|
|
Laboratory workplace for virtual reality
Budáč, Adam ; Polák, Ladislav (oponent) ; Kufa, Jan (vedoucí práce)
This bachelor’s thesis is about creation of laboratory workplace for virtual reality. In theoretical part, it’s analysis of various types of devices for use in VR, description of operation principle of these devices used in VR, analysis of modern coding algorithms suitable for use in VR, suitable use of these codecs for 2D, 3D a 360° multimedia content, options of obtaining and playback of this content, comparing the difference against classical displaying methods and quality measurement of this content using objective metrics at different qualities and compressions. In experimental part, there’s described database creation of video content, coding these video sequencies using objective metrics, creation of laboratory workplace for use with VR, creation and sample elaboration of laboratory task dedicated to virtual reality.
|
|
|
Deep Learning based compression of 360° images
Budáč, Adam ; Boleček, Libor (oponent) ; Kufa, Jan (vedoucí práce)
This diploma thesis deals with 360° image compression based on deep learning. The thesis describes the representation and deformations of a 360° image. It then describes conventional compression methods and methods using machine learning, deep learning and deep neural networks for 360° image compression. Part of the work is the creation of a dataset, which consists of ten 360° images, and the design of a framework that enables image compression using conventional codecs and algorithms based on machine learning, deep learning, and deep neural networks. Images from the dataset are compressed using five conventional codecs and four deep learning algorithms. The quality of compressed images is measured using seven objective metrics and one subjective test. As a result of the experiment, the conventional methods achieved higher compression quality than the methods using deep learning, and the computational complexity of the conventional methods is lower compared to the methods using deep learning.
|
|
|
Comparison Of Modern Video Codecs For 4k And 8k Video Content
Budáč, Adam
The purpose of this paper is to compare modern video codecs, that ensure delivery of 4Kand 8K video content in high quality at reasonable bitrates. There are compared three video codecs:HEVC, VVC and the AV1. Video quality assessment is ensured by using four objective video qualitymetrics, which are: PSNR, SSIM, VIF and VMAF. For tests, there were chosen two from total ofseven videosequencies with different parameters: very high resolutions, high framerates, wide rangeof spatial and temporal indexes and 3D and 360° projection.
|
|
|
Laboratory workplace for virtual reality
Budáč, Adam ; Polák, Ladislav (oponent) ; Kufa, Jan (vedoucí práce)
This bachelor’s thesis is about creation of laboratory workplace for virtual reality. In theoretical part, it’s analysis of various types of devices for use in VR, description of operation principle of these devices used in VR, analysis of modern coding algorithms suitable for use in VR, suitable use of these codecs for 2D, 3D a 360° multimedia content, options of obtaining and playback of this content, comparing the difference against classical displaying methods and quality measurement of this content using objective metrics at different qualities and compressions. In experimental part, there’s described database creation of video content, coding these video sequencies using objective metrics, creation of laboratory workplace for use with VR, creation and sample elaboration of laboratory task dedicated to virtual reality.
|
host ::
RSS.