An efficient framework for image/video inpainting

作者：

Highlights：

• To provide the theory analysis in transform domain for image inpainting.

• To summarize merits of spatial transforms: local coherence, area and boundary.

• To save traditional geometric methods nearly 93% run time in average.

• To improve the PSNR by up to 5.10 dB compared with traditional methods.

摘要

Author-Highlights•To provide the theory analysis in transform domain for image inpainting.•To summarize merits of spatial transforms: local coherence, area and boundary.•To save traditional geometric methods nearly 93% run time in average.•To improve the PSNR by up to 5.10 dB compared with traditional methods.

论文关键词：Image inpainting,Image interpolation,Image restoration,PDE,Diffusion function

论文评审过程：Received 11 September 2012, Revised 29 January 2013, Accepted 11 March 2013, Available online 8 April 2013.

论文官网地址：https://doi.org/10.1016/j.image.2013.03.002