-driven image super-resolution algorithms [7,8,9,10,11,12], The distortion-based network structure makes the restored image too smooth, losing considerable high-frequency information and texture information, which does not correspond with human visual perception [9]. It can be seen that the experimental results in this paper are closer to the structure of the original picture, reduce the generation of other textures, and correspond better with human visual perception
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.
