View-spatial–temporal post-refinement for view synthesis in 3D video systems
作者：
Highlights：
• Propose a post-refinement method to improve the rendering quality of synthesized image.
• Propose hole filling and boundary artifacts removal technologies.
• View-spatial–temporal correlations of mutiview video are efficiently used to improve the rendering quality.
• A good trade-off between complexity and rendering quality is achieved in reference frame selection.
• Both subjective and objective qualities of rendered images are significantly improved.
摘要
•Propose a post-refinement method to improve the rendering quality of synthesized image.•Propose hole filling and boundary artifacts removal technologies.•View-spatial–temporal correlations of mutiview video are efficiently used to improve the rendering quality.•A good trade-off between complexity and rendering quality is achieved in reference frame selection.•Both subjective and objective qualities of rendered images are significantly improved.
论文关键词：Multiview video,View synthesis,Depth image based rendering,Hole-filling,Boundary artifact
论文评审过程：Received 11 March 2013, Revised 15 August 2013, Accepted 15 August 2013, Available online 29 August 2013.
论文官网地址：https://doi.org/10.1016/j.image.2013.08.005
