Virtual sample based kernel discrimination method for face recognition

Abstract

The invention discloses a virtual sample based kernel discrimination method for face recognition, which is a virtual sample based quick kernel method. The method comprises the following steps of: constructing a virtual sample set for a training sample set at one time before constructing a kernel matrix for the training sample set; and training / testing through a kernel matrix theory based on the virtual sample set. Since the virtual sample set is an aggregate of a characteristic sample set (MES) and a public vector sample set (MCS) of the training sample set, the virtual sample set has extremely high description capacity for both a known training sample set and an unknown test sample set. Experimental verification of the method on an FERET database shows that the method is quick and effective; the computation speed of the kernel method is greatly increased by using the method; and meanwhile, compared with the conventional kernel method, the recognition rate is also increased.

Description

technical field [0001] The invention relates to a nuclear identification method, which is established based on virtual samples and used for feature extraction of face recognition, belonging to the field of face recognition in pattern recognition. Background technique [0002] (1) Research background: [0003] Face recognition includes three links: image preprocessing, feature extraction and recognition. Feature extraction is one of the most basic problems in pattern recognition research. For image recognition, extracting effective image features is the primary task of image recognition. Kernel-based feature extraction method is a very popular and effective nonlinear feature extraction method nowadays. The basic idea of ​​the kernel method is to use a nonlinear mapping to map the linearly inseparable samples in the input space R to a hidden feature space, so that the samples are linearly separable in the space F. In the kernel method, there is no need to explicitly calcul...

AI Technical Summary

Problems solved by technology

[0027] Although the kernel feature extraction method can transform the linearly inseparable problem of the original space into a linearly separable problem in the high-dimensional space, its projection vector is linearly expanded by all training samples, such as kernel principal component analysis (KPCA) and generalized The discriminant analysis method (GDA), especially in the case of multi-class, takes a lot of time to calculate the huge kernel matrix, so that the calculation of the kernel method becomes very large

In order to solve this problem, some kernel acceleration algorithms have been proposed, such as Greedy method, Nystrom method, sparse kernel feature analysis method (SKFA) and reduced set method (RSS and RSC), etc., but these acceleration algorithms are very expensive to search for projection vector expansion elements. Time

In order to reduce the expansion elements, these acceleration methods use iterative algorithms to select expansion elements one by one from the original sample set, which is a very time-consuming process, especially considering the calculation amount of each kernel mapping function, the calculation amount is even greater

Moreover, due to discarding part of the sample information, the recognition ability of these accelerated kernel methods has declined.

Images