15 March 2022 in Digital Signal Processing
1 May 2021 in Programming and Computer Software
In recent years, the field of image inpainting has developed rapidly, learning based approaches show impressive results in the task of filling missing parts in an image. But most deep methods are strongly tied to the resolution of the images on which they were trained. A slight resolution increase leads to serious artifacts and unsatisfactory filling quality. These methods are therefore unsuitable for interactive image processing. In this article, we propose a method that solves the problem of inpainting arbitrary-size images. We also describe a way to better restore texture fragments in the filled area. For this, we propose to use information from neighboring pixels by shifting the original image in four directions. Moreover, this approach can work with existing inpainting models, making them almost resolution independent without the need for retraining.
1 January 2022 in Keldysh Institute Preprints
This paper compares the effectiveness of different projections for different 360-degree videos and codecs, including AV1, HEVC, VP9, and AV. Content for 360 video is especially challenging because using codecs on reprojections magnifies video-stream errors. For this reason, we tested different projections to determine their stability over different encodings. Our results came from performance measurements using end-to-end metrics such as PSNR, SSIM and VMAF as well as subjective assessment of static viewports.