Mixed neighborhood topology cross decoded patterns for image-based face recognition

作者：

Highlights：

• Proposition of a new LBP like descriptor for face recognition referred to as MNTCDP.

• Evaluating the proposed method on ORL, FERET, YALE, Extended Yale B and AR databases.

• Providing a fair comparison between the proposed MNTCDP descriptor and 51 LBP variants.

• The proposed MNTCDP based system outperforms several recent state-of-the-art systems.

• The performance of MNTCDP descriptor is compared to the one of deep learning methods.

摘要

•Proposition of a new LBP like descriptor for face recognition referred to as MNTCDP.•Evaluating the proposed method on ORL, FERET, YALE, Extended Yale B and AR databases.•Providing a fair comparison between the proposed MNTCDP descriptor and 51 LBP variants.•The proposed MNTCDP based system outperforms several recent state-of-the-art systems.•The performance of MNTCDP descriptor is compared to the one of deep learning methods.

论文关键词：Face recognition,Feature extraction,Face descriptor,LBP,Classification,MNTCDP

论文评审过程：Received 26 December 2017, Revised 25 April 2018, Accepted 16 July 2018, Available online 18 July 2018, Version of Record 31 July 2018.

论文官网地址：https://doi.org/10.1016/j.eswa.2018.07.035