GPU-Based MPEG-2 to Secure Scalable Video Transcoding

Abstract

Most of the high definition video content are still produced in a single-layer MPEG-2 format. Multiple-layers Scalable Video Coding (SVC) offers a minor penalty in rate-distortion efficiency when compared to single-layer coding MPEG-2. A scaled version of the original SVC bitstream can easily be extracted by dropping layers from the bitstream. This paper proposes a parallel transcoder from MPEG-2 to SVC video with Graphics Processing Unit (GPU), named PTSVC. The objective of the transcoder is to migrate MPEG-2 format video to SVC format video such that clients with different network bandwidth and terminal devices can seamlessly access video content. Meanwhile, the transcoded SVC videos are encrypted such that only authorized users can access corresponding SVC layers. Using various scalabilities SVC test sequences, experimental results on TM5 and JSVM indicate that PTSVC is a higher efficient transcoding system compared with previous systems and only causes little quality loss.