Asymmetric spatial scalability in stereoscopic video coding

Under many viewing circumstances, asymmetric stereoscopic video coding provides similar perceived quality with significant decrease in bitrate, computational complexity, and memory usage compared to conventional stereoscopic video coding. However, in other circumstances, users may prefer conventional symmetric stereoscopic video to asymmetric stereoscopic video. This paper presents a stereoscopic video coding scheme including asymmetric spatial scalability to provide the bitstream adaptation capability to meet heterogeneous network conditions, receiver devices, and user preferences. The spatial scalability property is enabled for the non-base view, and hence the coded bitstream can be decoded as conventional or mixed-resolution stereoscopic video. Simulation results show that the proposed method outperforms the tested other coding schemes providing the same scalability property by at least 8.2% in terms of average bitrate saving.

1. INTRODUCTION

Bitrate scalability of coded bitstreams is a desirable property in heterogeneous transmission environments where content is pre-encoded and adapted for the prevailing network conditions at the time of sending. Bitrate adaptation of scalable bitstreams can also be utilized in gateways or other media-aware network elements when adapting the content for the downlink throughput. Moreover, the scalability property is also useful when the computational capability or display resolution of the receiving device(s) or the preferences of receiving end-user(s) are not known at the time of encoding. Consequently, services and transmission schemes for which scalable video coding can be beneficial include multiparty video conferencing, IP and application- layer multicast, television broadcast, and such video file transfer and playback services where the properties of decoding devices are not accurately specified. Bitrate scalability can be realized as temporal, quality, and/or spatial scalability. The Scalable Video Coding (SVC) extension [1] of the Advanced Video Coding (H.264/AVC) standard [2] provides all three types of scalability. Single-loop decoding is enabled in SVC, i.e., only the highest layer present in the bitstream has to be decoded without reconstructing any of the lower layer decoded pictures. For the improvement of coding efficiency of enhancement layers, SVC includes an inter-layer prediction mechanism.