Optimized segmentation of H.264/AVC video for HTTP adaptive streaming

HTTP adaptive streaming (HAS) is based on transmission of independently decodable segments of a video sequence. In this paper, we investigate the coding cost of segmentation in H.264/AVC coded videos and present an optimized segmentation framework. The proposed framework minimizes the coding cost associated with video segmentation. This is achieved by an optimized positioning of the Instantaneous Decoding Refresh (IDR) slices. In particular, a first-round coded bitstream is used to determine the frame positions where the temporal prediction modes fail to achieve high compression efficiency. In the second coding round, IDR frames are inserted on the determined positions to optimize the coding efficiency of the segmentation. In a second methodology, variable IDR frame rates for different video layers are investigated. Experimental results show that the proposed strategies yield moderate coding gain compared to the uniform segmentation used in current HAS implementations.