Back to search

University of Thessaly (UTH)

Parallelization and tile partitioning methods in video coding

Abstract

dc:description

The advent of 4K and 8K resolutions along with the rampant use of video content both in streaming and social media platforms render storage and transmission of video data a challenge of great scientific and commercial interest. As a matter of fact, the majority of data in internet traffic concerns the transmission of video data. This issue is tackled through the development of video codecs, which consist of formally defined specifications that establish shared methods that describe the encoding/decoding process, with the most successful coding standard being AVC/H.264, which was developed in 2004 and still enjoys wide use to this day. Its successor, HEVC/H.265 improved AVC’s compression efficiency to the range of 50%, while HEVC’s successor, VVC/H.266, manages to reduce bitrate to half for the same perceptual quality as its predecessor. The increase in performance that characterizes newer coding standards is due to the development of newer and more complex coding tools as well as the adaptation of tools present in older codecs in order to capture complex content and higher resolutions more accurately. However, while the newer coding standards offer great savings as far as coding overhead is concerned, they do so at the cost of an increased computational complexity. Motivated by this, this thesis harnesses the power of high level structures present in VVC, namely the Tile Structure, to reduce encoding time and computational complexity while maintaining the high coding performance that comes with the adoption of newer state of the art coding standards. The flexibility and power of the Tile Structure is harnessed through three aspects. Firstly, tiles along with other high level parallelization schemes supported in VVC, such as Wavefront Parallel Processing and CU Split, are utilized in developed algorithms to maximize speedup in a parallel system of homogeneous processing cores and improve upon scalability issues. Secondly, we tackle the aspect of content-aware tile partitioning in tandem with the aspect of pure speedup. The algorithm developed manages to derive a tilegrid that results in both better time performance and quality compared to uniform parallelization solutions. Lastly, we focused on traffic surveillance video data and utilized the Tile Structure to reduce bitrate and encoding time by deriving a tilegrid that maximizes the areas of a frame containing no objects. Tiles and CTUs according to their specific content are then encoded in an asymmetric manner in our scheme, where significant gains in coding efficiency and time performance are achieved with no adverse effects as it pertains to the object detection process that is of paramount importance to surveillance videos.

Degree

thesis:*
Grantor dc:publisher
University of Thessaly (UTH)
Year dc:date
2026

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Belememis, Panagiotis
  • Μπελεμέμης, Παναγιώτης

Subjects

dc:subject × 15

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/62925

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Belememis, Panagiotis; Μπελεμέμης, Παναγιώτης. Parallelization and tile partitioning methods in video coding. University of Thessaly (UTH), 2026. http://hdl.handle.net/10442/hedi/62925