Optimal multi-layer adaptation of SVC video over heterogeneous environments

Abstract

Scalable video coding (SVC) is a new video coding format which provides scalability in three dimensional (SNR-spatio-temporal) space. In this paper, we focus on the adaptation in SNR dimension. Usually, a SVC bitstream may contain multiple spatial layers, and each spatial layer may be enhanced by up to 3 FGS layers. To meet a bitrate constraint, the FGS data of different spatial layers can be truncated in various manners. However the contributions of FGS layers to the overall/collective video quality are different. In this work, we propose an optimized framework to control the SNR scalability across multiple spatial layers. Our proposed framework has the flexibility in allocating the resource (i.e. bitrate) among spatial layers, where the overall quality is defined as a function of all spatial layers' qualities and can be modified on the fly.

SVC adaptation: Standard tools and supporting methods

“…Each adaptation unit may contain one or more entities, which are the basic elements to be adapted. Depending on applications, entities can be pictures, temporal segments [19], spatial layers [20], regions of interest (ROI) [21,22], etc. Now, suppose that the adaptation unit to be adapted consists of N entities {E i , 1pipN}.…”

“…If MGS NUs of lower spatial layer(s) can be discarded, it will be possible to further reduce video bitrate without changing the spatiotemporal resolution. Moreover, when the QE data can be truncated simultaneously at multiple spatial layers, it is possible to have a flexible tradeoff between layers' qualities [20,26]. The drawback of multi-spatial-layer truncation is that layer dependency information should be modified to maintain the conformance of adapted bitstreams [37].…”

“…QCIF layer in Fig. 2), so the adaptation of discardable data can be easily integrated into the multi-layer adaptation framework of [20,26]. The only difference is that a discardable NU can have quality_id equal to 0.…”

“…In [15], a scheme for MPEG-4 FGS adaptation is proposed, where the operational R-D data of FGS frames are computed and stored in encoding process. In [20], the optimization of multi-layer adaptation is also based on R-D data of spatial layers which are obtained offline.…”

Optimal Rate Allocation for P2P Video Streaming

“…The main drawbacks of these and other approaches as [16] [17] [18] are that they require partial decoding of the bit-stream and that they are not flexible enough to allow for further constraints to be added to the optimization problem. A more flexible, but still computationally demanding, approach is proposed in [19]. In [20], the authors proposed an adaptation method based on distortion models for Medium Grain Scalability (MGS) introducing the possibility to control the quality between different groups of pictures.…”