Residual-guided In-loop Filter Using Convolution Neural Network

The block-based coding structure in the hybrid video framework inevitably introduces compression artifacts such as blocking, ringing, and so on. To compensate for those artifacts, extensive filtering techniques were proposed loop of codecs, which are capable boosting subjective objective qualities reconstructed videos. Recently, neural network-based filters presented with power deep learning from a large magnitude data. Though efficiency has been improved traditional methods High-Efficiency Video Coding (HEVC), rich features information generated by pipeline have not fully utilized design networks. Therefore, this article, we propose Residual-Reconstruction-based Convolutional Neural Network (RRNet) to further improve its full extent, where induced bitstream form prediction residual fed into network an additional input frame. In essence, signal can provide valuable about block partitions aid reconstruction edge texture regions picture. Thus, more adaptive parameters be trained handle different characteristics. experimental results show that our RRNet approach presents significant BD-rate savings compared HEVC state-of-the-art CNN-based schemes, indicating plays role enhancing frame reconstruction.