基于块编码特点的压缩视频质量增强算法

Compressed Video Quality Enhancement Method Based on Block Coding Features

针对现有压缩视频质量增强算法未能充分利用压缩视频特点的问题,研究了视频编码与压缩视频质量增强任务之间的本质关系,并针对性地设计了一种基于三维卷积神经网络(3D convolutional neural network, 3D-CNN)的非对齐压缩视频质量增强算法。实验结果表明:相较于高效视频编码(high efficiency video coding, HEVC)标准H.265,所提算法在低延迟(low delay, LD)配置下且量化参数(quantization parameter, QP)为37时,峰值信噪比(peak signal-to-noise ratio, PSNR)提升了0.465 2 dB;相较于数据压缩会议(data compression conference, DCC)中提出的多帧引导的注意力网络(multi-frame guided attention network, MGANet)方法,该算法PSNR的增长量提升了15.1%。

To solve the issue that existing compressed video quality enhancement algorithms do not fully utilize the characteristics of compressed videos,the intrinsic relationship between video encoding and the task of compressed video quality enhancement was studied and a targeted non-aligned compressed video quality enhancement algorithm was designed contrapuntally,utilizing a three-dimensional convolutional neural network(3D-CNN).Experimental results show that compared with the high efficiency video coding(HEVC)standard,the peak signal-to-noise ratio(PSNR)of the proposed method is improved to 0.4652 dB when low delay(LD)configuration and quantization parameter(QP)is 37.Compared with MGANet proposed in data compression conference(DCC),the PSNR increase of the proposed algorithm is improved by 15.1%.