摘要
现有的比较成熟的可伸缩编码框架是基于运动补偿时域滤波技术的。改进了运动补偿时域滤波中的更新操作,并引入去块滤波。更新操作过程中,要求更新块中的像素所对应的预测运动向量完全一致,否则不参与更新操作。更新操作前,对残差图像在边缘级和采样点级进行滤波控制。通过对高通信号的去块滤波,减少更新块中的伪边界成分,同时减轻块效应对后继预测操作的影响,改善编码效率。实验结果表明该方法能够改进图像的主观质量,提高低运动量的图像序列的编码性能。
Currently, the most recognized framework for scalable video coding is based on Motion Compensation Temporal Filtering （MCTF）. Some improvements on deblocking filter and update step for MCTF were proposed. During the update process, the constraint was requested that the pixels of the update block had the uniform prediction motion vector. Before update process, filtering residual image was performed in both edge and sample dot. The high-passed frames were deblocked to bring about less block artifacts into the low-pass frames, and better reference frames for the following motion compe.nsation. In the mean time, the deblock of the high-passed frames could also improve the coding efficiency for the low-activated video sequences. Experimental results show that the coding efficiency of the low-pass frames and the subject quality of frames are improved.
出处
《计算机应用》
CSCD
北大核心
2007年第11期2778-2780,共3页
journal of Computer Applications
基金
教育部科学技术研究重点项目(105047)
霍英东教育基金会优选资助课题(104019)
全国优秀博士学位论文作者专项资金资助项目(200439)
关键词
运动补偿时域滤波
更新操作
去块滤波
可伸缩编码
Motion Compensated Temporal Filtering （MCTF）
update step
deblocking filtering
scalable video coding
