摘要
讨论了如何在压缩域直接将 4∶ 2∶ 2采样结构的 MPEG码流转换为 4∶ 2∶ 0采样结构的码流。首先利用DCT变换的正交特性 ,推导出了在压缩域实现低通滤波运算的具体算法 ,在这种算法中 ,利用压缩域的矩阵乘法运算实现空间域的线性低通滤波运算。然后分析了在压缩域实现色度空间采样结构转换时转换失真的产生过程 ,定性地说明了转换失真产生的原因 ,并由此得出选取阶数低的滤波器有助于减小转换失真的结论。最后通过仿真实验验证了本文的方法能够取得较好的结果 ,尽管色度分量图像质量有一些损失 ,但大大减少了计算量。
How to convert 4∶2∶2 MPEG stream into 4∶2∶0 MPEG stream in the compressed domain is discussed. Firstly,the algorithm for accomplishing spatial domain low-pass filtering in the compressed domain is formulated based on the properties of the DCT. In this algorithm,using multiplication of matrix with 8×8 dimensions in the DCT domain implements spatial domain linear filtering. Secondly,after analyzing the reason to produce drift in the processing of conversion,low order filter is chosen for reducing filter processing drift in the compressed domain. Finally,the simulation validates that the approach can produce the reasonable image quality. Although the quality of the chroma picture is degraded a little,the computation complexity is reduced.
出处
《数据采集与处理》
CSCD
2004年第2期140-144,共5页
Journal of Data Acquisition and Processing
基金
上海市青年科技启明星计划 ( 0 1 QD1 40 3 0 )资助项目
