Heindel A (2018)
Publication Language: English
Publication Type: Thesis
Publication year: 2018
Publisher: Friedrich-Alexander-Universität Erlangen-Nürnberg
City/Town: Errlangen; Nürnberg
URI: https://nbn-resolving.org/urn:nbn:de:bvb:29-opus4-110590
Video communication applications are prevalent in our digital society. Smart phones and other mobile devices like tablet computers are equipped with both high-resolution displays and cameras, which allow digital video communication almost whenever and wherever desired. Leading-edge technologies from multiple fields like digital communications and digital signal processing jointly enable high quality video applications as we know it today. However, capabilities like processing power, energy supply, display size, or the supported video formats of the devices in the field are very heterogeneous. Moreover, also the network conditions may vary significantly. This challenge is tackled by scalable video coding, which creates bit stream layers of the same content but with different properties, denoted as base layer and enhancement layer. Depending on the network and receiver device capabilities, only the base layer or both layers may be transmitted and decoded. However, scalable video coding with the possibility for lossless reconstruction of individual pictures is not optimally supported by state-of-the-art approaches. Additionally, the specific scalability of the color resolution is not possible by the state of the art. Consequently, two scalable coding schemes that support these requirements are proposed in this thesis. The first scalable coding framework encodes the residual signal of the base layer in a lossless manner. Pixel-wise intra picture prediction is applied, that is efficient for lossless coding and retains the individual access to single pictures. The experimental results showed that there is no bit rate overhead for the scalable representation compared to state-of-the-art direct lossless coding with only one layer, when intra picture coding is considered. Exploiting temporal correlations by means of inter picture coding of the base layer allows to unite the advantages of both approaches, base layer video coding optimized for lossy compression and lossless enhancement layer coding. An extension of this approach supporting near-lossless coding is also presented in this thesis and outperforms the state of the art in most cases. The second enhancement layer codec proposed in this thesis allows to reconstruct the high-resolution color format when the base layer is compressed using a reduced color component resolution. Experiments showed that scalable coding can increase the average quality in terms of the peak signal-to-noise ratio by 1.82 dB compared to base layer coding only. Additionally, a novel metric to measure the artifact level originating from the reduced color component resolution is introduced in this context.
APA:
Heindel, A. (2018). Methods for Enhancement Layer Compression in Two-Layer Quality Scalable Video Coding (Dissertation).
MLA:
Heindel, Andreas. Methods for Enhancement Layer Compression in Two-Layer Quality Scalable Video Coding. Dissertation, Errlangen; Nürnberg: Friedrich-Alexander-Universität Erlangen-Nürnberg, 2018.
BibTeX: Download