关键词: 可分级视频编码, 空间分级, 帧间模式选择, 层间残差预测, 率失真代价

Abstract: Inter-Layer Residual Prediction (ILRP) in Scalable Video Coding (SVC) reduces the number of bits while significantly increasing the encoding complexity. To reduce the complexity, the paper proposed a fast mode decision algorithm. This algorithm analyzed the difference of Lagrangian Rate-Distortion cost (RDCost) when using residual prediction and not using residual prediction in enhancement layer. According to the difference ILRP was dynamically selected in the proposed algorithm. Meanwhile inter mode correlation between lower spatial layer and higher spatial layer was also analyzed. Based on the correlation the proposed algorithm used the best inter mode in lower layer to decide the best inter mode in higher layer and this would save more encoding time. The experimental results show that when compared with the algorithm in Joint Scalable Video Model (JSVM), the proposed algorithm can save more than 50% encoding time on average with neglected Peak Signal to Noise Ratio (PSNR) change and small bit-rate loss. The proposed algorithm provides valuable references for further study of optimizing encoder.

Key words: Scalable Video Coding (SVC), spatial scalability, inter prediction mode decision, Inter-Layer Residual Prediction (ILRP), rate-distortion cost