Robust joint representation with triple local feature for face recognition with single sample per person

作者：

Highlights：

• RJR-TLF is proposed for FR with SSPP, making innovations in both feature extraction and classifier design.

• Triple local feature exploits the robustness and discrimination of local scale, orientation and space of the face image.

• Robust joint representation jointly combines local features with their weights adaptively distributed to further enhance the robustness.

• The proposed RJR-TLF is evaluated extensively on popular databases with promising results.

摘要

•RJR-TLF is proposed for FR with SSPP, making innovations in both feature extraction and classifier design.•Triple local feature exploits the robustness and discrimination of local scale, orientation and space of the face image.•Robust joint representation jointly combines local features with their weights adaptively distributed to further enhance the robustness.•The proposed RJR-TLF is evaluated extensively on popular databases with promising results.

论文关键词：Triple local feature,Robust joint representation,Face recognition,Single sample per person

论文评审过程：Received 6 December 2018, Revised 27 May 2019, Accepted 28 May 2019, Available online 30 May 2019, Version of Record 16 August 2019.

论文官网地址：https://doi.org/10.1016/j.knosys.2019.05.033