提升小波运动补偿时域滤波可伸缩视频编码

Scalable Video Coding of Lifting Wavelet Motion Compensation Temporal Filter

为了改善传统可伸缩性视频编码方法效率不高的问题,将提升小波应用于运动补偿时域滤波中。利用5/3提升小波进行时域滤波,针对像素的连接类型不同采取不同的处理方式,并根据运动补偿效果自适应选取时域分解级数。实验结果表明,该算法不仅实现了码率良好的可伸缩性,而且与MPEG-4(Moving Pictures Expert Group-4)和3DWT(3 Dimensions Wavlet Transformation)方法相比,在相同码率的条件下,重建图像的信噪比提高了1～2 dB,从而有效地提高了编码效率。

In order to improve the traditional scalable video coding performance, the lifting wavelet is used in MCTF （Motion Compensated Temporal Filtering）. The 5/3 lifting wavelet is carried on the temporal filter. Different motion-handled modes are adopted to deal with pixels with different connected states, and according to motion compensation effect the progression of temporal domain decomposition is auto-adapted. The experimental results show that the algorithm achieves a rate of scalability, and the PSNR is 1 -2 dB better than MPEG-4 （Moving Pictures Expert Group-4） and 3DWT （3 Dimensions Wavlet Transformation） with the same compensation rate, and effectively enhance the code performance.