Real-time interactive regions of interest in H.264/AVC

Abstract

The concept of regions of interest (ROIs) within a video sequence is useful for many application scenarios. This paper concentrates on the exploitation of ROI coding within the first version of the H.264/AVC specification, for which it was already shown in literature that the flexible macroblock ordering (FMO) tool can be used to achieve ROIs in H.264/AVC video streams. We extend the existing methods with two approaches in order to better match the denotation of ROI scalability. The first approach allows to change the size of the output video pane while the second approach makes it possible to select an ROI at run time without the need for an encoder to provide that specific ROI in the bitstream. It is shown that both approaches allow for real-time adaptation of H.264/AVC bitstreams. Measurements also show that significant bit rate savings can be achieved when performing ROI-based adaptation, that the decoding speed is positively affected, and that the coding overhead can be controlled.