Abstract: By analyzing the data partitioning tool in H.264, a novel data partitioning method based on the content of a picture is presented. It regroups video bitstreams into three separate sub-bitstreams according to importance: header information, data of Intra-macroblocks and part of Inter-macroblocks, data of the remainder of Inter-macroblocks. The two parts of Inter-macroblocks belonging to different sub-bitstreams are differentiated according to their impacts on the quality of the picture, and the number is decided by an optimal algorithm. Simulation results show that compared with the data partitioning tool in H. 264, the proposed method is more adaptive to the change of picture content and network conditions. Together with unequal error protection, it can improve the quality of video streaming to a great extent.

Key words: video streaming, data partitioning, error resilience, picture content, unequal error protection