Face recognition method, device and equipment applied to enterprise management system

Abstract

The invention relates to the technical field of image recognition, in particular to a face recognition method, device and equipment applied to an enterprise management system, and the method comprises the steps: obtaining a to-be-detected face image; obtaining non-skin pixel points based on the color features of the pixel points; partitioning the image, and combining a convex hull algorithm to obtain a human face shielding degree; performing connected domain extraction on non-skin pixel points in all convex hulls in the to-be-detected face grayscale image; according to the gradient features and the color features of the edge pixel points of each connected domain in each convex hull, obtaining the suspected degree of the five sense organs of the human face; according to the distribution of the face five-sense-organ suspected degree of the connected domain where the pixel points in the pixel blocks are located, obtaining the information carrying degree; and combining the LBP coding histograms of the pixel blocks of the to-be-detected face grayscale image and the standard face grayscale image to obtain a face recognition result. The method aims at extracting the feature information amount borne by different areas of the face, and the accuracy of face recognition based on the LBPH algorithm is improved.

Description

Technical Field

[0001] This application relates to the field of image recognition technology, and particularly relates to a face recognition method, device, and equipment applied to an enterprise management system. Background Art

[0002] Face recognition is a biometric technology that uses feature extraction methods to represent face features and thus automatically recognize faces. It has the advantages of high user acceptance, good concealment, and the ability to perform concurrent recognition, and is widely used in fields such as content retrieval and identity recognition. The wide application of face recognition technology has brought efficient, safe, and convenient management methods to enterprises, such as realizing multiple functions including employee attendance, access control systems, meeting check-in, visitor management, and security monitoring. Face recognition generally includes three parts: face detection, feature extraction, and pattern recognition. Among them, the feature extraction of the RGB image of the face to be detected is a key link in face recognition.

[0003] The LBPH (Local Binary Pattern Histogram) algorithm is a commonly used feature extraction method in current face recognition and is suitable for the feature extraction of real-time face images. The traditional LBPH algorithm extracts features by calculating the LBP coding histogram of the entire face image. However, it assigns the same importance to the feature information at each position of the face image, which means that the features in different regions contribute equally to recognition. However, in actual applications, the amount of feature information carried by different regions of the face is not the same. Since the LBPH algorithm fails to distinguish these differences, it may cause the omission of regional features that are more critical for recognition during feature extraction. Therefore, this limitation of the traditional LBPH algorithm makes its effect not ideal when extracting the most discriminative features, thereby reducing the accuracy of face recognition. Summary of the Invention

[0004] In view of the above, it is necessary to provide a face recognition method, device, and equipment applied to an enterprise management system to solve the above problems.

[0005] According to one aspect of the present application, there is provided a face recognition method applied to an enterprise management system, the method including:

[0006] Obtain the RGB image and grayscale image corresponding to the face to be detected;

[0007] Based on the color features of each pixel point in the RGB image of the face to be detected, obtain skin pixel points and non-skin pixel points; divide the grayscale image of the face to be detected into blocks, and use the convex hull algorithm for all skin pixel points in the grayscale image of the face to be detected. According to the number distribution of the pixel points in the convex hull in each pixel block, obtain the face occlusion degree of each pixel block;

[0008] Extract the connected components of the non-skin pixel points within all the convex hulls in the face grayscale image to be detected; based on the uniformity of the gradient distribution of the edge pixel points of each connected component within each convex hull and the color characteristics of the pixel points, combined with the color similarity degree between the pixel points, obtain the suspected degree of facial features for each connected component; based on the distribution of the suspected degree of facial features of the connected components where the pixel points in each pixel block are located, combined with the degree of face occlusion, obtain the information carrying degree of each pixel block;

[0009] Apply the LBPH algorithm to the pixel blocks of the face grayscale image to be detected and the standard face grayscale image, and based on the similarity degree between the LBP-coded histograms, combined with the information carrying degree, obtain the result of face recognition.

[0010] Among them, the specific method for obtaining the skin pixel points and non-skin pixel points is as follows:

[0011] Convert the face RGB image to be detected into a YIQ image, preset an I-component interval, and regard all the pixel points whose I-component values of the pixel points in the face RGB image to be detected are within the I-component interval as skin pixel points, otherwise, as non-skin pixel points.

[0012] Among them, the degree of face occlusion of each pixel block is specifically the ratio of the number of pixel points that do not belong to the convex hull in each pixel block to the total number of pixel points in the pixel block.

[0013] Among them, the specific method for obtaining the suspected degree of facial features for each connected component is as follows:

[0014] Based on the uniformity of the gradient distribution of the edge pixel points of each connected component within each convex hull, obtain the contour symmetry degree of each connected component;

[0015] For each connected component, regard the vector composed of the I-component value and the Q-component value of each pixel point in the YIQ image as the color feature vector of each pixel point;

[0016] Based on the similarity degree between the color feature vectors of each pixel point and the pixel points within the local range in each connected component, obtain the color uniformity coefficient of each pixel point;

[0017] For each connected component, calculate the mean value of the cosine similarities between the color feature vectors of all pairwise pixel points, denoted as the first mean value;

[0018] Calculate the mean value of the color uniformity coefficients of all pixel points, denoted as the second mean value;

[0019] Based on the contour symmetry degree of each connected component and the first mean value and the second mean value, obtain the suspected degree of facial features for each connected component; among them, the suspected degree of facial features is positively correlated with the contour symmetry degree and negatively correlated with the first mean value and the second mean value.

[0020] Among them, obtaining the contour symmetry degree of each connected component specifically includes:

[0021] For each convex hull, obtain the gradient direction angles of all edge pixel points of each connected component, obtain the histogram of the gradient direction angles of all edge pixel points, and obtain the contour symmetry degree of each connected component according to the concentration degree of the frequencies corresponding to all intervals in the histogram.

[0022] Among them, obtaining the color uniformity coefficient of each pixel point includes:

[0023] Preset the window of each pixel point, extract the pixel points within the window of each pixel point that are in the same connected component as the pixel point, and take the average level of the similarity degree of the color feature vectors between each pixel point and each extracted pixel point as the color uniformity coefficient of each pixel point.

[0024] Among them, obtaining the information carrying degree of each pixel block specifically includes:

[0025] The suspected degree of facial features of each non-skin pixel point in the convex hull is specifically the suspected degree of facial features of the connected component where the non-skin pixel point is located; the suspected degree of facial features of each skin pixel point in the convex hull is specifically 0;

[0026] Calculate the mean value of the suspected degrees of facial features of all non-skin pixel points in the convex hull in each pixel block, denoted as the third mean value; obtain the ratio of the number of all pixel points within the convex hull in each pixel block to the number of pixel points in the pixel block; fuse the third mean value and the ratio; take the normalized value of the fusion of the negative correlation mapping value of the facial occlusion degree of each pixel block and the fusion result as the information carrying degree of each pixel block.

[0027] Among them, obtaining the result of face recognition specifically includes:

[0028] Denote the face matching degree between the face grayscale image to be detected and the standard face grayscale image as R(m), and its formula form is: In the formula, Lk and L(m,k) respectively represent the LBP coding histograms of the k-th pixel block of the face grayscale image to be detected and the standard face grayscale image; χ() represents the chi-square function; Xk represents the information carrying degree of the k-th pixel block of the face grayscale image to be detected; K represents the number of pixel blocks in the face grayscale image to be detected; α is a preset constant greater than zero; norm() is the normalization function;

[0029] Calculate the maximum value of the face matching degrees between the face grayscale image to be detected and all standard face grayscale images. When the maximum value is greater than the preset matching threshold, the face recognition passes; otherwise, the face recognition fails.

[0030] According to another aspect of the present application, a face recognition device applied to an enterprise management system is provided, including:

[0031] A face image acquisition module, configured to obtain an RGB image and a grayscale image corresponding to a face to be detected;

[0032] A face image analysis module, configured to obtain skin pixel points and non-skin pixel points based on the color features of each pixel point in the RGB image of the face to be detected; divide the grayscale image of the face to be detected into blocks, and adopt a convex hull algorithm for all skin pixel points of the grayscale image of the face to be detected, and obtain the face occlusion degree of each pixel block according to the number distribution of pixel points in the convex hull in each pixel block; perform connected component extraction on non-skin pixel points within all convex hulls in the grayscale image of the face to be detected; obtain the suspected degree of facial features of each connected component according to the uniformity of the gradient distribution of edge pixel points of each connected component within each convex hull and the color features of pixel points, and in combination with the color similarity degree between pixel points; obtain the information carrying degree of each pixel block according to the distribution of the suspected degree of facial features of the connected component where the pixel points in each pixel block are located, in combination with the face occlusion degree;

[0033] A face image recognition module, configured to adopt the LBPH algorithm for pixel blocks of the grayscale image of the face to be detected and the standard grayscale face image, and obtain the result of face recognition based on the similarity degree between LBP encoded histograms, in combination with the information carrying degree.

[0034] According to still another aspect of the present application, a face recognition device applied to an enterprise management system is provided, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the operations corresponding to the above-mentioned face recognition method applied to an enterprise management system are implemented.

[0035] In the above scheme, the RGB image and grayscale image corresponding to the face to be detected are first obtained, and non-skin pixels are obtained based on the color features of the pixels in the RGB image of the face to be detected, which helps to improve the accuracy of subsequent facial feature extraction; the grayscale image of the face to be detected is divided into blocks, and combined with the convex hull algorithm, the face occlusion degree is obtained, which has the beneficial effect of measuring the possibility that the RGB image of the face to be detected is blocked by the occluder; further, the connected domain of all non-skin pixels in the convex hull is extracted, and the facial features suspicion is obtained according to the shape and color features of the connected domain, which has the beneficial effect of distinguishing whether each connected domain is a facial feature, distinguishing the facial feature area and the occluder area from the shape and color features, and providing facial features. The discrimination between the pixel points corresponding to the occluders and the facial features of the person in the convex hull is improved, and the influence of the subsequent adaptive acquisition of the information carrying degree of the LBP coding histogram of the pixel blocks in the grayscale image of the face to be detected is reduced; based on the information carrying degree and the LBP coding histogram, the face recognition of the employee to be detected is completed, and its beneficial effect lies in that the information carrying degree of the facial contour occluders and the facial areas where the facial features are located in the face recognition image is taken into account, and the facial feature information of different pixel blocks in the grayscale image of the face to be detected is accurately evaluated, thereby improving the utilization of the most distinctive information in the detected face image, and thereby improving the accuracy of the facial features extracted based on the LBPH algorithm in rapid face recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 A flowchart of the steps of a face recognition method applied to an enterprise management system provided in this application;

[0037] Figure 2 A schematic diagram of the LBP coding histogram of the pixel block provided in this application;

[0038] Figure 3 A block diagram of a face recognition device applied to an enterprise management system provided in this application. DETAILED DESCRIPTION

[0039] In the description of the embodiments of the present application, words such as "exemplary", "or", "for example" and the like are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary", "or", "for example" and the like is intended to present related concepts in a concrete manner.

[0040] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art in the present application. The terms used in the specification of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application.

[0041] In addition, it should be noted that the terms "first" and "second" in this application and the accompanying drawings are used to distinguish similar objects, rather than to describe a specific order or sequence. For the methods disclosed in the embodiments of this application or shown in the flowcharts, including one or more steps for implementing the methods, without departing from the scope of this application, the execution order of multiple steps can be interchanged with each other, and some steps can also be deleted.

[0042] Please refer to Figure 1 , which shows the step flowchart of a face recognition method applied to an enterprise management system provided by an embodiment of this application. The method includes the following steps:

[0043] Step 1: Obtain the RGB image and grayscale image corresponding to the face to be detected.

[0044] A camera is used to obtain the facial image of the employee to be detected, which is an RGB image. The facial image of the employee to be detected is used as the input of a face detection network (Multi-task convolutional neural network, MTCNN), and the face region image of the employee to be detected is output. Then, geometric normalization processing is performed on the output image respectively to correct the size difference and angle tilt caused by the change of face pose, and the image is normalized to a unified format size to obtain the RGB image of the face to be detected of the employee to be detected.

[0045] Step 2: Based on the color features of each pixel point in the RGB image of the face to be detected, obtain skin pixel points and non-skin pixel points; divide the grayscale image of the face to be detected into blocks, and use the convex hull algorithm for all skin pixel points in the grayscale image of the face to be detected. According to the number distribution of pixel points in the convex hull in each pixel block, obtain the face occlusion degree of each pixel block.

[0046] To improve the utilization degree of the most discriminative information in the RGB image of the face to be detected by the Local Binary Pattern Histogram (LBPH) algorithm, this application divides the grayscale image of the face to be detected transformed from the RGB image of the face to be detected into pixel blocks. By analyzing the distribution of occluders and facial features in the grayscale image of the face to be detected, different weights are assigned to the LBP encoded histograms extracted from different pixel blocks, and the recognition of the grayscale image of the face to be detected of the employee to be detected is completed according to the LBP encoded histograms and weights of all pixel blocks in the grayscale image of the face to be detected.

[0047] Specifically, the RGB image of the face to be detected is converted into a YIQ image. Among them, the conversion of an RGB image to a YIQ image is a well-known technology, and the specific process will not be elaborated here. Since the hue of the facial skin color of Asians generally lies between red and yellow and contains a relatively large amount of yellow component, the I component value of the facial skin color varies within a specific range in the YIQ image. Therefore, a preset I component value interval is used to distinguish skin pixel points and non-skin pixel points in the RGB image of the face to be detected: all pixel points in the RGB image of the face to be detected with I component values within the I component value interval are recorded as skin pixel points; otherwise, they are recorded as non-skin pixel points. Among them, the I component value interval is set to [30, 100] in this embodiment, and the implementer can adjust it by himself.

[0048] The RGB image of the face to be detected is grayscaled to obtain a grayscale image of the face to be detected. Among them, the grayscaling of an RGB image is a well-known technology, and the specific process will not be elaborated here. The grayscale image of the face to be detected is evenly divided into M pixel blocks and numbered. In this embodiment, the divided pixel blocks are numbered in the order from top to bottom and from left to right in the grayscale image of the face to be detected, and all pixel blocks of the grayscale image of the face to be detected are obtained. M takes the value of 16. Among them, the number of pixel blocks and the numbering order of the pixel blocks can be set by the implementer himself, and this application does not limit it.

[0049] Since the face area blocked by facial contour occluders is usually at the contour edge of the face, such as occluders like masks and hair that can block part of the contour edge of the face, the convex hull algorithm is used to obtain all convex hulls of the skin pixel points in the grayscale image of the face to be detected, which is used to represent the face area in the grayscale image of the face to be detected that is not blocked by facial contour occluders. Among them, the convex hull algorithm is a well-known technology, and the specific process will not be elaborated here.

[0050] The ratio of the number of pixel points in each pixel block that do not belong to the convex hull to the total number of pixel points in the pixel block is used as the face occlusion degree of each pixel block.

[0051] It should be understood that the face occlusion degree is used to represent the degree to which the face area where the pixel block is located is blocked by facial contour occluders. The larger the value of the face occlusion degree, the smaller the weight of the LBP coding histogram of the pixel block in subsequent face recognition of the grayscale image of the face to be detected.

[0052] Step 3: Perform connected component extraction on non-skin pixel points within all convex hulls in the face grayscale image to be detected; based on the uniformity of the gradient distribution of edge pixel points in each connected component within each convex hull and the color features of the pixel points, combined with the color similarity degree between pixel points, obtain the suspected degree of facial features for each connected component; based on the distribution of the suspected degree of facial features of the connected components where the pixel points in each pixel block are located, combined with the face occlusion degree, obtain the information carrying degree of each pixel block.

[0053] However, the boundary generated by the convex hull algorithm will include the minimum convex boundary of all objects. However, if the contour of the object is occluded or not fully displayed, the convex hull may extend to the occluder or other non-target areas, thus including some areas that do not belong to the object. Therefore, when using the convex hull algorithm, the area where facial features are located may be partially covered by the occluder, resulting in part of the non-skin area being included in the convex hull boundary.

[0054] Based on the above analysis, perform connected component extraction on non-skin pixel points within all convex hulls in the face grayscale image to be detected. Each connected component represents a non-skin area within the convex hull in the face grayscale image to be detected, and all connected components of the convex hull are obtained. Among them, the extraction of connected components is a well-known technology, and this application will not elaborate on it.

[0055] Generally, the contour edges of facial features have more obvious central symmetry features compared to the contour of the facial occluder, and the facial area where the facial contour occluder is located has a more uniform color distribution compared to the facial feature area.

[0056] Based on the above analysis, for each convex hull, obtain the gradient direction angles of all edge pixel points of each connected component, obtain the histogram of the gradient direction angles of all edge pixel points, and based on the concentration degree of the frequencies corresponding to all intervals in the histogram, obtain the contour symmetry degree of each connected component.

[0057] It should be noted that in this embodiment, the gradient direction angle of each edge pixel point is obtained using the Sobel operator; the concentration degree of multiple variables is measured by the reciprocal of the standard deviation. Among them, the extraction of the Sobel operator and the histogram is a well-known technology, and the specific process will not be elaborated.

[0058] It should be understood that if the number distribution of edge pixel points of all edge pixel points of the connected component in each gradient direction angle is more consistent, the standard deviation is smaller, the central symmetry distribution feature of the edge contour of the connected component is more obvious, and its contour symmetry degree is larger.

[0059] For each connected component, the vector formed by the I - component value and the Q - component value of each pixel in the YIQ image is used as the color feature vector of each pixel; a window for each pixel is preset, and the pixels within the window of each pixel that are in the same connected component as the pixel are extracted. The average level of the similarity degree of the color feature vectors between each pixel and each of the extracted pixels is used as the color uniformity coefficient of each pixel.

[0060] In this embodiment, the window of each pixel is specifically a 3×3 window centered on each pixel; the window size can be set by the implementer himself; the similarity between vectors is measured by the absolute value of the cosine similarity. The larger the absolute value of the cosine similarity, the closer the colors of the pixels are; the mean value of the absolute values of the cosine similarities between each pixel and all other pixels is recorded as the color uniformity coefficient of each pixel, which is used to characterize the uniformity of the color distribution in the face area corresponding to the window area where the pixel is located. Among them, the calculation of the cosine similarity is a well - known technology, and the specific process will not be elaborated here.

[0061] Furthermore, according to the contour symmetry degree of each connected component and the color uniformity coefficients of all pixels, combined with the color similarity degree between pixels, the suspected degree of facial features of each connected component is obtained:

[0062] For each connected component, calculate the mean value of the cosine similarities between the color feature vectors of all pixels, and record it as the first mean value; calculate the mean value of the color uniformity coefficients of all pixels, and record it as the second mean value; based on the contour symmetry degree of each connected component and the first mean value and the second mean value, obtain the suspected degree of facial features of each connected component; among them, the suspected degree of facial features is positively correlated with the contour symmetry degree and negatively correlated with the first mean value and the second mean value.

[0063] In this embodiment, the suspected degree of facial features is specifically calculated by multiplying the first mean value and the second mean value, and taking the ratio of the contour symmetry degree to the product as the suspected degree of facial features. It should be noted that in order to prevent the denominator from being 0, a preset value of 0.01 needs to be added to the denominator, and the implementer can adjust it by himself.

[0064] It should be understood that the more obvious the central - symmetric distribution feature of the edge contour corresponding to the connected component, that is, the greater the contour symmetry degree, and the more uneven the color distribution within the connected component, that is, the smaller the overall color uniformity coefficient of all pixels within the connected component, and the smaller the cosine similarity of the color feature vectors between all the pixels, the greater the possibility that the connected component belongs to facial features in the grayscale face image to be detected, that is, the greater the suspected degree of facial features.

[0065] According to the distribution of the suspected degrees of facial features of the pixel points in each pixel block within the connected component, and in combination with the degree of facial occlusion, the information carrying degree of each pixel block is obtained:

[0066] If the nth pixel point in the convex hull is a non-skin pixel point, then the suspected degree of facial features of the connected component where the corresponding pixel point is located is recorded as the suspected degree of facial features of the pixel point; if the nth pixel point in the convex hull is a skin pixel point, then the suspected degree of facial features of the corresponding pixel point is assigned a value of 0.

[0067] Calculate the average value of the suspected degrees of facial features of all non-skin pixel points in the convex hull in each pixel block, and record it as the third average value; obtain the ratio of the number of all pixel points within the convex hull in each pixel block to the number of pixel points in the pixel block; fuse the average value and the ratio; the normalized value obtained by fusing the negative correlation mapping value of the degree of facial occlusion of each pixel block with the fusion result is used as the information carrying degree of each pixel block.

[0068] In this embodiment, when fusing multiple variables, the multiplication method is adopted, and the negative correlation mapping value is measured by the reciprocal; the normalization method adopts the maximum-minimum normalization method. Specifically, the information carrying degree of each pixel block refers to calculating the product of the third average value and the ratio, and using the ratio of the product to the degree of facial occlusion as the information carrying degree of each pixel block. It should be noted that in order to prevent the denominator from being 0, a preset value needs to be added to the denominator, and the value is 0.01, and the implementer can adjust it by himself.

[0069] It should be understood that in face recognition, since the facial features in the RGB image of the face to be detected can better reflect the face characteristics, and the occluders in the RGB image of the face to be detected will submerge or distort the face characteristics, the smaller the degree of occlusion of the pixel block by the facial contour occluder, and the larger the proportion and the suspected degree of the pixel points belonging to the facial features of the face in the pixel block, the more facial feature information carried by the LBP coding histogram of the pixel block, that is, the larger the information carrying degree, the more the LBP coding histogram of the pixel block can reflect the face characteristics of the face area corresponding to the pixel block.

[0070] Step 4: Apply the LBPH algorithm to the pixel blocks of the grayscale image of the face to be detected and the standard grayscale face image, and based on the similarity degree between the LBP coding histograms, in combination with the information carrying degree, obtain the result of face recognition.

[0071] Use the LBPH algorithm to extract the LBP-coded histograms of each pixel block of the grayscale face image to be detected respectively. Among them, in this embodiment, the scale of the LBPH algorithm is selected as (8,1), and the uniform pattern is adopted to obtain the LBP-coded histograms of each pixel block for subsequent face recognition of the grayscale face image to be detected. The LBPH algorithm is a well-known technology, and the specific process will not be elaborated here; in addition, the schematic diagram of the LBP-coded histogram of the pixel block is as shown in Figure 2 shown.

[0072] For the m-th standard grayscale face image in the database of the face recognition module, use the same method as the LBP-coded histogram of the grayscale face image to be detected of the employee to be detected to obtain the LBP-coded histogram of each pixel block of the standard grayscale face image, and obtain the face matching degree between the grayscale face image to be detected and the standard grayscale face image:

[0073] According to the similarity degree between the LBP-coded histograms of the pixel blocks of the grayscale face image to be detected and the standard grayscale face image, combined with the information carrying degree, obtain the face matching degree between the grayscale face image to be detected and the standard grayscale face image:

[0074] Denote the face matching degree between the grayscale face image to be detected and the standard grayscale face image as R(m), and its formula form is: In the formula, Lk and L(m,k) respectively represent the LBP-coded histograms of the k-th pixel block of the grayscale face image to be detected and the standard grayscale face image; χ() represents the chi-square function, and the smaller the result of the χ() function, the more similar the two histograms are. The chi-square comparison method is a well-known technology, and the specific process will not be elaborated here; Xk represents the information carrying degree of the k-th pixel block of the grayscale face image to be detected; K represents the number of pixel blocks in the grayscale face image to be detected; α is a preset constant greater than zero. To prevent the denominator from being 0, where α takes the value of 0.01; norm() is a normalization function, and the sigmoid function is adopted.

[0075] It should be understood that the greater the information carrying degree of the LBP-coded histogram of the pixel block, the greater the influence of the histogram similarity degree between the LBP-coded histograms of the k pixel blocks on the face matching result between the grayscale face image to be detected and the standard grayscale face image. If the similarity degree between the LBP-coded histograms is higher, that is, χ(Lk,L(m,k)) is smaller, the grayscale face image to be detected and the standard grayscale face image have more similar face features, that is, the face matching degree is greater, and the employee corresponding to the standard grayscale face image is more likely to be the face recognition result of the employee to be detected.

[0076] Calculate the face matching degree between the grayscale image of the face to be detected and each standard face grayscale image in the database respectively. Set the matching threshold r = 0.9, which can be set by the implementer himself / herself. If the maximum value among the obtained face matching degrees is greater than the matching threshold, obtain the employee name corresponding to the standard face grayscale image corresponding to the maximum value from the face recognition module. For example, if it is Li Ming, then the voice broadcast module of the face recognition module will emit the voice "Li Ming, face recognition passed", otherwise the voice broadcast module will emit the voice "Face recognition failed", thus completing the rapid face recognition of the employee to be detected.

[0077] Based on the same concept as the method embodiment of this application, a face recognition device applied to an enterprise management system is proposed, including:

[0078] A face image acquisition module, used to acquire the RGB image and grayscale image corresponding to the face to be detected;

[0079] A face image analysis module, which is used to obtain skin pixel points and non-skin pixel points based on the color characteristics of each pixel point in the RGB image of the face to be detected; divide the grayscale image of the face to be detected into blocks, and use the convex hull algorithm for all skin pixel points of the grayscale image of the face to be detected. According to the number distribution of pixel points in the convex hull in each pixel block, obtain the face occlusion degree of each pixel block; perform connected component extraction on the non-skin pixel points within all convex hulls in the grayscale image of the face to be detected; according to the uniformity of the gradient distribution of the edge pixel points of each connected component within each convex hull and the color characteristics of the pixel points, combined with the color similarity degree between the pixel points, obtain the suspected degree of facial features of each connected component; according to the distribution of the suspected degree of facial features of the connected components where the pixel points are located in each pixel block, combined with the face occlusion degree, obtain the information carrying degree of each pixel block;

[0080] A face image recognition module, which is used to adopt the LBPH algorithm for the pixel blocks of the grayscale image of the face to be detected and the standard face grayscale image, and based on the similarity degree between the LBP encoded histograms, combined with the information carrying degree, obtain the result of face recognition.

[0081] Among them, a block diagram of a face recognition device applied to an enterprise management system is as Figure 3 shown.

[0082] Based on the same concept as the method embodiment of this application, a face recognition device applied to an enterprise management system is proposed, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the operations corresponding to the above-mentioned face recognition method applied to an enterprise management system.

[0083] In summary, first, obtain the RGB image and grayscale image corresponding to the face to be detected. Based on the color characteristics of the pixel points in the RGB image of the face to be detected, non-skin pixel points are obtained, which helps to improve the accuracy of subsequent face feature extraction. Divide the grayscale image of the face to be detected into blocks, and combine with the convex hull algorithm to obtain the face occlusion degree. The beneficial effect is to measure the possibility that the RGB image of the face to be detected is occluded by an occluder. Further, extract the connected components of all non-skin pixel points within the convex hull. According to the shape characteristics and color characteristics of the connected components, obtain the suspected degree of facial features. The beneficial effect is to distinguish whether each connected component is a facial feature, distinguish the facial feature area and the occluder area from the shape and color characteristics, improve the distinguishability between the occluder and the pixel points corresponding to the facial features of a person in the convex hull, and reduce the influence of the information carrying degree of the subsequent adaptive acquisition of the LBP coding histogram of the pixel blocks in the grayscale image of the face to be detected. Based on the information carrying degree and the LBP coding histogram, complete the face recognition of the employee to be detected. The beneficial effect is to consider the information carrying degree of the facial contour occluder and the face area where the facial feature area is located in the face recognition image, accurately evaluate the amount of face feature information of different pixel blocks in the grayscale image of the face to be detected, improve the utilization degree of the most discriminative information in the detected face image, and thus improve the accuracy of the face fast recognition based on the face features extracted by the LBPH algorithm.

[0084] It should be noted that the flowcharts and block diagrams in the accompanying drawings show the possible architectures, functions, and operations of the systems, methods, and computer program products according to the embodiments of the present application. In this regard, each block in the flowchart or block diagram may represent a module, a program segment, or a part of code, and the module, program segment, or part of code contains one or more executable instructions for implementing the specified logical function. In some alternative implementations, the functions marked in the blocks may also occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, which may depend on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different blocks may also occur in a different order than that disclosed in the description. Sometimes, there is no specific order between different operations or steps. For example, two consecutive operations or steps may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, which may depend on the functions involved. Each block in the block diagram and / or flowchart, as well as the combination of the blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0085] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the various embodiments of the present application, and should all be included within the protection scope of the present application.

Claims

1. A face recognition method applied to an enterprise management system, characterized in that, The method includes the following steps: Obtain the RGB image and grayscale image corresponding to the face to be detected; Based on the color features of each pixel point in the RGB image of the face to be detected, obtain skin pixel points and non-skin pixel points; divide the grayscale image of the face to be detected into blocks, and use the convex hull algorithm for all skin pixel points in the grayscale image of the face to be detected. According to the distribution of the number of pixel points in each pixel block that are within the convex hull, obtain the face occlusion degree of each pixel block; Extract the connected components of all non-skin pixel points within all convex hulls in the grayscale image of the face to be detected; according to the uniformity of the gradient distribution of the edge pixel points of each connected component within each convex hull and the color features of the pixel points, combined with the color similarity degree between pixel points, obtain the suspected degree of facial features of each connected component; according to the distribution of the suspected degree of facial features of the connected components where the pixel points in each pixel block are located, combined with the face occlusion degree, obtain the information carrying degree of each pixel block; Use the LBPH algorithm for the pixel blocks of the grayscale image of the face to be detected and the standard grayscale image of the face. Based on the similarity degree between the LBP encoded histograms, combined with the information carrying degree, obtain the result of face recognition.

2. The face recognition method applied to an enterprise management system according to claim 1, characterized in that, The obtaining of the skin pixel points and non-skin pixel points is specifically as follows: Convert the RGB image of the face to be detected into a YIQ image, preset an I component interval, and take all pixel points in the RGB image of the face to be detected whose I component values are within the I component interval as skin pixel points, otherwise, as non-skin pixel points.

3. The face recognition method applied to an enterprise management system according to claim 1, characterized in that, The face occlusion degree of each pixel block is specifically the ratio of the number of pixel points that do not belong to the convex hull in each pixel block to the total number of pixel points in the pixel block.

4. The face recognition method applied to an enterprise management system according to claim 2, characterized in that, The obtaining of the suspected degree of facial features of each connected component is specifically as follows: According to the uniformity of the gradient distribution of the edge pixel points of each connected component within each convex hull, obtain the contour symmetry degree of each connected component; For each connected component, use the vector composed of the I component value and the Q component value of each pixel point in the YIQ image as the color feature vector of each pixel point; According to the similarity degree between the color feature vectors of each pixel point within each connected component and the pixel points within the local range, obtain the color uniformity coefficient of each pixel point; For each connected component, calculate the mean value of the cosine similarities between the color feature vectors of all pairwise pixel points, denoted as the first mean value; Calculate the mean value of the color uniformity coefficients of all pixel points, denoted as the second mean value; Based on the contour symmetry degree of each connected component and the first mean value and the second mean value, obtain the suspected degree of facial features of each connected component; wherein, the suspected degree of facial features is positively correlated with the contour symmetry degree and negatively correlated with the first mean value and the second mean value.

5. The face recognition method applied to an enterprise management system according to claim 4, wherein, The obtaining of the contour symmetry degree of each connected component is specifically as follows: For each convex hull, obtain the gradient direction angles of all edge pixel points of each connected component, obtain the histogram of the gradient direction angles of all edge pixel points, and according to the concentration degree of the frequencies corresponding to all intervals in the histogram, obtain the contour symmetry degree of each connected component.

6. The face recognition method applied to an enterprise management system according to claim 4, wherein, The obtaining of the color uniformity coefficient of each pixel point includes: Preset windows for each pixel point, extract the pixel points within the window of each pixel point that are in the same connected component as the pixel point, and use the average level of the similarity degree of the color feature vectors between each pixel point and each extracted pixel point as the color uniformity coefficient of each pixel point.

7. The face recognition method applied to an enterprise management system according to claim 1, characterized in that, The obtaining of the information carrying degree of each pixel block is specifically as follows: The suspected degree of human face features of each non-skin pixel point in the convex hull is specifically the suspected degree of human face features of the connected component where the non-skin pixel point is located; the suspected degree of human face features of each skin pixel point in the convex hull is specifically 0; Calculate the mean value of the suspected degrees of human face features of all non-skin pixel points in the convex hull in each pixel block, denoted as the third mean value; obtain the ratio of the number of all pixel points within the convex hull in each pixel block to the number of pixel points in the pixel block; fuse the third mean value and the ratio; Use the normalized value obtained by fusing the negative correlation mapping value of the face occlusion degree of each pixel block with the fusion result as the information carrying degree of each pixel block.

8. The face recognition method applied to an enterprise management system according to claim 1, characterized in that, The obtaining of the result of face recognition is specifically as follows: The face matching degree between the grayscale image of the face to be detected and the grayscale image of the standard face is denoted as R(m), and its formula form is as follows: In the formula, Lk and L(m,k) respectively represent the LBP coding histograms of the k-th pixel block of the grayscale image of the face to be detected and the grayscale image of the standard face; χ() represents the chi-square function; Xk represents the information carrying degree of the k-th pixel block of the grayscale image of the face to be detected; K represents the number of pixel blocks in the grayscale image of the face to be detected; α is a preset constant greater than zero; norm() is a normalization function; Calculate the maximum value of the face matching degrees between the face gray-scale image to be detected and all standard face gray-scale images. When the maximum value is greater than the preset matching threshold, the face recognition passes; otherwise, the face recognition fails.

9. A face recognition device applied to an enterprise management system, characterized in that, The device includes: A face image acquisition module, configured to acquire the RGB image and the gray-scale image corresponding to the face to be detected; A face image analysis module, configured to obtain skin pixel points and non-skin pixel points based on the color features of each pixel point in the RGB image of the face to be detected; divide the gray-scale image of the face to be detected into blocks, and use the convex hull algorithm for all skin pixel points in the gray-scale image of the face to be detected. According to the number distribution of the pixel points within the convex hull in each pixel block, obtain the face occlusion degree of each pixel block; perform connected component extraction on all non-skin pixel points within the convex hull in the gray-scale image of the face to be detected; according to the uniformity of the gradient distribution of the edge pixel points of each connected component within each convex hull and the color features of the pixel points, combined with the color similarity degree between the pixel points, obtain the suspected degree of human face features of each connected component; according to the distribution of the suspected degrees of human face features of the connected components where the pixel points in each pixel block are located, combined with the face occlusion degree, obtain the information carrying degree of each pixel block; A face image recognition module, configured to use the LBPH algorithm for the pixel blocks of the gray-scale image of the face to be detected and the standard face gray-scale images, and based on the similarity degree between the LBP encoded histograms, combined with the information carrying degree, obtain the result of face recognition.

10. A face recognition device applied to an enterprise management system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the operations corresponding to the face recognition method applied to an enterprise management system according to any one of claims 1-8.

Images