Honggui Li

Title : Cswin2sr: Circular swin2sr for compressed image super-resolution

Abstract:

With the exponential growth of realistic and virtual image data, it is necessary to compress the original image for the purpose of highly-efficient image storage, transmission, processing, analysis, understanding and some emergent applications, such as augment reality and metaverse. In order to gain a high ratio of image compression, lossy image compression methods are preferentially considered. However, lossy image compression will unavoidably lead to the distortion to the original image. Hence, there is an urgent demand for compressed image restoration, including compressed image denoising, deblurring, deblocking, inpainting, artifact removing and super-resolution (SR). Classical compressed image SR methods utilized an open-loop architecture to obtain a high-quality image from a low-quality image. Closed-loop negative feedback mechanism is extensively utilized in automatic control systems and brings about extraordinary dynamic and static performance. In order to further improve the reconstruction capability of current methods of compressed image super-resolution, a circular Swin2SR (CSwin2SR) approach is proposed. The CSwin2SR contains a serial Swin2SR for initial super-resolution reestablishment and circular Swin2SR for enhanced super-resolution reestablishment. Simulated experimental results show that the proposed CSwin2SR dramatically outperforms the classical Swin2SR in the capacity of super-resolution recovery. On DIV2K valid dataset, the average increment of PSNR is greater than 0.18 dB and the related average increment of SSIM is greater than 0.004.

Audience Take Away Notes :

• The proposed closed-loop architecture can dramatically improve the performance, such as PSNR and SSIM, of existing compressed image super-resolution algorithms.
• The proposed closed-loop architecture is simple and easy to understand.
• The proposed closed-loop architecture can be efficiently integrated into any existing methods of compressed image super-resolution.

Biography:

Honggui Li received a B.S. degree in Electronic Engineering from Yangzhou University and received a Ph.D. degree in Mechatronic Engineering from Nanjing University of Science and Technology. He is a senior member of the Chinese Institute of Electronics. He is a visiting scholar and a postdoctoral fellow at the Institut Supérieur d'Électronique de Paris. He is an Associate Professor and postgraduate supervisor of Electronic Engineering at Yangzhou University. He is a reviewer for several international journals and conferences. He is the author of over 30 refereed journal and conference articles. His current research interests include machine learning and computer vision.