Global plus local: A complete framework for feature extraction and recognition

作者：

Highlights：

• There are three kinds of local information that exist for linear discriminant analysis.

• For incorporating the local similarity information, we propose the EWLDA algorithm.

• For incorporating the three kinds of local information, we propose the CGLDA algorithm.

摘要

•There are three kinds of local information that exist for linear discriminant analysis.•For incorporating the local similarity information, we propose the EWLDA algorithm.•For incorporating the three kinds of local information, we propose the CGLDA algorithm.

论文关键词：Feature extraction,LDA,Global information,Local information,Pattern classification

论文评审过程：Received 4 February 2013, Revised 16 May 2013, Accepted 2 October 2013, Available online 17 October 2013.

论文官网地址：https://doi.org/10.1016/j.patcog.2013.10.005