Selective Deblocking for Focused Lenslet Video Coding

Lenslet video data captured by a focused plenoptic camera usually undergo a preprocessing step of cropping their micro images and concatenating them into micro image patches before being fed into a video encoder. However, subsequent direct encoding of such preprocessed lenslet video by conventional coding methods can cause a problem of undesirable blurred boundaries between micro image patches due to deblocking filtering. In this paper, we investigate on preventing such an undesirable blurring at the micro image boundaries. The experiment results show that under random access configuration, our proposed method not only outperforms the VTM-23.0 anchor by $1.24 \%$ in terms of the BDrate in luma channel but also reconstructs the lenslet video with better visual quality.