WO2018187951A1

WO2018187951A1 - Facial recognition method based on kernel principal component analysis

Info

Publication number: WO2018187951A1
Application number: PCT/CN2017/080175
Authority: WO
Inventors: 邹霞
Original assignee: 邹霞
Priority date: 2017-04-12
Filing date: 2017-04-12
Publication date: 2018-10-18

Abstract

A facial recognition method based on kernel principal component analysis. The method comprises: calculating a kernel matrix for M pieces of given training set data X[X1, X2, … , XM]; constructing a centering matrix H and solving a characteristic equation; calculating a vector; extracting a principal component to form a characteristic subspace and obtain a sample data set Y retained after principal component analysis of facial data; projecting a test data set X' to the characteristic subspace of the training set to obtain a test data set Z' after characteristic extraction; and classifying and recognizing the sample Z' by means of a nearest neighbor classifier. The facial recognition method based on kernel principal component analysis can significantly shorten the recognition time. The application of a kernel method can skillfully remedy the defect that nonlinear information in data cannot be utilized in the principal component analysis method and the linear discriminant analysis method.

Description

Face recognition method based on kernel principal component analysis

Technical field

The invention relates to a face recognition method based on kernel principal component analysis, belonging to the field of biometric identification.

Background technique

Face recognition is a computer technology that achieves the purpose of identity identification by analyzing human facial visual features. The academic community gives a specific definition of face recognition in both broad and narrow sense. Generalized face recognition includes face detection, face representation, face identification, face expression analysis, and physical classification. Narrow face recognition is defined as a technology or system that enables identity verification, identity comparison, and identity lookup through facial features.

At present, since face recognition technology can distinguish individuals by the biological characteristics of organisms (generally referred to as humans), and improve the accuracy of organism recognition, the technology has been widely concerned and respected, and the field has become Hotspots in the study of biometric features. In humans, for example, biometrics mainly come from the following aspects: face, retina, iris, palmprint, fingerprint, voice, body shape, habits, etc. Therefore, based on the above, research is focused on identifying faces, retinas, and irises. The computer recognition technology of the corresponding features such as palm print, fingerprint, voice, body shape, keyboard stroke, signature, etc., has achieved important results.

The advantage of face recognition lies in its natural and friendly characteristics. The so-called natural nature means that human beings also identify and confirm the identity of each other by observing and comparing human facial features. For example, speech recognition and body shape recognition also have natural characteristics, while humans or other creatures usually do not pass fingerprints. Features such as iris distinguish individuals, so the above feature recognition does not have natural characteristics. Sign.

The so-called friendliness means that the identification method does not increase the psychological burden of the authenticated person due to special treatment, and thus it is easier to obtain direct and true feature information. Fingerprint or iris recognition needs to use special techniques such as electronic pressure sensor or infrared to collect information. The above special collection technology is easy to be discovered, which greatly increases the possibility that the authenticated person avoids identity identification and reduces the efficiency of identity authentication.

However, face recognition can directly obtain the face information of the authenticated person through simple image or video technology. This information collection method is not easy to be perceived, which increases the authenticity and reliability of the information.

Although the face recognition technology has the above advantages, the implementation of the technology is not easy. Mainly limited by the biological characteristics of the face, as follows:

First, the structure of the same type of face has a high similarity. This feature can be used for face localization, but it greatly increases the difficulty of using individual facial features to identify individuals.

Second, due to factors such as age, mood, temperature and illumination conditions, and coverings, the shape of the face is very unstable. Even at different viewing angles, the image features of the face are significantly different, and the face recognition technology is added. The complexity of the application.

In order for face recognition technology to better serve the required fields, it is necessary to study the above two limitations to seek a breakthrough.

Summary of the invention

In view of the above deficiencies of the prior art, an object of the present invention is to provide a face recognition method based on kernel principal component analysis, comprising:

Step 1: Calculate a kernel matrix for a given M training set data X[x ₁ , x ₂ , . . . , x _M ];

Step 2: Construct a centralization matrix H to solve the characteristic equation;

Step three, calculating a vector;

Step 4: Extract the principal component, form the feature subspace, and obtain the principal component analysis of the face data. After the retained sample data set Y;

Step 5: For the test data set X', project it into the feature subspace of the training set to obtain a test data set Y' after feature extraction;

Step 6. Classify the sample Y' by the nearest neighbor classifier.

Preferably, the above step 3 calculates the vector k _x =[K(x,x ₁ ), K(x,x ₂ ),...,K(x,x _M )] ^T and

Preferably, the above step 4 is to extract the first k total contribution rates of 90% or more.

Compared with the prior art, the face recognition method based on kernel principal component analysis provided by the invention can greatly shorten the recognition time, and the core method is used to make up for the fact that the principal component analysis method and the linear discriminant analysis method cannot utilize the data in the middle. The shortcoming of linear information.

detailed description

The present invention provides a face recognition method based on kernel principal component analysis, and the present invention will be further described in detail in the following examples in order to clarify and clarify the objects, technical solutions and effects of the present invention. It is understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

As shown in FIG. 1, the face recognition method based on kernel principal component analysis provided by the present invention maps sample data from a low-dimensional space to a high-dimensional space by a kernel method, so that the PCA algorithm has the processing capability for nonlinear data.

In principal component analysis, let x _k ∈R ^N ,k=1,2,...,M,

The principal component is the diagonalized covariance matrix of the above formula

Elements that are located on the diagonal. Usually, the top k (larger eigenvalues) along the diagonal direction The principal component contains most of the useful information with informational value. The principal component analysis is to find the eigenvalues and eigenvectors of the matrix for the covariance matrix C:

Wherein the eigenvalue λ ≠ 0, the eigenvector v ∈ R ^N \{0},

Let the data x _i in the original space R ^N pass the nonlinear mapping Φ(x), and the image in the feature space F is (Φx _i ), assuming that the mapping data is zero-mean, then the data matrix Φ(x) in the feature space F The covariance matrix is:

It is known from the linear principal component analysis theory that principal component analysis of the mapped data matrix Φ(x) is equivalent to solving the eigenvectors for its covariance matrix C. Let λ be the eigenvalue of C, and matrix V is the eigenvector matrix, then

Incorporate each mapping sample with equation (4),

Expanded formula (4), available

Combined with the above analysis, the method of face recognition based on kernel principal component analysis is as follows:

(1) Calculate the kernel matrix K for a given M training set data X[x ₁ , x ₂ ,..., x _M ], where K _ij =(Φ(x _i )·Φ(x _j )) .

(2) Construct a centralization matrix H to solve the characteristic equation.

(3) Calculate the vector k _x =[K(x,x ₁ ), K(x,x ₂ ),...,K(x,x _M )] ^T and

(4) Extract the first k principal components whose total contribution rate reaches 90% or more, form the feature subspace, and obtain the sample data set Y retained after the principal component analysis of the face data.

(5) For the test data set X', it is projected into the feature subspace of the training set to obtain the feature extracted test data set Y'.

(6) The sample Y' is classified and identified by the nearest neighbor classifier.

It is to be understood that those skilled in the art can make equivalent substitutions or changes to the inventions and the inventions of the present invention, and all such changes or substitutions fall within the scope of the appended claims.

Claims

A face recognition method based on kernel principal component analysis, characterized in that the method comprises the following steps:

Step 1: Calculate a kernel matrix for a given M training set data X[x 1 , x 2 , . . . , x M ];

Step 2: Construct a centralization matrix H to solve the characteristic equation;

Step three, calculating a vector;

Step four, extracting the main component, forming a feature subspace, and obtaining a sample data set Y retained after the principal component analysis of the face data;

Step 5: For the test data set X', project it into the feature subspace of the training set to obtain a test data set Y' after feature extraction;

Step 6. Classify the sample Y' by the nearest neighbor classifier.
The face recognition method based on kernel principal component analysis according to claim 1, wherein said step three calculates a vector k x = [K(x, x 1 ), K(x, x 2 ), .. ..K(x,x M )] T and
The face recognition method based on kernel principal component analysis according to claim 1, wherein the step 4 is to extract the top k total contribution rates of 90% or more.