| 引用本文: |
|
| Citation: |
|
摘要: 对视频序列离散余弦变换（DCT）系数统计分布进行了分析，提出了一种基于Weibull概率密度的视频DCT系数统计模型，该模型较Laplace概率密度与Cauchy概率密度更好地描述了视频序列的DCT特征.随后以该统计模型为基础，依据熵编码的理论分别给出了视频序列的速率-量化关系和失真-量化关系，并根据实际视频序列特性对其进行合理地简化近似，得到一种新的较为精确的视频序列率失真模型.经大量仿真实验证明，文中提出的基于Weibull分布的率失真模型对于Intra帧和Inter帧两种编码的真实视频序列率失真特性都具有较好的描述.
Abstract: According to analysis of discrete cosine transform （DCT） coefficients of video sequence, a new DCT coefficients statistical model based on Weibull probability density function（PDF） is introduced. Compared with Laplacian PDF and Cauchy PDF, the model can describe properties of DCT of video sequences much better. In addition, according to the new statistical model and theory of entropy coding, the rate quantization （RQ） and distortion quantization （DQ） functions are proposed separately. Based on the simplified and approximate RQ and DQ which are obtained by the theory of rate distortion （RD） model and the properties of real video sequence, the novel RD model is presented. The experimental results show that the real video sequence encoded by intra or inter can be described much better by using rate distortion model based on Weibull PDF.
下载:
