基于小波变换的MPEG-视频流预测模型

摘要

提出了一种基于小波变换的MPEG视频流的预测模型,该模型对原始数据做Haar小波变换以后,充分考虑尺度系数和乘子的不同统计特性,对尺度系数和乘子采用不同的预测模型分别预测,最终通过多重分形小波模型建模过程合成MPEG视频流序列.利用仿真和数学分析的方法对预测模型在刻画MPEG视频流的主要统计特性方面进行性能分析,从误差、数据的概率分布特性以及自相似特性等方面,证实了该预测模型的优越性能.%A prediction model of MPEG video traces based on wavelet transform is proposed. At first, the video traces are transformed by the wavelet transform. According to the different distribution features, we predict the scaling coefficients and multipliers by different models. Then the MPEG video traces are predicted by the model. Our simulation and analytic studies demonstrate the efficiency and excellent performance of the proposed model from different aspects, such as error, probability distribution, as well as self-similarity characteristics.