Learning to remove reflections from windshield images

作者：

Highlights：

• Depth of field features benefit reflection removal for windshield images.

• Semantic constraints help preserve image contents in windshield reflection removal.

• Model trained by synthetic windshields generates well on real surveillance data.

摘要

•Depth of field features benefit reflection removal for windshield images.•Semantic constraints help preserve image contents in windshield reflection removal.•Model trained by synthetic windshields generates well on real surveillance data.

论文关键词：Reflection removal,Windshield image,Depth of field,Surveillance

论文评审过程：Received 9 February 2019, Revised 28 May 2019, Accepted 13 June 2019, Available online 14 June 2019, Version of Record 27 June 2019.

论文官网地址：https://doi.org/10.1016/j.image.2019.06.007