Fisher discrimination based low rank matrix recovery for face recognition
作者：
Highlights：
• A low rank matrix recovery algorithm with Fisher discrimination (FDLR) is proposed.
• FDLR promotes the discrimination power in the learned representation dictionary.
• FDLR is relaxed to be solved under augmented Lagrange multipliers framework.
• A sparse error feedback strategy is presented and combined with FDLR.
• FDLR shows robustness to severe occlusions of images in training and testing sets.
摘要
Highlights•A low rank matrix recovery algorithm with Fisher discrimination (FDLR) is proposed.•FDLR promotes the discrimination power in the learned representation dictionary.•FDLR is relaxed to be solved under augmented Lagrange multipliers framework.•A sparse error feedback strategy is presented and combined with FDLR.•FDLR shows robustness to severe occlusions of images in training and testing sets.
论文关键词：Low rank,Fisher discrimination,Sparse,Augmented Lagrange multiplier
论文评审过程：Received 8 May 2013, Revised 7 March 2014, Accepted 6 May 2014, Available online 15 May 2014.
论文官网地址：https://doi.org/10.1016/j.patcog.2014.05.001
