摘要
论述了在压缩域内将 MPEG- 2码流中的 B帧或 P帧转换为 I帧的方法 .根据 DCT变换的正交特性 ,在将运动补偿运算转换为矩阵相乘运算后 ,推导出压缩域帧类型转换的计算公式 .针对采用半像素精度运动估计的帧 ,对相应公式做了修正 .由于量化阶的选择极大影响转换过程中的失真程度和码率的大小 ,提出了一种重新计算量化阶的方法 ,利用压缩域的信息计算宏块的活动性 ,并由该活动性信息调整量化阶的大小 .仿真实验验证了所提出方法的有效性 .
An approach to convert B frame or P frame of MPEG-2 bitstream into I frame in the compressed domain was presented. Based on the orthogonal properties of the DCT, an algorithm for changing frame type was deduced. Because the de-quantized step is very important in the processing of frame type changing, a method which uses an activity of macroblock in the compressed domain to select quant scale was presented. The simulation suggests that this approach can produce a reasonable image quality.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2003年第8期1293-1296,共4页
Journal of Shanghai Jiaotong University
基金
上海市青年科技启明星计划资助项目 (0 1QD14 0 30 )
关键词
压缩域
帧类型转换
量化阶
Digital television
Image compression
Motion estimation
Signal to noise ratio