Optimization-based methodology for training set selection to synthesize composite correlation filters for face recognition
作者：
Highlights：
• An optimization-based methodology is proposed to automatically select training sets.
• Three objective functions are proposed to be used as suitable optimization criteria.
• The exploration ability of the optimization algorithms is efficiently exploited.
• The results on face recognition show the high quality of the training sets selected.
• The results on illumination and expression confirm the performance in practice.
摘要
Highlights•An optimization-based methodology is proposed to automatically select training sets.•Three objective functions are proposed to be used as suitable optimization criteria.•The exploration ability of the optimization algorithms is efficiently exploited.•The results on face recognition show the high quality of the training sets selected.•The results on illumination and expression confirm the performance in practice.
论文关键词：Training set selection,Face recognition,Composite correlation filter,Optimization algorithm,Pattern recognition
论文评审过程：Received 21 June 2015, Revised 13 December 2015, Accepted 10 February 2016, Available online 19 February 2016, Version of Record 6 April 2016.
论文官网地址：https://doi.org/10.1016/j.image.2016.02.002
