Neighborhood-Based Weighted Regularization of Video Sequence Super-Resolution

This paper presents a neighbourhood-based weighted super-resolution approach that estimates a high resolution video sequence from the low resolution video sequence. Firstly, In order to reduce artifacts in the super-resolved outcome due to the inaccurate estimation in the non-global motion fields, a neighbourhood-based weighted functional in terms of local mean residual is used to weight each low resolution channel. Secondly, a locally adaptive regularization functional based on the local mean residual is determined within each low resolution channel instead of the overall regularization parameter. The proposed approach has significantly improved performance in both global motion based image sequence and video sequences containing complex motions. Experimental results indicate the obvious performance improvement in both PSNR and visual effect compared to non-channel-weighted method and overall-channel-weighted method.