基于内容统计分析的高清视频SVC快速模式选择算法

摘要

Scalable video coding (SVC) greatly enhances the flexibility of video transmission and decoding. However,SVC also increases the encoding complexity significantly,making real-time implementation difficult. To address this problem we propose a fast mode decision algorithm to speed up the SVC encoder. Using the statistical mode distribution of the enhancement layer,the proposed method evaluates the complexity of the video content,and then sets the rate distortion RD cost thresholds for inter coding modes. Results are shown using four resolution layers (1080p,720p, VGA and QVGA),two temporal layers and two quality layers. It is found that the proposed algorithm can reduce the computational complexity signiifcantly while maintaining the encoder efifciency.%可伸缩编码（SVC）极大地增强了视频传输和解码的灵活性，而同时也极大地增加了编码的复杂度，使得实时应用变得更为困难。为此，本文提出了一个快速模式选择算法以加快编码器的速度。本算法首先统计增强层宏块的模式分布情况，然后估计视频序列的复杂度，并以此为根据为帧间最佳模式设置率失真代价门限。算法使用4个分辨率层（1080p，720p，VGA and QVGA），4个时域层，以及两个质量层进行测试，结果表明本算法能够在保持现有编码效率基本不变的情况下有效提高编码速度。