Semi-dense and robust image registration by shift adapted weighted aggregation and variational completion

作者：

Highlights：

• The disparity adapted weighted aggregation reduces the fattening effect.

• The multi-scale strategy highly decreases errors due to match ambiguities.

• Our local and global methods are robust to additive illumination changes.

• The proposed validation criteria rejects inaccurate estimations.

摘要

•The disparity adapted weighted aggregation reduces the fattening effect.•The multi-scale strategy highly decreases errors due to match ambiguities.•Our local and global methods are robust to additive illumination changes.•The proposed validation criteria rejects inaccurate estimations.

论文关键词：Image correspondences,Stereo,Optical flow,Block-matching,Interpolation

论文评审过程：Received 8 March 2019, Revised 13 June 2019, Accepted 20 July 2019, Available online 1 August 2019, Version of Record 23 August 2019.

论文官网地址：https://doi.org/10.1016/j.imavis.2019.07.005