MPEG-21数字项适配技术研究

Research on MPEG-21 digital item adaptation technology

研究了MPEG-21数字项适配DIA(Digital Item Adaptation)框架及媒体格式独立(format-independent)的资源适配机制,结合H.264/AVC视频感知运动能量模型PME(Perceived Motion Energy),提出了一套基于MPEG-21 DIA的H.264/AVC视频适配模型和系统,并设计了多媒体资源适配引擎,提出了相应的资源适配决策算法,即在满足网络可用带宽的约束条件下,丢弃低感知质量层的视频帧,而尽可能多地保留高感知质量层的视频帧.实验结果表明,大部分测试视频序列在丢帧率平均达到50 %～60 %的时候,其用户主观感受质量评分仍可以达到可接受的程度(一般),而视频传输速率平均有24 %的减少.该系统不仅保证了视频适配后较高的用户视觉主观感知质量,而且实现了独立于媒体编码格式的视频适配机制,提高了多媒体适配系统的灵活性和可扩展性.

MPEG-21 digital item adaptation framework and format-independent media resource adaptation mechanism were studied. The perceived motion energy （PME） model was incorporated into the proposed H. 264/AVC video adaptation system based on MPEG-21 DIA framework, the adaptation engine for multimedia resources was designed. An adaptation decision algorithm was also presented which was realized by retaining most of the higher quality layers, and dropping lower quality layers as far as possible. The experiment results showed that most of the video sequences, with an average up to 50 % ～60 % of dropped frames, will get an acceptable grading （fair）, resulting in a corresponding reduction of an average of 24％ in the transmitting bit rate. The proposed system not only ensures the viewers＇ perceived qualities as high as possible but also realizes format-independent video adaptation mechanism and enhances the flexibility and scalability of multimedia adaptation system.