CAJ | 학술논문

网络丢包引发的视频失真是网络视频流媒体服务质量评价的研究重点。网络丢包引发的人眼视觉感受与丢包所在帧类型、视频帧的运动剧烈状态和纹理丰富程度有着显著关联，提出了一种运动矢量和DCT融合的无参考视频质量评价模型，以发生丢包的宏块为基本研究对象，以运动矢量表征其运动剧烈程度，以DCT直流系数刻画其纹理丰富程度，无需获取原始视频任何信息，无需完全解码，关联运动矢量和DCT完成视频质量的客观评价。实验结果表明，该模型的评价结果与主观评价结果有较好的一致性。
망락주포인발적시빈실진시망락시빈류매체복무질량평개적연구중점。망락주포인발적인안시각감수여주포소재정류형、시빈정적운동극렬상태화문리봉부정도유착현저관련，제출료일충운동시량화DCT융합적무삼고시빈질량평개모형，이발생주포적굉괴위기본연구대상，이운동시량표정기운동극렬정도，이DCT직류계수각화기문리봉부정도，무수획취원시시빈임하신식，무수완전해마，관련운동시량화DCT완성시빈질량적객관평개。실험결과표명，해모형적평개결과여주관평개결과유교호적일치성。
Network video distortion caused by packet loss is a research priority of online video streaming service quality evaluation. An associated quality assessment model of motion vector and DCT without reference video is provided. This model associates the human visual perception caused by network packet loss with the frame type whose packet is lost, the movement state and the richness of texture. This model takes the macro blocks that are lost as the basic study object. And it calculates the motion intenseness by using the macro blocks? motion vector and calculates the richness of its texture by using the DCT coefficients. And this model evaluates videos?quality objectively by associating the motion vector and DCT without any information of the original video and doesn?t need decoding completely. Experimental results show that this model's evaluation results have a good consistency with the subjective evaluation results.