| dc.description.abstract | Due to the increasing number of new services and devices that allow the creation, distribution and consumption of video content, the amount of video information being transmitted all over the world is constantly growing. Video compression technology is essential to cope with the ever increasing volume of digital video data being distributed in today's networks, as more e cient video compression techniques allow support for higher volumes of video data under the same memory/bandwidth constraints. This is especially relevant with the introduction of new and more immersive video formats associated with signi cantly higher amounts of data. In this thesis, novel techniques for improving the e ciency of current and future video coding technologies are investigated. Several aspects that in uence the way conventional video coding methods work are considered. In particular, the properties and limitations of the Human Visual System are exploited to tune the performance of video encoders towards better subjective quality. Additionally, it is shown how the visibility of speci c types of visual artefacts can be prevented during the video encoding process, in order to avoid subjective quality degradations in the compressed content. Techniques for higher video compression e ciency are also explored, targeting to improve the compression capabilities of state-of-the-art video coding standards. Finally, the application of video coding technologies to practical use-cases is considered. Accurate estimation models are devised to control the encoding time and bit rate associated with compressed video signals, in order to meet speci c encoding time and transmission time restrictions. | en_US |
