Title:
KERNEL NON-NEGATIVE MATRIX FACTORIZATION BASED FACE RECOGNITION METHOD AND SYSTEM, AND STORAGE MEDIUM
Document Type and Number:
WIPO Patent Application WO/2019/033261
Kind Code:
A1
Abstract:
A kernel non-negative matrix factorization based face recognition method and system, and a storage medium. Said method comprises: constructing a fractional order inner product kernel function, the fractional order inner product kernel function having no limitation to the power exponent parameters; obtaining, by means of a combination of a fractional order inner product kernel function and kernel non-negative matrix factorization, a fractional order inner product kernel non-negative matrix factorization algorithm; and performing face recognition by means of the fractional order inner product kernel non-negative matrix factorization algorithm. The method overcomes the problem that the power parameter of a polynomial kernel function can only be an integer, making the selection of the power parameter more flexible, effectively overcoming the defects caused by changes in postures and illumination during face recognition, and having a very fast convergence rate and a superior recognition performance.
Inventors:
CHEN WENSHENG (CN)
LIU JINGMIN (CN)
WANG QIAN (CN)
LIU JINGMIN (CN)
WANG QIAN (CN)
Application Number:
PCT/CN2017/097485
Publication Date:
February 21, 2019
Filing Date:
August 15, 2017
Export Citation:
Assignee:
UNIV SHENZHEN (CN)
International Classes:
G06K9/62
Foreign References:
| CN105893954A | 2016-08-24 | |||
| CN106897685A | 2017-06-27 | |||
| CN101739555A | 2010-06-16 | |||
| US20120041906A1 | 2012-02-16 |
Attorney, Agent or Firm:
SZ KINDWALF INTELLECTUAL PROPERTY FIRM (CN)
Download PDF: