视频压缩中环效应的分析与消除

分析了降低视频主观质量的环效应的成因 ,澄清了边缘增强导致环效应的误解 ,认为环效应是由于图像中的锐利边缘所包含的高频分量在量化过程中受到过度损失引起的 ,进而提出基于宏块的在编码器端消除环效应的算法。该算法根据图像中各个宏块分别含有锐利边缘的多少来调节每个宏块的量化精度。在 MPEG- 2编解码器上的实验结果表明 ,该算法在保持视频平均码率和客观质量几乎不变的情况下可以有效地消除环效应 ,提高视频主观质量。该算法无需修改视频编码标准的码流语法 ,因此便于在实际应用中推广

Analysis and reduction of the ringing effect in video coding

The ringing effect which degrades video subjective quality was analyzed to counter the misunderstanding that edge enhancement leads to the ringing effect. Ringing effect is indeed due to the loss of high-frequency components during quantization with those components contained by the sharp edges in the image. A macroblock-based algorithm was developed to reduce the ringing effect at the encoder end. The algorithm adjusts the quantization precision of each macroblock according to the number of sharp edges in the block. The experimental results on an MPEG-2 codec show that the algorithm efficiently reduces the ringing effect and improves video subjective quality while keeping the video average bitrate and video objective quality almost unchanged. The algorithm does not need to modify the bitstream syntax of existing video coding standards, which makes it appropriate for a wide range of practical applications.