Data-driven image completion for complex objects
作者：
Highlights：
•
• A novel data-driven image completion method is proposed for complex objects.
• The degraded image is reconstructed by coarse-to-fine object and patch matching.
• A multi-attribute matching measure is presented to retrieve the compatible object.
• The degraded region is fixed by patch-based contour propagation and region filling.
• The effectiveness of our method is verified through extensive experiments.
摘要
Highlights • A novel data-driven image completion method is proposed for complex objects. • The degraded image is reconstructed by coarse-to-fine object and patch matching. • A multi-attribute matching measure is presented to retrieve the compatible object. • The degraded region is fixed by patch-based contour propagation and region filling. • The effectiveness of our method is verified through extensive experiments.
论文关键词：Image inpainting,Data-driven,Object matching,Exemplar-based
论文评审过程：Received 11 November 2016, Revised 25 April 2017, Accepted 25 April 2017, Available online 27 April 2017, Version of Record 6 May 2017.
论文官网地址：https://doi.org/10.1016/j.image.2017.04.011
