Deep view synthesis with compact and adaptive Multiplane Images

作者：

Highlights：

• A learning-based method that computes compact and adaptive MPIs.

• The network promotes sparsity in the MPI to only keep the essential scene information to render views of high quality.

• The depth sampling is adapted to the given scene to optimize the available memory and increase the synthesis quality.

• The proposed method does not need individual training per scene.

摘要

•A learning-based method that computes compact and adaptive MPIs.•The network promotes sparsity in the MPI to only keep the essential scene information to render views of high quality.•The depth sampling is adapted to the given scene to optimize the available memory and increase the synthesis quality.•The proposed method does not need individual training per scene.

论文关键词：View synthesis,Multiplane Image,Deep learning

论文评审过程：Received 30 September 2021, Revised 8 March 2022, Accepted 1 June 2022, Available online 8 June 2022, Version of Record 17 June 2022.

论文官网地址：https://doi.org/10.1016/j.image.2022.116763