WO2017106996A1 - Human facial recognition method and human facial recognition device - Google Patents

Abstract

A human facial recognition method and a human facial recognition device, the method comprising: acquire a facial image from a user (101); perform a filtering process on the facial image to acquire a target facial image (102); acquire a concentric dual cross pattern (DCP) feature from the target facial image (103); use a chi-squared test to calculate a similarity score between the DCP feature corresponding to the target facial image and a DCP feature corresponding to an original facial image, and use the similarity score to perform facial recognition of the target facial image, wherein the DCP feature corresponding to an original facial image is acquired in advance (104). The method acquires the DCP feature to achieve effective facial recognition, raise the robustness of facial recognition, enhance practicality of a facial recognition solution, and improve user experience.

Description

Face recognition method and face recognition device Technical field

The embodiments of the present invention relate to the field of biometrics and computer technology, and in particular, to a method for recognizing a face and a face recognition device.

Background technique

Face recognition technology has developed rapidly in recent years. Face recognition technology is based on human facial features and recognizes input face images or video streams. First, it is judged whether or not there is a human face, and if there is a human face, the position and size of each face and the position information of each main facial organ are further given. Based on this information, the identity features contained in each face are further extracted and compared with known faces to identify the identity of each face.

In the prior art, the face image of the face is first established. That is, the camera collects the face image files of the face of the unit personnel or takes their photos to form an image file, and stores these face image files to obtain the face code; and obtains the current body image, that is, the current access captured by the camera. The face image of the person, or take a photo input, and encode the current face image file; use the current face code to compare with the file inventory, that is, the face code of the current face image and the face in the file inventory The pattern is coded for retrieval.

However, in the prior art, in terms of face code coding, although the texture structure of the face can be characterized, the influence on the noise and the light change has not been well solved, and the face pose and the expression change. At the same time, it is difficult to perform face recognition by simple calculation, and it can be seen that the robustness is not good enough.

Summary of the invention

Embodiments of the present invention provide a method for recognizing a face and a face recognition device, which are used for effectively recognizing a face by extracting a DCP feature, so that the face recognition is more robust and the scheme is increased. Practicality to enhance the user experience.

In view of this, the first aspect of the present invention provides a method for face recognition, including:

Obtaining a face image of the user;

Filtering the face image and obtaining a target face image;

Extracting a concentric double crisscross mode DCP feature from the target face image;

Calculating a similarity score between the DCP feature corresponding to the target face image and the DCP feature corresponding to the original face image by using a chi-square test, and comparing the target face map according to the similarity score For example, the recognition is performed, wherein the DCP feature corresponding to the original face image is obtained in advance.

With reference to the first aspect of the embodiments of the present invention, in a first possible implementation manner, the filtering, processing, and obtaining a target facial image includes:

The filtered gradient image of the face image is calculated as follows:

Wherein FDG(θ) represents a filtered gradient image of the face image corresponding to the direction of the θ angle,

Represents the standard direction vector in the filter, G represents the two-dimensional Gaussian filter, and ▽ represents the gradient operator symbol.

With reference to the first possible implementation manner of the first aspect of the embodiments of the present invention, in the second possible implementation manner, the θ angle is four angle values, which are 0 degrees, 45 degrees, 90 degrees, and 135 degrees, respectively.

With reference to the first aspect of the embodiments of the present invention, in a third possible implementation, the extracting the concentric double cross mode DCP feature from the target facial image includes:

Taking the center point of the target face image as a center, respectively obtaining an inner circle and an outer circle having different radii;

Obtaining 8 inner circle sampling points with equal angular intervals from the inner circle;

Obtaining 8 outer-circle DCP sampling points with equal angular intervals from the outer circle, wherein the inner-circular sampling points have a corresponding relationship with the outer-circular sampling points;

The inner circle sampling point and the outer circle sampling point are respectively encoded, and the DCP feature is obtained.

With reference to the third possible implementation manner of the first aspect of the embodiments of the present invention, in a fourth possible implementation, the inner circle sampling point and the outer circle sampling point are separately encoded, and the DCP features, including:

The DCP codes on the inner circle sampling point and the outer circle sampling point are respectively calculated as follows:

or,

Where DCP _i represents the DCP code of the ith sample point, and S(x) represents the gray scale intensity function.

And I _O respectively represent gray values of the sampling points A _i , B _i and O;

The DCP features are calculated as follows:

Where DCP represents the DCP feature and i represents the ith sample point.

The DCP encoding representing the sampling points in the horizontal and vertical directions,

The DCP code representing the sample points in the diagonal direction.

With reference to the fourth possible implementation manner of the first aspect of the embodiments of the present invention, in a fifth possible implementation manner,

The value of the gray level intensity function S(x) is calculated as follows:

Wherein, S(x) represents a grayscale intensity function, b(x) represents a constant value function, f _1,d (x) and f _0,d (x) are represented as fuzzy membership functions, and d is a boundary threshold.

A second aspect of the present invention provides a face recognition device, including:

An obtaining module, configured to acquire a face image of the user;

a filtering module, configured to perform filtering processing on the face image acquired by the acquiring module, and obtain a target face image;

An extraction module, configured to extract a concentric double cross mode DCP feature from the target face image obtained by filtering by the filtering module;

a calculation module, configured to calculate, by using a chi-square test, a similarity score between a DCP feature corresponding to the target face image extracted by the extraction module and a DCP feature corresponding to the original face image, and according to the similarity score pair The target face image is identified, wherein the DCP feature corresponding to the original face image is obtained in advance.

With reference to the second aspect of the embodiments of the present invention, in a first possible implementation, the filtering module includes:

a calculating unit, configured to calculate a filtered gradient image of the face image as follows:

Wherein FDG(θ) represents a filtered gradient image of the face image corresponding to the direction of the θ angle,

Represents the standard direction vector in the filter, G represents the two-dimensional Gaussian filter, and ▽ represents the gradient operator symbol.

With reference to the first possible implementation manner of the second aspect of the embodiment of the present invention, in the second possible implementation manner, the θ angle is four angle values, which are 0 degrees, 45 degrees, 90 degrees, and 135 degrees, respectively.

With reference to the second aspect of the embodiments of the present invention, in a third possible implementation, the extracting module includes:

a first acquiring unit, configured to obtain an inner circle and an outer circle having different radii by using a center point of the target face image as a center;

a second acquiring unit, configured to acquire, from the inner circle acquired by the first acquiring unit, eight inner circle sampling points with equal angular intervals;

a third acquiring unit, configured to acquire, from the outer circle acquired by the first acquiring unit, eight outer circle DCP sampling points with equal angular intervals, wherein the inner circle sampling point and the outer circle sampling point have Correspondence relationship

a coding unit, configured to respectively encode the inner circle sampling point acquired by the second acquiring unit and the outer circle sampling point acquired by the third acquiring unit, and obtain the DCP feature.

With reference to the third possible implementation manner of the second aspect of the embodiments of the present invention, in a fourth possible implementation, the coding unit includes:

Calculating a subunit for respectively calculating DCP codes on the inner circle sampling point and the outer circle sampling point as follows:

or,

Wherein, _i represents DCP DCP encoding the i-th sampling point, S (x) represents the gray scale intensity function,

And I _O respectively represent gray values of the sampling points A _i , B _i and O;

The DCP features are calculated as follows:

Where DCP represents the DCP feature and i represents the ith sample point.

The DCP encoding representing the sampling points in the horizontal and vertical directions,

The DCP code representing the sample points in the diagonal direction.

With reference to the fourth possible implementation manner of the second aspect of the embodiments of the present invention, in a fifth possible implementation manner,

The calculating subunit is further configured to calculate a value of the gray level intensity function S(x) as follows:

Wherein, S(x) represents a grayscale intensity function, b(x) represents a constant value function, f _1,d (x) and f _0,d (x) are represented as fuzzy membership functions, and d is a boundary threshold.

A third aspect of the present invention provides a face recognition device, including:

Processor and memory;

The memory is used to store a program;

The processor is configured to execute a program in the memory, such that the face recognition device performs the face recognition according to the first aspect of the present invention, the first to fifth possible implementation manners of the first aspect Methods.

A fourth aspect of the present invention provides a storage medium storing one or more programs, including:

The one or more programs include instructions that, when executed by the face recognition device including one or more processors, cause the face recognition device to perform the method of any one of claims 1 to Face recognition method.

It can be seen from the above technical solutions that the embodiments of the present invention have the following advantages:

In the embodiment of the present invention, a method for face recognition is provided. First, a face image of a user is acquired, and then the face image is filtered, and a target face image is obtained, and then the DCP feature is extracted from the target face image. Finally, the chi-square test is used to calculate the similarity score between the DCP feature corresponding to the target face image and the DCP feature corresponding to the original face image, and the target face image is identified according to the similarity score, wherein the original face image The corresponding DCP features are obtained in advance. The face recognition is effectively used to effectively identify the face, which makes the face recognition more robust, increases the practicability of the solution, and enhances the user experience.

DRAWINGS

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly described below. It is obvious that the drawings in the following description are only some embodiments of the present invention, Those skilled in the art can also obtain other drawings according to these drawings without paying any creative work.

1 is a schematic diagram of an embodiment of a method for recognizing a face according to an embodiment of the present invention;

2 is a schematic diagram of local sampling of DCP features in an embodiment of the present invention;

3 is a schematic diagram of two modes of adopting DCP features in an embodiment of the present invention;

4 is a schematic diagram of an extraction process of a DCP feature according to an embodiment of the present invention;

Figure 5 is a schematic diagram showing the results of FAR and FRR evaluation indicators of Experiment (1);

Figure 6 is a schematic diagram showing the results of FAR and FRR evaluation indicators of Experiment (2);

7 is a schematic diagram showing the results of FAR and FRR evaluation indexes of Experiment (3);

Figure 8 is a schematic diagram showing the results of FAR and FRR evaluation indexes of Experiment (4);

9 is a schematic diagram showing the results of FAR and FRR evaluation indexes of Experiment (5);

FIG. 10 is a schematic diagram of an embodiment of a face recognition device according to an embodiment of the present invention; FIG.

FIG. 11 is a schematic diagram of another embodiment of a face recognition device according to an embodiment of the present invention; FIG.

FIG. 12 is a schematic diagram of another embodiment of a face recognition device according to an embodiment of the present invention; FIG.

FIG. 13 is a schematic diagram of another embodiment of a face recognition device according to an embodiment of the present invention; FIG.

FIG. 14 is a schematic structural diagram of a face recognition device according to an embodiment of the present invention.

detailed description

The technical solutions in the embodiments of the present invention are clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present invention. It is obvious that the described embodiments are only a part of the embodiments of the present invention, but not all embodiments. All other embodiments obtained by a person skilled in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention.

The terms "first", "second", "third", "fourth", etc. (if present) in the specification and claims of the present invention and the above figures are used to distinguish similar objects without being used for Describe a specific order or order. It is to be understood that the data so used may be interchanged as appropriate, such that the embodiments of the invention described herein can be implemented, for example, in a sequence other than those illustrated or described herein. In addition, the terms "including" and "having" and any variants thereof are intended to cover no An exclusive inclusion, for example, a process, method, system, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include those that are not clearly listed or Other steps or units inherent to the method, product or device.

Embodiments of the present invention provide a method for recognizing a face and a face recognition device, which are used for effectively recognizing a face by extracting a DCP feature, so that the face recognition is more robust and the scheme is increased. Practicality to enhance the user experience.

The technical solutions in the embodiments of the present invention are clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present invention. It is obvious that the described embodiments are only a part of the embodiments of the present invention, but not all embodiments. All other embodiments obtained by a person skilled in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention.

The method for the face recognition in the present invention is described in detail below. Referring to FIG. 1, an embodiment of a method for recognizing a face according to an embodiment of the present invention includes:

101. Obtain a face image of the user;

In this embodiment, the camera in the face recognition device captures a face image of the user, wherein the face image should include at least an eye, a nose, and a mouth.

Obtaining a user's face image mainly includes the following two steps:

First, face image acquisition, different face images can be collected through the camera lens, such as static images, dynamic images, different positions, different expressions, etc. can be well collected. When the user is within the shooting range of the acquisition device, the acquisition device automatically searches for and captures the user's face image.

Second, face detection, face detection is mainly used in the preprocessing of face recognition in practice, that is, the position and size of the face are accurately calibrated in the image. The pattern features contained in the face image are very rich, such as histogram feature, color feature, template feature, structural feature and rectangular feature (English name: Haar). Face detection is to pick out the useful information and use these features to achieve face detection.

The mainstream face detection method is based on the above features using adaptive weak classification algorithm (English full name: Adaptive boosting, English abbreviation: Adaboost), Adaboost algorithm is a method used for classification, it combines some weak classification methods , combined with a new strong classification method. In the face detection process, the Adaboost algorithm is used to select some Haar features that best represent the face. The weak classifier is constructed as a strong classifier according to the weighted voting method, and then several strong classifiers obtained by training are connected in series to form a cascaded classifier of cascade structure, which effectively improves the detection speed of the classifier.

102. Perform filtering processing on the face image, and obtain a target face image;

In this embodiment, the acquired face image needs to be pre-processed, and the image is processed based on the face image and the process of final service and feature extraction is performed. The original image acquired by the system is often not directly used due to various conditions and random interference. Image preprocessing such as gradation correction, noise filtering and filtering must be performed in the early stage of image processing. For face images, the preprocessing process mainly includes ray compensation, gradation transformation, histogram equalization, normalization, geometric correction, filtering and sharpening of face images, and finally can be used to extract features. Target face image.

103. Extracting a concentric double crisscross mode DCP feature from the target face image;

In this embodiment, the feature of the concentric double cross mode (Dual-Cross Patterns, English abbreviation: DCP) is extracted from the target face image. Among them, the DCP feature is improved on the basis of the characteristics of the local binary pattern (English full name: Local Binary Patterns, English abbreviation: LBP). First, the field of the unit 8 is changed into the double-area 8 fields, and then horizontally. In the vertical and diagonal directions, two sets of sub-DCP features of the same dimension are extracted according to the local quaternary coding method, and finally connected, which is a DCP feature.

It can be understood that in addition to extracting DCP features, the features that the face recognition device can use include visual features, pixel statistical features, face image transform coefficient features, and face image algebra features. Face feature extraction is performed on certain features of the face. Face feature extraction, also known as face representation, is a process of character modeling a face. The methods of face feature extraction are summarized into two categories: one is based on knowledge representation and the other is based on algebraic features or statistical learning.

104. Calculate the similarity score between the DCP feature corresponding to the target face image and the DCP feature corresponding to the original face image by using the chi-square test, and identify the target face image according to the similarity score, wherein the original face image The corresponding DCP features are obtained in advance.

In this embodiment, the similarity score of the DCP feature corresponding to the DCP feature corresponding to the original face image stored in the database is calculated by using the chi-square test, and a threshold is set by setting a threshold value. When the similarity score exceeds this threshold, the result of the matching is output.

Face recognition is to compare the face features to be recognized with the obtained face feature templates, and judge the identity information of the faces according to the degree of similarity. This process is divided into two categories: one is confirmation, one-to-one process of image comparison, and the other is recognition, which is a one-to-many process of image matching and comparison. The specific matching method is not limited here.

In the embodiment of the present invention, a method for face recognition is provided. First, a face image of a user is acquired, and then the face image is filtered, and a target face image is obtained, and then the DCP feature is extracted from the target face image. Finally, the chi-square test is used to calculate the similarity score between the DCP feature corresponding to the target face image and the DCP feature corresponding to the original face image, and the target face image is identified according to the similarity score, wherein the original face image The corresponding DCP features are obtained in advance. The face recognition is effectively used to effectively identify the face, which makes the face recognition more robust, increases the practicability of the solution, and enhances the user experience.

Optionally, in the first optional embodiment of the method for recognizing a face according to the embodiment of the present invention, the face image is filtered and the target face image is obtained. Can include:

The filtered gradient image of the face image is calculated as follows:

Where FDG(θ) represents a filtered gradient image of the face image corresponding to the direction of the θ angle,

Represents the standard direction vector in the filter, G represents the two-dimensional Gaussian filter, and ▽ represents the gradient operator symbol.

In this embodiment, in order to suppress the influence of noise and illumination changes, it is necessary to use a Gaussian first derivative operator for the input face image from multiple angles (English full name: the first derivative of Gaussian operator, English abbreviation :FDG) Calculate the FDG filter gradient image, the formula is as follows:

FDG(θ) represents an FDG filtered gradient image of a face image corresponding to the direction of the θ angle,

Represents the standard direction vector in the filter, G represents the two-dimensional Gaussian filter, and ▽ represents the gradient operator symbol.

among them,

Is the standard direction vector used in the filter, which is expressed as:

G is a two-dimensional Gaussian filter whose calculation formula is as follows:

A Gaussian filter is a type of linear smoothing filter that selects weights based on the shape of a Gaussian function. Gaussian smoothing filters are very effective at suppressing noise that obeys a normal distribution. The one-dimensional zero-mean Gaussian function is:

Among them, the Gaussian distribution parameter Sigma determines the width of the Gaussian function. For image processing, a two-dimensional zero-mean discrete Gaussian function is commonly used as a smoothing filter. The Gaussian function has five important properties that make it particularly useful in early image processing. These properties indicate that Gaussian smoothing filters are very efficient low-pass filters in both the spatial and frequency domains, and Gaussian functions. There are five very important properties, namely:

(1) The two-dimensional Gaussian function has rotational symmetry, that is, the smoothness of the filter in all directions is the same. In general, the edge direction of an image is not known in advance, so it is impossible to determine that more smoothing is required in one direction than in the other direction before filtering. Rotational symmetry means that the Gaussian smoothing filter does not deflect in either direction in subsequent edge detection.

(2) The Gaussian function is a single-valued function, which indicates that the Gaussian filter replaces the pixel value of the point with the weighted mean of the pixel neighborhood, and the weight of each neighborhood pixel is monotonically increased with the distance between the point and the center point. Less. This property is important because the edge is an image local feature. If the smoothing operation still has a large effect on pixels far from the center of the operator, the smoothing operation will distort the image.

(3) The Fourier transform spectrum of the Gaussian function is single-lobed. This property is a direct inference of the fact that the Gaussian function Fourier transform is equal to the Gaussian function itself. Images are often contaminated by unwanted high-frequency signals (such as noise and fine texture), and desired image features (such as edges) contain both low-frequency components and high-frequency components. Gaussian function Fourier transform single-lobes This means that the smoothed image is not contaminated by unwanted high frequency signals while retaining most of the desired signal.

(4) The Gaussian filter width (determining the degree of smoothness) is characterized by the parameter σ, and the relationship between σ and the degree of smoothing is very simple. The larger the σ, the wider the band of the Gaussian filter and the smoothness The better. By adjusting the smoothness parameter σ, a compromise can be obtained between excessive blurring of the image features (too smooth) and excessive undesired abrupt changes (not enough smoothing) due to noise and fine texture in the smoothed image.

(5) Due to the separability of the Gaussian function, a larger-sized Gaussian filter can be effectively realized. The two-dimensional Gaussian function convolution can be performed in two steps. First, the image is convolved with the one-dimensional Gaussian function, and then the convolution result is convolved with the same one-dimensional Gaussian function perpendicular to the direction. Therefore, the calculation of the two-dimensional Gaussian filter The amount grows linearly with the width of the filter template rather than squared.

Secondly, in the embodiment of the present invention, a method for obtaining a target face image by filtering a face image is provided, and a filter gradient image is separately calculated from four directions by using a correlation formula, and the filter is correspondingly performed. The optimization can suppress the influence of noise and illumination changes, improve the quality of the face image, facilitate the extraction of image features, and enhance the practicability of the scheme.

Optionally, in the second optional embodiment of the method for recognizing a face according to the foregoing embodiment of the present invention, the θ angle is four angle values, respectively. It is 0 degrees, 45 degrees, 90 degrees and 135 degrees.

In the present embodiment, a filter in which the θ angle is equal to 0, 45, 90, and 135 degrees, respectively, will be described.

According to the first alternative embodiment corresponding to FIG. 1 above, the FDG filter gradient image has the following formula:

When calculating the FDG gradient filtered images of 0 degrees and 90 degrees, the corresponding values can be directly substituted into the formula. However, when calculating the FDG gradient filtered images of 45 degrees and 135 degrees, the following formula can be used:

F _θ =F _X cosθ+F _Y sinθ

Wherein F _X represents FDG gradient filtering in the horizontal direction, and F _Y represents FDG gradient filtering in the vertical direction.

Since the FDG filter is optimized, it is verified by experiments that the size of the filter is 5×5, the mathematical expectation in the two-dimensional Gaussian filter is μ=4, and the variance σ=1, the filter effect is better. Filters in 4 directions can be obtained, which are expressed as:

horizontal direction:

Vertical direction:

Diagonal direction:

Diagonal direction:

It should be noted that the filter in the four directions is only one schematic. In practical applications, the filter may be configured by other parameters, which is not limited herein.

In addition, in the embodiment of the present invention, the filtering process for the face image may specifically involve filtering in four directions, namely 0 degrees, 45 degrees, 90 degrees, and 135 degrees, and is verified by experiments, from these four angles. The input face image is filtered to achieve a cost-effective effect. Although the filtering process in multiple directions can achieve better when the angular interval is smaller. The ability to suppress noise and suppress illumination changes, but it increases the computational burden. Not conducive to practical applications. However, filtering using a direction with a larger angular interval may weaken the suppression of noise and illumination variations, which is not conducive to image processing. Therefore, the filtering process in the four directions of the face image provided by the solution of the present invention has operability and practicability.

Optionally, on the basis of the foregoing embodiment corresponding to FIG. 1 , in a third optional embodiment of the method for recognizing a face according to the embodiment of the present invention, extracting a concentric double crisscross mode DCP feature from a target face image Can include:

Taking the center point of the target face image as the center, respectively obtaining the inner circle and the outer circle having different radii;

Obtaining 8 inner circle sampling points with equal angular intervals from the inner circle;

Obtaining 8 outer circle DCP sampling points with equal angular intervals from the outer circle, wherein the inner circle sampling points have corresponding correspondences with the outer circle sampling points;

The inner circle sampling point and the outer circle sampling point are respectively coded, and the DCP feature is obtained.

In this embodiment, the DCP features are extracted by using the sampling method of the eight circles of the double circle. First, the diagonals of the target face images are respectively connected, and the intersection of the diagonal lines, that is, the center point is obtained. Two circles with different radii are drawn with the center point as the center, the inner circle is the smaller radius, and the outer circle is the larger radius. For the inner circle and the outer circle at the same time in every 45 degree angle, the radius of the inner circle and the outer circle are both sampled with two points. Specifically, please refer to FIG. 2. FIG. 2 is a schematic diagram of local sampling of DCP features according to an embodiment of the present invention. As shown in the figure, the center point is O, the inner circle sampling point is A _i , and the outer circle sampling point is B _i , i= 0,1,...,7 are symmetrically arranged in 8 directions, that is, the inner circle sampling point has a corresponding relationship with the outer circle sampling point. For example, A ₀ and B ₀ have a corresponding relationship, and both correspond to a 0 degree angle, A ₁ Corresponding to B ₁ , each corresponds to a 45 degree angle, and so on, until A ₇ and B ₇ have a corresponding relationship, which corresponds to a 335 degree angle.

In this scheme, the sampling method is to extract DCP features in 8 fields of double circle. The advantage of extracting features from the unit is to increase the context information of the local domain and characterize the local intensity contrast. However, in the improvement, the sampling method of eight fields of three circles was tried, but the test effect was general. The reason was that the coding method of the center point and the three points in each direction did not have a suitable calculation form, so the sampling method was affected. Evaluation.

In addition, the eight fields of the ellipse were tried, and the test results were not significantly improved. As for the 16 areas that are inconsistent with the eight directions of this program, they are not introduced.

It can be understood that when the radius of the inner circle is 4 and the radius of the outer circle is 6 pixels, the test result is better.

Finally, the inner circle sampling point and the outer circle sampling point are respectively encoded, and the DCP feature is obtained.

Secondly, in the embodiment of the present invention, the DCP feature is extracted by using a method of local sampling from 8 fields of double circle. In contrast, in the prior art, LBP features are extracted from 16 fields of a single circle or 16 fields of a single circle. The extraction of DCP features can better cater to the trend of facial texture. For the face, there are two main pieces of key information, one is the structure of the facial organs, and the other is the shape of the facial organs. In general, the shape of the facial organs is regular, and their ends converge substantially in a diagonal direction so that features can be extracted from the diagonal direction. In addition, the wrinkles on the forehead are flat, but they are convex or inclined on the cheeks. Therefore, the method of local sampling of 8 fields in the double circle can better describe the main texture information of the face and improve the feasibility of the scheme.

Optionally, in the fourth optional embodiment of the method for recognizing a face provided by the embodiment of the present invention, the inner circle sampling point and the outer circle are respectively selected on the basis of the third optional embodiment corresponding to FIG. The sampling points are encoded and the DCP features are obtained, which may include:

Calculate the DCP code on the inner and outer sampling points as follows:

or,

Where DCP _i represents the DCP code of the ith sample point, and S(x) represents the gray scale intensity function.

And I _O respectively represent gray values of the sampling points A _i , B _i and O;

The DCP features are calculated as follows:

Where DCP represents the DCP feature and i represents the ith sample point.

DCP encoding indicating the sampling points in the horizontal and vertical directions,

Indicates the DCP encoding of the sample points in the diagonal direction.

In this embodiment, the inner circle sampling point and the outer circle sampling point are respectively coded, and the DCP feature is obtained, which may be specifically divided into the following two steps. The first step is to divide the inner circle and the outer circle according to the eight directions. The upper sampling points are independently coded; the second step is to connect the 8 direction codes on the inner circle and the 8 direction codes on the outer circle to obtain DCP coding.

In each direction, the DCP encoding is calculated as:

or,

The similarity scores calculated by the two formulas are consistent, where DCP _i represents the DCP code of the i-th sample point, and S(x) represents the gray-scale intensity function.

And I _O respectively represent gray values of the sampling points A _i , B _i and O;

The calculation formula of the gray intensity function S(x) is as follows:

According to the calculation formula of DCP coding, the value of DCP can take 0, 1, 2 and 3, a total of four values, which can be encoded in quaternary. If it is coded in 8 directions, there are 4 ⁸ = 65536 dimensions, which is more difficult in practical applications. If the direction is divided into two groups, each group has four directions, which are horizontal and vertical directions (0, π/2, π, 3π/2), and diagonal directions (π/4, 3π/4, 5π/4, 7π/4), a total of 4 ⁴ × 2 = 512 dimensions, so that the dimension can be greatly reduced.

For the above-mentioned block coding strategy, the maximum joint entropy theory is utilized. In order to reduce the loss of information, the interval between the four directions of each group needs to be the largest, that is, perpendicular to each other, so that each has a certain independence. In addition, for an image, the more sparsely dispersed the pixels, the stronger the independence between them, and the joint entropy can be maximized. Please refer to FIG. 3. FIG. 3 is a schematic diagram of two modes of adopting DCP features according to an embodiment of the present invention. As shown in the figure, according to the above analysis, DCP coding can be divided into the following two groups:

Horizontal and vertical directions: DCP _-1 = {DCP ₀ , DCP ₂ , DCP ₄ , DCP ₆ }

Diagonal direction: DCP _-2 = {DCP ₁ , DCP ₃ , DCP ₅ , DCP ₇ }

Then the respective calculation formula is as follows:

The final composition of the DCP is:

Please refer to FIG. 4. FIG. 4 is a schematic diagram of a process for extracting a DCP feature according to an embodiment of the present invention. First, a face image is acquired, and a target face image is obtained by filtering. The target face image is divided into two circles and 8 fields, and DCP coding of two sets of four directions is obtained, and finally the target face image DCP feature is formed. Face recognition is performed by comparison with DCP features in the database.

Again, in the embodiment of the present invention, the inner circle sampling point and the outer circle sampling point are respectively encoded, and a DCP feature is obtained. The sampling points on the inner circle and the outer circle are independently coded in 8 directions, 8 direction codes on the inner circle are connected, and 8 direction codes on the outer circle are obtained to obtain DCP coding. By obtaining the DCP feature in the above manner, the calculation dimension can be greatly reduced, and the calculation efficiency can be provided. Although this split coding strategy loses some of the texture information, it makes DCP coding more compact and more robust on the face of the person.

Optionally, in the fifth optional embodiment of the method for recognizing a face provided by the embodiment of the present invention, in the fourth optional embodiment corresponding to FIG.

Calculate the value of the grayscale intensity function S(x) as follows:

Where S(x) represents the gray level intensity function, b(x) represents the constant value function, f _1,d (x) and f _0,d (x) are represented as fuzzy membership functions, and d is the boundary threshold.

In this embodiment, in the field of double circle 8, when the gray value of the prime point and the center point are close, the intensity contrast is easily affected by noise, so the "soft boundary" coding method is used to improve the gray intensity function, specifically Calculate the value of the grayscale intensity function S(x) using the following formula:

Among them, the calculation method of the constant value function b(x) is as follows:

The fuzzy membership functions f _1,d (x) and f _0,d (x) are calculated as follows:

f _0,d (x)=1-f _1,d (x)

Where d is the boundary threshold, which affects the membership degree of the module. In the experiment, d is 0.0005.

Further, in the embodiment of the present invention, when the gray point value of the pixel point and the center point are close, the intensity contrast is susceptible to noise, so the "soft boundary" coding method is used to improve the gray intensity function, so that the pixel point and the center When the gray value of the point is close, it is not susceptible to noise, which makes the extraction process of the DCP feature more robust, and improves the feasibility and practicability of the solution of the present invention.

For ease of understanding, a method for face recognition in the present invention will be described in detail below by an experimental procedure, specifically as follows:

This experiment is mainly used to verify the optimal data involved in the above various embodiments. In the experiment, the corresponding image block mainly adopts 3×3 blocks, and other block combinations are not used because of the general test effect. When each feature is matched, the weights are assigned accordingly. Before using the FDG operator, if the feature detection (English name: Difference of Gaussia, English abbreviation: DoG) is used to process the face image, the recognition effect is obtained. Will be worse. On the face registration database containing 38225 face images, the error rejection rate after testing (English full name: False Rejection Rate, English abbreviation: FRR) and the error acceptance rate (English full name: False Acceptance Rate, English abbreviation: FAR) The evaluation results are shown in Fig. 5. Please refer to Fig. 5. Fig. 5 is a schematic diagram showing the results of the FAR and FRR evaluation indexes of the experiment (1). Among them, FRR and FAR are the two main parameters used to evaluate the performance of fingerprint or face recognition algorithms.

However, after switching to the filter in the scheme of the present invention, please refer to FIG. 6. FIG. 6 is a schematic diagram of the FAR and FRR evaluation index results of the experiment (2), and FIG. 6 shows the FAR and FRR evaluations obtained on the face registration database. result.

In the feature matching, after optimizing the weight of each block, the obtained FAR and FRR evaluation results are shown in Fig. 7, and Fig. 7 is a schematic diagram of the FAR and FRR evaluation index results of experiment (3).

After the adjustment of the 3×3 block size and the optimization of each block weight, the FAR and FRR evaluation index results of the experiment (4) corresponding to FIG. 8 are obtained.

After using the soft boundary fuzzy membership function, please refer to FIG. 9. FIG. 9 is a schematic diagram of the FAR and FRR evaluation index results of experiment (5). It can be understood that if the face image is not aligned, it is necessary to first perform similar transformation and affine transformation on the face image, and then perform image cropping.

The following is a detailed description of the face recognition device in the present invention. Referring to FIG. 10, the face recognition device 200 in the embodiment of the present invention includes:

The obtaining module 201 is configured to acquire a face image of the user;

The filtering module 202 is configured to perform filtering processing on the face image acquired by the acquiring module 201, and obtain a target face image;

The extraction module 203 is configured to extract a concentric double cross mode DCP feature from the target face image obtained by filtering by the filtering module 202;

a calculation module 204, configured to calculate, by using a chi-square test, a similarity score between a DCP feature corresponding to the target face image extracted by the extraction module 203 and a DCP feature corresponding to the original face image, and according to the similarity The score identifies the target face image, wherein the DCP feature corresponding to the original face image is obtained in advance.

In this embodiment, the acquiring module 201 acquires a face image of the user; the filtering module 202 performs filtering processing on the face image acquired by the obtaining module 201, and obtains a target face image; and the target obtained by the filtering module 202 is filtered by the filtering module 202. The concentric double cross mode DCP feature is extracted from the face image; the calculation module 204 uses the chi-square test to calculate the similarity score between the DCP feature corresponding to the target face image extracted by the extraction module 203 and the DCP feature corresponding to the original face image. And the target face image is identified according to the similarity score, wherein the DCP feature corresponding to the original face image is obtained in advance.

In the embodiment of the present invention, a method for face recognition is provided. First, a face image of a user is acquired, and then the face image is filtered, and a target face image is obtained, and then the DCP feature is extracted from the target face image. Finally, the chi-square test is used to calculate the similarity score between the DCP feature corresponding to the target face image and the DCP feature corresponding to the original face image, and the target is scored according to the similarity score. The face image is identified, wherein the DCP feature corresponding to the original face image is obtained in advance. The face recognition is effectively used to effectively identify the face, which makes the face recognition more robust, increases the practicability of the solution, and enhances the user experience.

Referring to FIG. 11, another embodiment of the face recognition device of the present invention includes:

The obtaining module 201 is configured to acquire a face image of the user;

The filtering module 202 is configured to perform filtering processing on the face image acquired by the acquiring module 201, and obtain a target face image;

The extraction module 203 is configured to extract a concentric double cross mode DCP feature from the target face image obtained by filtering by the filtering module 202;

a calculation module 204, configured to calculate, by using a chi-square test, a similarity score between a DCP feature corresponding to the target face image extracted by the extraction module 203 and a DCP feature corresponding to the original face image, and according to the similarity a score is used to identify the target face image, wherein the DCP feature corresponding to the original face image is obtained in advance;

The filtering module 202 includes:

The calculating unit 2021 is configured to calculate a filtered gradient image of the face image as follows:

Wherein FDG(θ) represents a filtered gradient image of the face image corresponding to the direction of the θ angle,

Represents the standard direction vector in the filter, G represents the two-dimensional Gaussian filter, and ▽ represents the gradient operator symbol.

Secondly, in the embodiment of the present invention, a method for obtaining a target face image by filtering a face image is provided, and a filter gradient image is separately calculated from four directions by using a correlation formula, and the filter is correspondingly performed. The optimization can suppress the influence of noise and illumination changes, improve the quality of the face image, facilitate the extraction of image features, and enhance the practicability of the scheme.

Optionally, in the first optional embodiment of the face recognition device provided by the embodiment of the present invention, the θ angle is four angle values, respectively, on the basis of the foregoing embodiment corresponding to FIG. 45 degrees, 90 degrees and 135 degrees.

In addition, in the embodiment of the present invention, the filtering process for the face image may specifically involve filtering in four directions, namely 0 degrees, 45 degrees, 90 degrees, and 135 degrees, and is verified by experiments, from the fourth The angle of the input face image is filtered, which can achieve the effect of high cost performance. Although the filtering process in multiple directions can achieve better functions of suppressing noise and suppressing illumination changes in the case where the angular interval is smaller, the calculation load is increased. Not conducive to practical applications. However, filtering using a direction with a larger angular interval may weaken the suppression of noise and illumination variations, which is not conducive to image processing. Therefore, the filtering process in the four directions of the face image provided by the solution of the present invention has operability and practicability.

Referring to FIG. 12, another embodiment of the face recognition device of the present invention includes:

The obtaining module 201 is configured to acquire a face image of the user;

The filtering module 202 is configured to perform filtering processing on the face image acquired by the acquiring module 201, and obtain a target face image;

The extraction module 203 is configured to extract a concentric double cross mode DCP feature from the target face image obtained by filtering by the filtering module 202;

a calculation module 204, configured to calculate, by using a chi-square test, a similarity score between a DCP feature corresponding to the target face image extracted by the extraction module 203 and a DCP feature corresponding to the original face image, and according to the similarity a score is used to identify the target face image, wherein the DCP feature corresponding to the original face image is obtained in advance;

The extraction module 203 includes:

The first obtaining unit 2031 is configured to obtain an inner circle and an outer circle having different radii, respectively, with the center point of the target face image as a center;

a second acquiring unit 2032, configured to acquire, from the inner circle acquired by the first acquiring unit 2031, eight inner circle sampling points with equal angular intervals;

a third obtaining unit 2033, configured to acquire, from the outer circle acquired by the first acquiring unit 2031, eight outer circle DCP sampling points with equal angular intervals, wherein the inner circle sampling point and the outer circle sample Points have a corresponding relationship;

The encoding unit 2034 is configured to respectively encode the inner circle sampling point acquired by the second acquiring unit 2032 and the outer circle sampling point acquired by the third acquiring unit 2033, and obtain the DCP feature.

Secondly, in the embodiment of the present invention, the DCP feature is extracted by using a method of local sampling from 8 fields of double circle. In contrast, in the prior art, LBP is extracted from 16 fields of a single circle or 16 fields of a single circle. Features, extracting DCP features can better cater to the direction of facial texture. For the face, there are two main pieces of key information, one is the structure of the facial organs, and the other is the shape of the facial organs. In general, the shape of the facial organs is regular, and their ends converge substantially in a diagonal direction so that features can be extracted from the diagonal direction. In addition, the wrinkles on the forehead are flat, but they are convex or inclined on the cheeks. Therefore, the method of local sampling of 8 fields in the double circle can better describe the main texture information of the face and improve the feasibility of the scheme.

Referring to FIG. 13, another embodiment of the face recognition device of the present invention includes:

The obtaining module 201 is configured to acquire a face image of the user;

The filtering module 202 is configured to perform filtering processing on the face image acquired by the acquiring module 201, and obtain a target face image;

The extraction module 203 is configured to extract a concentric double cross mode DCP feature from the target face image obtained by filtering by the filtering module 202;

a calculation module 204, configured to calculate, by using a chi-square test, a similarity score between a DCP feature corresponding to the target face image extracted by the extraction module 203 and a DCP feature corresponding to the original face image, and according to the similarity a score is used to identify the target face image, wherein the DCP feature corresponding to the original face image is obtained in advance;

The extraction module 203 includes:

The first obtaining unit 2031 is configured to obtain an inner circle and an outer circle having different radii, respectively, with the center point of the target face image as a center;

a second acquiring unit 2032, configured to acquire, from the inner circle acquired by the first acquiring unit 2031, eight inner circle sampling points with equal angular intervals;

a third obtaining unit 2033, configured to acquire, from the outer circle acquired by the first acquiring unit 2031, eight outer circle DCP sampling points with equal angular intervals, wherein the inner circle sampling point and the outer circle sample Points have a corresponding relationship;

The encoding unit 2034 is configured to respectively encode the inner circle sampling point acquired by the second acquiring unit 2032 and the outer circle sampling point acquired by the third acquiring unit 2033, and obtain the DCP feature;

The coding unit 2034 includes:

a calculating subunit 20341, configured to separately calculate the inner circle sampling point and the outer side as follows DCP encoding at the round sampling point:

or,

Where DCP _i represents the DCP code of the ith sample point, and S(x) represents the gray scale intensity function.

And I _O respectively represent gray values of the sampling points A _i , B _i and O;

The DCP features are calculated as follows:

Where DCP represents the DCP feature and i represents the ith sample point.

The DCP encoding representing the sampling points in the horizontal and vertical directions,

The DCP code representing the sample points in the diagonal direction.

Again, in the embodiment of the present invention, the inner circle sampling point and the outer circle sampling point are respectively encoded, and a DCP feature is obtained. The sampling points on the inner circle and the outer circle are independently coded in 8 directions, 8 direction codes on the inner circle are connected, and 8 direction codes on the outer circle are obtained to obtain DCP coding. By obtaining the DCP feature in the above manner, the calculation dimension can be greatly reduced, and the calculation efficiency can be provided. Although this split coding strategy loses some of the texture information, it makes DCP coding more compact and more robust on the face of the person.

Optionally, in the second optional embodiment of the face recognition device provided by the embodiment of the present invention, on the basis of the foregoing embodiment corresponding to FIG.

The obtaining module 201 is configured to acquire a face image of the user;

The filtering module 202 is configured to perform filtering processing on the face image acquired by the acquiring module 201, and obtain a target face image;

The extraction module 203 is configured to extract a concentric double cross mode DCP feature from the target face image obtained by filtering by the filtering module 202;

The calculating module 204 is configured to calculate a similarity between the DCP feature corresponding to the target face image extracted by the extraction module 203 and the DCP feature corresponding to the original face image by using a chi-square test Number, and identifying the target face image according to the similarity score, wherein the DCP feature corresponding to the original face image is obtained in advance;

The extraction module 203 includes:

The first obtaining unit 2031 is configured to obtain an inner circle and an outer circle having different radii, respectively, with the center point of the target face image as a center;

a second acquiring unit 2032, configured to acquire, from the inner circle acquired by the first acquiring unit 2031, eight inner circle sampling points with equal angular intervals;

a third obtaining unit 2033, configured to acquire, from the outer circle acquired by the first acquiring unit 2031, eight outer circle DCP sampling points with equal angular intervals, wherein the inner circle sampling point and the outer circle sample Points have a corresponding relationship;

The encoding unit 2034 is configured to respectively encode the inner circle sampling point acquired by the second acquiring unit 2032 and the outer circle sampling point acquired by the third acquiring unit 2033, and obtain the DCP feature;

The coding unit 2034 includes:

The calculating subunit 20341 is configured to separately calculate DCP codes on the inner circle sampling point and the outer circle sampling point as follows:

or,

Where DCP _i represents the DCP code of the ith sample point, and S(x) represents the gray scale intensity function.

And I _O respectively represent gray values of the sampling points A _i , B _i and O;

The DCP features are calculated as follows:

Where DCP represents the DCP feature and i represents the ith sample point.

The DCP encoding representing the sampling points in the horizontal and vertical directions,

The DCP code representing the sample points in the diagonal direction.

The calculating subunit 20341 is further configured to calculate the gray level intensity function S(x) as follows Value:

Wherein, S(x) represents a grayscale intensity function, b(x) represents a constant value function, f _1,d (x) and f _0,d (x) are represented as fuzzy membership functions, and d is a boundary threshold.

Further, in the embodiment of the present invention, when the gray point value of the pixel point and the center point are close, the intensity contrast is susceptible to noise, so the "soft boundary" coding method is used to improve the gray intensity function, so that the pixel point and the center When the gray value of the point is close, it is not susceptible to noise, which makes the extraction process of the DCP feature more robust, and improves the feasibility and practicability of the solution of the present invention.

FIG. 14 is a schematic structural diagram of a face recognition device 30 according to an embodiment of the present invention. The face recognition device 30 can include an input device 310, an output device 320, a processor 330, and a memory 340. The output device in the embodiment of the present invention may be a display device.

Memory 340 can include read only memory and random access memory and provides instructions and data to processor 330. A portion of the memory 340 may also include a non-volatile random access memory (English name: Non-Volatile Random Access Memory, English abbreviation: NVRAM).

The memory 340 stores the following elements, executable modules or data structures, or a subset thereof, or an extended set thereof:

Operation instructions: include various operation instructions for implementing various operations.

Operating system: Includes a variety of system programs for implementing various basic services and handling hardware-based tasks.

In the embodiment of the present invention, the processor 330 is configured to:

Obtaining a face image of the user;

Filtering the face image and obtaining a target face image;

Extracting a concentric double crisscross mode DCP feature from the target face image;

Calculating a similarity score between the DCP feature corresponding to the target face image and the DCP feature corresponding to the original face image by using a chi-square test, and identifying the target face image according to the similarity score, wherein The DCP feature corresponding to the original face image is obtained in advance.

The processor 330 controls the operation of the face recognition device 30. The processor 330 may also be referred to as a central processing unit (English full name: Central Processing Unit: CPU). Memory 340 can include read only memory and random access memory and provides instructions and data to processor 330. A portion of the memory 340 may also include an NVRAM. In a specific application, the components of the face recognition device 30 are coupled together by a bus system 350. The bus system 350 may include a power bus, a control bus, a status signal bus, and the like in addition to the data bus. However, for clarity of description, various buses are labeled as bus system 350 in the figure.

The method disclosed in the foregoing embodiments of the present invention may be applied to the processor 330 or implemented by the processor 330. Processor 330 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the foregoing method may be completed by an integrated logic circuit of hardware in the processor 330 or an instruction in a form of software. The processor 330 may be a general-purpose processor, a digital signal processor (English name: digital signal processing, English abbreviation: DSP), an application-specific integrated circuit (English name: Application Specific Integrated Circuit, English abbreviation: ASIC), ready-made programmable Gate array (English name: Field-Programmable Gate Array, English abbreviation: FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. The methods, steps, and logical block diagrams disclosed in the embodiments of the present invention may be implemented or carried out. The general purpose processor may be a microprocessor or the processor or any conventional processor or the like. The steps of the method disclosed in the embodiments of the present invention may be directly implemented by the hardware decoding processor, or may be performed by a combination of hardware and software modules in the decoding processor. The software module can be located in a conventional storage medium such as random access memory, flash memory, read only memory, programmable read only memory or electrically erasable programmable memory, registers, and the like. The storage medium is located in memory 340, and processor 330 reads the information in memory 340 and, in conjunction with its hardware, performs the steps of the above method.

Optionally, the processor 330 is further configured to:

The filtered gradient image of the face image is calculated as follows:

Wherein FDG(θ) represents a filtered gradient image of the face image corresponding to the direction of the θ angle,

Indicates the standard direction vector in the filter, G denotes a two-dimensional Gaussian filter, and ▽ denotes a gradient operator symbol.

Optionally, the processor 330 is further configured to:

Taking the center point of the target face image as a center, respectively obtaining an inner circle and an outer circle having different radii;

Obtaining 8 inner circle sampling points with equal angular intervals from the inner circle;

Obtaining 8 outer-circle DCP sampling points with equal angular intervals from the outer circle, wherein the inner-circular sampling points have a corresponding relationship with the outer-circular sampling points;

The inner circle sampling point and the outer circle sampling point are respectively encoded, and the DCP feature is obtained.

Optionally, the processor 330 is further configured to:

The DCP codes on the inner circle sampling point and the outer circle sampling point are respectively calculated as follows:

or,

Where DCP _i represents the DCP code of the ith sample point, and S(x) represents the gray scale intensity function.

And I _O respectively represent gray values of the sampling points A _i , B _i and O;

The DCP features are calculated as follows:

Where DCP represents the DCP feature and i represents the ith sample point.

The DCP encoding representing the sampling points in the horizontal and vertical directions,

The DCP code representing the sample points in the diagonal direction.

Optionally, the processor 330 is further configured to:

The value of the gray level intensity function S(x) is calculated as follows:

Wherein, S(x) represents a grayscale intensity function, b(x) represents a constant value function, f _1,d (x) and f _0,d (x) are represented as fuzzy membership functions, and d is a boundary threshold.

The related description of FIG. 14 can be understood by referring to the related description and effect of the method part of FIG. 1 , and no further description is made herein.

A person skilled in the art can clearly understand that for the convenience and brevity of the description, the specific working process of the system, the device and the unit described above can refer to the corresponding process in the foregoing method embodiment, and details are not described herein again.

In the several embodiments provided by the present application, it should be understood that the disclosed system, apparatus, and method may be implemented in other manners. For example, the device embodiments described above are merely illustrative. For example, the division of the unit is only a logical function division. In actual implementation, there may be another division manner, for example, multiple units or components may be combined or Can be integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, device or unit, and may be in an electrical, mechanical or other form.

The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

The integrated unit, if implemented in the form of a software functional unit and sold or used as a standalone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention, which is essential or contributes to the prior art, or all or part of the technical solution, may be embodied in the form of a software product stored in a storage medium. A number of instructions are included to cause a computer device (which may be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: a U disk, a mobile hard disk, a read only memory (English full name: Read-Only Memory, English abbreviation: ROM), a random access memory (English full name: Random Access Memory, English abbreviation: RAM), magnetic A variety of media that can store program code, such as a disc or a disc.

The method for face recognition provided by the present invention is described in detail above. The principles and embodiments of the present invention are described in the following. The description of the above embodiments is only for helping to understand the method of the present invention. And the core idea; at the same time, for those skilled in the art, according to the idea of the embodiment of the present invention, there are some changes in the specific implementation manner and the application scope. In summary, the content of the present specification should not be construed as Limitations of the invention.

Claims (14)

1. A method for face recognition, comprising:

   Obtaining a face image of the user;

   Filtering the face image and obtaining a target face image;

   Extracting a concentric double crisscross mode DCP feature from the target face image;

   Calculating a similarity score between the DCP feature corresponding to the target face image and the DCP feature corresponding to the original face image by using a chi-square test, and identifying the target face image according to the similarity score, wherein The DCP feature corresponding to the original face image is obtained in advance.
2. The method according to claim 1, wherein the filtering the face image and obtaining the target face image comprises:

   The filtered gradient image of the face image is calculated as follows:

   

   Wherein FDG(θ) represents a filtered gradient image of the face image corresponding to the direction of the θ angle,

   

   Represents the standard direction vector in the filter, and G represents the two-dimensional Gaussian filter.

   

   Represents the gradient operator symbol.
3. The method of claim 2 wherein said θ angle is four angular values of 0 degrees, 45 degrees, 90 degrees, and 135 degrees, respectively.
4. The method according to claim 1, wherein the extracting the concentric double cross mode DCP feature from the target face image comprises:

   Taking the center point of the target face image as a center, respectively obtaining an inner circle and an outer circle having different radii;

   Obtaining 8 inner circle sampling points with equal angular intervals from the inner circle;

   Obtaining 8 outer-circle DCP sampling points with equal angular intervals from the outer circle, wherein the inner-circular sampling points have a corresponding relationship with the outer-circular sampling points;

   The inner circle sampling point and the outer circle sampling point are respectively encoded, and the DCP feature is obtained.
5. The method according to claim 4, wherein the encoding the inner circle sampling point and the outer circle sampling point respectively, and obtaining the DCP feature, comprises:

   Calculating the DCP series on the inner circle sampling point and the outer circle sampling point respectively as follows code:

   

   or,

   

   Where DCP _i represents the DCP code of the ith sample point, and S(x) represents the gray scale intensity function.

   

   

   And I _O respectively represent gray values of the sampling points A _i , B _i and O;

   The DCP features are calculated as follows:

   

   Where DCP represents the DCP feature and i represents the ith sample point.

   

   The DCP encoding representing the sampling points in the horizontal and vertical directions,

   

   The DCP code representing the sample points in the diagonal direction.
6. The method of claim 5 wherein:

   The value of the gray level intensity function S(x) is calculated as follows:

   

   Wherein, S(x) represents a grayscale intensity function, b(x) represents a constant value function, f _1,d (x) and f _0,d (x) are represented as fuzzy membership functions, and d is a boundary threshold.
7. A face recognition device, comprising:

   An obtaining module, configured to acquire a face image of the user;

   a filtering module, configured to perform filtering processing on the face image acquired by the acquiring module, and obtain a target face image;

   An extraction module, configured to extract a concentric double cross mode DCP feature from the target face image obtained by filtering by the filtering module;

   a calculation module, configured to calculate, by using a chi-square test, a similarity score between a DCP feature corresponding to the target face image extracted by the extraction module and a DCP feature corresponding to the original face image, and according to the similarity score pair Identifying the target face image, wherein the original face image corresponds to The DCP characteristics are obtained in advance.
8. The face recognition device according to claim 7, wherein the filtering module comprises:

   a calculating unit, configured to calculate a filtered gradient image of the face image as follows:

   

   Wherein FDG(θ) represents a filtered gradient image of the face image corresponding to the direction of the θ angle,

   

   Represents the standard direction vector in the filter, and G represents the two-dimensional Gaussian filter.

   

   Represents the gradient operator symbol.
9. The method of claim 8 wherein said θ angle is four angle values of 0 degrees, 45 degrees, 90 degrees, and 135 degrees, respectively.
10. The face recognition device according to claim 7, wherein the extraction module comprises:

    a first acquiring unit, configured to obtain an inner circle and an outer circle having different radii by using a center point of the target face image as a center;

    a second acquiring unit, configured to acquire, from the inner circle acquired by the first acquiring unit, eight inner circle sampling points with equal angular intervals;

    a third acquiring unit, configured to acquire, from the outer circle acquired by the first acquiring unit, eight outer circle DCP sampling points with equal angular intervals, wherein the inner circle sampling point and the outer circle sampling point have Correspondence relationship

    a coding unit, configured to respectively encode the inner circle sampling point acquired by the second acquiring unit and the outer circle sampling point acquired by the third acquiring unit, and obtain the DCP feature.
11. The face recognition device according to claim 10, wherein the coding unit comprises:

    Calculating a subunit for respectively calculating DCP codes on the inner circle sampling point and the outer circle sampling point as follows:

    

    or,

    

    Where DCP _i represents the DCP code of the ith sample point, and S(x) represents the gray scale intensity function.

    

    

    And I _O respectively represent gray values of the sampling points A _i , B _i and O;

    The DCP features are calculated as follows:

    

    Where DCP represents the DCP feature and i represents the ith sample point.

    

    The DCP encoding representing the sampling points in the horizontal and vertical directions,

    

    The DCP code representing the sample points in the diagonal direction.
12. A face recognition device according to claim 11, wherein

    The calculating subunit is further configured to calculate a value of the gray level intensity function S(x) as follows:

    

    Wherein, S(x) represents a grayscale intensity function, b(x) represents a constant value function, f _1,d (x) and f _0,d (x) are represented as fuzzy membership functions, and d is a boundary threshold.
13. A face recognition device, comprising:

    Processor and memory;

    The memory is used to store a program;

    The processor is configured to execute a program in the memory such that the face recognition device performs the method of face recognition according to any one of claims 1 to 6.
14. A storage medium storing one or more programs, the one or more programs including instructions that cause the face recognition when executed by the face recognition device including one or more processors The apparatus performs the method of face recognition according to any one of claims 1 to 6.

Images