Search

Transcoding of H.264/AVC bitstreams is frequently used for adapting video streams to varying network conditions or different end-user device characteristics. Transcoding requires a relatively high complexity (and thus high operational cost) for each adaptation operation, and possibly reduces the video quality. In comparison, transcoding the input bitstream to SVC requires only one adaptation step. Afterwards, the resulting SVC bitstream can be adapted instantly. To reduce the H.264/AVC-to-SVC transcoding complexity, a transcoding architecture which combines existing open-loop transcoding techniques with optimized closed-loop transcoding is proposed. Furthermore, this approach allows to scale the complexity of the system. Doing so, compared with a cascaded decoder-encoder only 8.48% of the complexity is required for the highest complexity, while a high rate-distortion efficiency is maintained and a high degree of scalability is guaranteed. When only limited resources are available, the transcoding complexity can be scaled such that over 99% of the complexity is reduced compared to a cascaded decoder-encoder. [ABSTRACT FROM PUBLISHER]

Scalable video coding is an important mechanism to provide several types of end-user devices with different versions of the same encoded bitstream. However, scalable video encoding remains a computationally expensive operation. To decrease the complexity we propose generic techniques. These techniques are generic in a sense that they can be combined with existing fast mode decision methods and optimizations. We show that extending such an existing fast mode decision technique yields an average complexity reduction of 87.27%, while only an additional 0.74% of bit rate increase and a decrease of 0.11dB in PSNR is required, compared to the original fast mode decision method. [ABSTRACT FROM AUTHOR]