A face recognition method, system, storage medium and computing device

Abstract

The application discloses a face recognition method and system, a storage medium and a computing device, wherein the application periodically corrects the registered face image feature data in the database according to the aging index, avoids the recognition error caused by the age change, corrects the registered face image feature data based on the pitch angle threshold of the image acquisition device and the face to be recognized during face recognition, and performs face recognition based on the corrected feature data, so that the error caused by the pitch angle of the image acquisition device and the face to be recognized is reduced, and the recognition is more accurate compared with the traditional recognition method.

Description

Technical Field

[0001] This invention relates to a face recognition method, system, storage medium, and computing device, belonging to the field of computer technology. Background Technology

[0002] Facial recognition is a biometric technology that identifies individuals based on their facial features. It involves capturing images containing human faces, automatically detecting and tracking faces within those images, and then performing facial recognition. However, current facial recognition systems often fail to account for errors caused by age changes or the tilt angle between the face and the capturing device, resulting in relatively poor accuracy. Summary of the Invention

[0003] This invention provides a face recognition method, system, storage medium, and computing device, which solves the problems disclosed in the background art.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0005] A facial recognition method includes:

[0006] Acquire the face image to be identified from the image acquisition device, and extract the feature data of the face image to be identified;

[0007] The first similarity between the feature data of the face image to be identified and the feature data of each registered face image is calculated respectively; wherein, the registered face image feature data are face image feature data pre-stored in the base database, and the registered face image feature data in the base database is periodically corrected according to a preset aging index;

[0008] If all first similarities are less than the similarity threshold, obtain the N first similarities that are closest to the similarity threshold, and calculate the percentage difference between the N first similarities and the similarity threshold for each; where N is a preset number.

[0009] If the M percentage differences are less than a preset percentage, the registered face image feature data corresponding to the M percentage differences are corrected according to the tilt angle threshold between the image acquisition device and the face to be identified; where 0 <M≤N;

[0010] Calculate the second similarity between the feature data of the face image to be identified and the feature data of the corrected registered face image;

[0011] If any second similarity score is not less than the similarity threshold, the face image to be identified matches the corresponding corrected registered face image feature data, and the face recognition is successful.

[0012] The pre-stored facial image feature data in the base database includes:

[0013] Acquire registered facial images from image acquisition devices and extract feature data from the registered facial images;

[0014] If the brightness value in the registered face image feature data is within the preset brightness range and the clarity is within the preset high-resolution range, then the registered face image feature data will be stored in the base database.

[0015] If the brightness value in the registered face image feature data is within a preset brightness range and the clarity is within a preset medium clarity range, the contrast of the registered face image is enhanced based on histogram equalization until the clarity is within a preset high clarity range, and then the registered face image feature data is stored in the base database.

[0016] The aging index includes age groups and age-matched facial features; the facial features include eyebrows, eyes, nose, lips, and the lower edge of the facial contour. The aging index is the percentage increase or decrease in the area of ​​the facial features in the following year compared to the previous year within the age group.

[0017] Based on the pitch angle threshold between the image acquisition device and the face to be identified, the registered face image feature data corresponding to M difference percentages are corrected, including:

[0018] Calculate the cosine value of the pitch angle threshold between the image acquisition device and the face to be identified;

[0019] The cosine value is used to correct the registered face image feature data corresponding to the M difference percentages.

[0020] A facial recognition system, comprising:

[0021] The feature extraction module acquires the face image to be identified from the image acquisition device and extracts the feature data of the face image to be identified;

[0022] The first similarity module calculates the first similarity between the feature data of the face image to be identified and the feature data of each registered face image; wherein, the registered face image feature data are face image feature data pre-stored in the base database, and the registered face image feature data in the base database are periodically corrected according to a preset aging index;

[0023] In the filtering module, if all first similarities are less than the similarity threshold, the N first similarities closest to the similarity threshold are obtained, and the percentage difference between the N first similarities and the similarity threshold is calculated for each; where N is a preset number.

[0024] The geometric correction module, if the M percentage differences are less than a preset percentage, corrects the registered face image feature data corresponding to the M percentage differences based on the tilt angle threshold between the image acquisition device and the face to be identified; where 0 <M≤N;

[0025] The second similarity module calculates the second similarity between the feature data of the face image to be identified and the feature data of the corrected registered face image.

[0026] If any second similarity is not less than the similarity threshold, the face image to be identified matches the corresponding corrected registered face image feature data, and the face recognition is successful.

[0027] It also includes a pre-storage module for pre-storing facial image feature data into the base database. The pre-storage process includes:

[0028] Acquire registered facial images from image acquisition devices and extract feature data from the registered facial images;

[0029] If the brightness value in the registered face image feature data is within the preset brightness range and the clarity is within the preset high-resolution range, then the registered face image feature data will be stored in the base database.

[0030] If the brightness value in the registered face image feature data is within a preset brightness range and the clarity is within a preset medium clarity range, the contrast of the registered face image is enhanced based on histogram equalization until the clarity is within a preset high clarity range, and then the registered face image feature data is stored in the base database.

[0031] The aging index includes age groups and age-matched facial features; the facial features include eyebrows, eyes, nose, lips, and the lower edge of the facial contour. The aging index is the percentage increase or decrease in the area of ​​the facial features in the following year compared to the previous year within the age group.

[0032] In the geometric correction module, based on the pitch angle threshold between the image acquisition device and the face to be identified, the registered face image feature data corresponding to M difference percentages are corrected, including:

[0033] Calculate the cosine value of the pitch angle threshold between the image acquisition device and the face to be identified;

[0034] The cosine value is used to correct the registered face image feature data corresponding to the M difference percentages.

[0035] A computer-readable storage medium storing one or more programs, the one or more programs including instructions that, when executed by a computing device, cause the computing device to perform a face recognition method.

[0036] A computing device includes one or more processors, one or more memories, and one or more programs, wherein the one or more programs are stored in the one or more memories and configured to be executed by the one or more processors, and the one or more programs include instructions for performing a face recognition method.

[0037] The beneficial effects achieved by this invention are as follows: This invention periodically corrects the registered facial image feature data in the database based on the aging index, avoiding recognition errors caused by age changes. During facial recognition, the registered facial image feature data is corrected based on the pitch angle threshold between the image acquisition device and the face to be recognized. Facial recognition is then performed based on the corrected feature data, reducing errors caused by the pitch angle between the image acquisition device and the face to be recognized. Compared with traditional recognition methods, the recognition is more accurate. Attached Figure Description

[0038] Figure 1 A flowchart of a face recognition method;

[0039] Figure 2 Functional block diagram of the company's intelligent service system;

[0040] Figure 3 Facial image feature data for 68 individuals, including male A, aged 25.

[0041] Figure 4 Facial image feature data for 68 individuals, including male A, aged 26.

[0042] Figure 5 Facial image feature data for 68 individuals, including male A, aged 30.

[0043] Figure 6 The facial image feature data in the base database;

[0044] Figure 7 This is facial image feature data corrected based on pitch angle threshold. Detailed Implementation

[0045] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0046] like Figure 1 As shown, a face recognition method includes:

[0047] Step 1: Acquire the face image to be identified from the image acquisition device and extract the feature data of the face image to be identified;

[0048] Step 2: Calculate the first similarity between the feature data of the face image to be identified and the feature data of each registered face image; wherein, the registered face image feature data are face image feature data pre-stored in the base database, and the registered face image feature data in the base database are periodically corrected according to a preset aging index;

[0049] Step 3: If all first similarities are less than the similarity threshold, obtain the N first similarities that are closest to the similarity threshold, and calculate the percentage difference between the N first similarities and the similarity threshold for each; where N is a preset number.

[0050] If any of the first similarity scores is not less than the similarity threshold, the face image to be identified matches the corresponding registered face image feature data, and the face recognition is successful.

[0051] Step 4: If the M percentage differences are less than a preset percentage, adjust the registered face image feature data corresponding to the M percentage differences based on the pitch angle threshold between the image acquisition device and the face to be identified; where 0 <M≤N;

[0052] If all the difference percentages are not less than the preset percentage, then the face recognition fails, meaning that the face to be recognized is not registered.

[0053] Step 5: Calculate the second similarity between the feature data of the face image to be identified and the feature data of the corrected registered face image;

[0054] Step 6: If any second similarity is not less than the similarity threshold, the face image to be identified matches the corresponding corrected registered face image feature data, and the face recognition is successful.

[0055] If all second similarities are less than the similarity threshold, then face recognition fails, meaning the face to be recognized is not registered.

[0056] The above method periodically corrects the registered face image feature data in the database based on the aging index to avoid recognition errors caused by age changes. During face recognition, the registered face image feature data is corrected based on the pitch angle threshold between the image acquisition device and the face to be recognized. Face recognition is then performed based on the corrected feature data, reducing the error caused by the pitch angle between the image acquisition device and the face to be recognized. Compared with traditional recognition methods, the recognition is more accurate.

[0057] Before implementing the above-mentioned face recognition, it is necessary to build a base database, that is, to register faces. The base database pre-stores face image feature data, which can select 68 feature points related to the eyebrows, eyes, nose, lips and lower edge of the outer contour of the face.

[0058] The pre-stored facial image feature data in the base database can be as follows:

[0059] S1) Obtain the registered face image from the image acquisition device and extract the feature data of the registered face image.

[0060] S2) If the brightness value in the registered face image feature data is within the preset brightness range, proceed to S3). If the brightness value in the registered face image feature data is not within the preset brightness range, discard the feature data and re-acquire the face image.

[0061] This step mainly evaluates the brightness value within the face area. If the brightness value is uniform and normal, it passes; if it is partially or completely too bright or too dark, it fails. The normal brightness range is generally 0-255, and the brightness range that does not affect the recognition result is 60-210. That is, the preset brightness range is set at 60-210. Values ​​less than 60 or greater than 210 will fail.

[0062] S3) If the sharpness of the registered face image feature data is within the preset range of poor sharpness, discard the feature data and re-acquire the face image; wherein, the range of poor sharpness is a sharpness of less than 0.08;

[0063] If the clarity of the registered face image feature data is within the preset high-resolution range, the registered face image feature data will be stored in the base database; where the high-resolution range is defined as a clarity greater than 0.2.

[0064] If the clarity of the registered face image feature data is within a preset medium clarity range, the contrast of the registered face image is enhanced based on histogram equalization until the clarity is within a preset high clarity range, and then the registered face image feature data is stored in the base database; wherein, the medium clarity range is 0.08 to 0.2.

[0065] Facial features change with age. According to the Stanford University timeline of human growth, development, and aging, without considering the effects of makeup or cosmetic procedures: Before age 18, facial features tend towards maturity, specifically manifested as thicker eyebrows, unchanged eye spacing, wider nose, wider lips, and a wider lower edge of the face. From 18 to 25 years old, these features are relatively stable. From 25 to 40 years old, eyebrows are relatively stable, eye sockets become smaller, the tip of the nose droops, the corners of the mouth droop, and the lower edge of the face widens. After age 40, the aforementioned changes in facial features continue, with eyebrows becoming longer. After age 60, aging becomes more pronounced, with the most significant changes occurring at the lower edge of the face.

[0066] The changes in facial features with age are quantified by setting age groups and facial feature aging indices that match those age groups. Facial features include eyebrows, eyes, nose, lips, and the lower edge of the facial contour. The aging index is the percentage increase or decrease in the area of ​​a facial feature in the following year compared to the previous year within the age group. See Table 1 below for details:

[0067] Table 1. Aging Index and its Relationship with Table 2

[0068]

[0069]

[0070] In the table, the first column is the age range, and the aging index is the change in the area of ​​the face. A positive value indicates the proportion of area increase, 0 indicates that it remains unchanged, and a negative value indicates the proportion of area decrease. For example, in the age range (40, 60), the eyebrows of a 42-year-old are 0.6% larger than those of a 41-year-old, and the eyes of a 42-year-old are 1.3% smaller than those of a 41-year-old.

[0071] Based on the table above, the registered facial image feature data in the database is periodically corrected. The correction cycle is generally set to one year, that is, the database is updated once a year.

[0072] like Figure 3 The facial image feature data of 68 individuals, including male A, aged 25, was corrected one year later based on an aging index. The corrected facial image feature data is as follows: Figure 4 As shown, the facial image feature data corrected after 5 years is as follows: Figure 5 As shown.

[0073] The process of facial recognition based on the latest database is as follows:

[0074] A1) Obtain the face image to be identified and extract the feature data of the face image to be identified.

[0075] A2) Calculate the first similarity between the feature data of the face image to be identified and the feature data of each registered face image.

[0076] A3) If any first similarity score is not less than the similarity threshold, the face image to be identified matches the corresponding registered face image feature data, and the face recognition is successful.

[0077] If all first similarities are less than the similarity threshold, obtain the N first similarities that are closest to the similarity threshold, and calculate the percentage difference between the N first similarities and the similarity threshold (subtract the first similarity from the similarity threshold and divide by the similarity threshold), then go to A4); where N can be defined as 10;

[0078] The similarity threshold varies depending on the situation. For facial recognition in access control systems, the similarity threshold is 0.56. For facial recognition during facial authorization and signature verification, the similarity threshold is 0.62. For facial recognition during attendance verification, the similarity threshold is 0.59.

[0079] A4) If all the difference percentages are not less than the preset percentage, then the face recognition fails, that is, the face to be recognized is not registered.

[0080] If the percentage of the M differences is less than the preset percentage, the registered face image feature data corresponding to the percentage of the M differences is corrected according to the pitch angle threshold between the image acquisition device and the face to be identified (the pitch angle range is -30 degrees to 30 degrees, i.e., the pitch angle threshold is 30 degrees), and then proceed to A5; where the preset percentage is 20%.

[0081] When image acquisition devices capture faces, the plane where the lens is located is often not parallel to the plane where the face is located. Since the true pitch angle between the image acquisition device and the face to be recognized cannot be captured, a pitch angle threshold is set here. The cosine value of the pitch angle threshold between the image acquisition device and the face to be recognized can be calculated. The cosine value is used to correct the registered face image feature data corresponding to M difference percentages, thereby reducing the error caused by the pitch angle.

[0082] like Figure 6 The facial image feature data in the uncorrected base database, after cosine correction using the pitch angle threshold, is as follows: Figure 7 As shown.

[0083] A5) Calculate the second similarity between the feature data of the face image to be identified and the feature data of the corrected registered face image.

[0084] A6) If any second similarity is not less than the similarity threshold, the face image to be identified matches the corresponding corrected registered face image feature data, and the face recognition is successful;

[0085] If all second similarities are less than the similarity threshold, then face recognition fails, meaning the face to be recognized is not registered.

[0086] The above method can be applied to the company's intelligent service system for facial recognition access control, attendance tracking, and authorization signature processing; these functions share a common facial recognition database. The functional modules of the company's intelligent service system are as follows: Figure 2The process involves collecting employee facial information during registration. This information is then linked to the employee's ID, organizational affiliation, and personal details in the personnel information management module. Facial information processing, including face detection and key point localization, is then performed. For face detection, a higher minimum face parameter setting results in faster computation, but also increases the performance requirements of the handheld device. Considering both computational speed and compatibility, the parameter value is set to 75 pixels. The maximum detectable image width and maximum detectable image height are related settings, both set to 1500 pixels. The maximum height and width represent the actual height detected by the algorithm. To ensure computational speed and device performance, if the width or height of the input image exceeds the limit, the image will be automatically scaled down to the resolution of the device. To guarantee good face matting results, the aspect ratio is set to 3:4. Sixty-eight feature points representing the eyebrows, eyes, nose, lips, and lower edge of the facial contour were selected as key points. These features were extracted and stored as a one-dimensional array.

[0087] Considering that the application scenario is mainly users aged 20-60, every year, the base database features are biologically corrected according to rows 3-5 of Table 1 below, the area of ​​organs is corrected, and the base database facial feature data is updated.

[0088] Pre-storage corrections are performed as needed, evaluating and correcting the brightness and sharpness of facial feature data. Only corrected facial features that pass the evaluation are saved in the facial feature database. The passing range for brightness evaluation is 60–210 pixels, and a sharpness greater than 0.2 is considered passing. For facial features with sharpness between 0.08 and 0.2, histogram equalization is used to enhance image contrast, transforming the image's grayscale from the center of the comparison set to a uniform distribution across the entire grayscale range.

[0089] In the application field, the faces captured by the acquisition device are first preprocessed, including face detection and key point localization. The specific operations and parameter settings are the same as those in the personnel registration and face database establishment process. After extracting the facial feature data, face comparison is performed. During this comparison process, geometric correction is performed as needed (i.e., the cosine value correction mentioned above), and the user's geometrically corrected facial features are stored in the face database, because in the application scenario, the user is highly likely to use the geometrically corrected facial feature data again.

[0090] When facial recognition access control systems are used to pass through turnstiles during peak commuting hours, high throughput is required, so the similarity threshold is set at 0.56. When using facial recognition for authorization and signature, high accuracy is the primary goal, so the similarity threshold is set at 0.62. In facial recognition attendance applications, both accuracy and speed need to be considered, so the similarity threshold is set at 0.59.

[0091] In facial recognition access control applications, new records are inserted into the personnel entry and exit database; in attendance check-in applications, new records are added into the attendance check-in database; and in authorized signature applications, new records are added into the signature record database.

[0092] Application results show that correcting facial features in the database can indeed improve the accuracy of facial recognition, especially in corporate facial recognition access control scenarios, reducing the probability of failure and subsequent manual verification.

[0093] Based on the same technical solution, this invention also discloses a software system for the above method, a face recognition system, comprising:

[0094] The feature extraction module acquires the face image to be identified from the image acquisition device and extracts the feature data of the face image to be identified.

[0095] The first similarity module calculates the first similarity between the feature data of the face image to be identified and the feature data of each registered face image. The registered face image feature data are face image feature data pre-stored in the base database. The registered face image feature data in the base database is periodically corrected according to a preset aging index. The aging index includes age group and aging index of facial parts matching the age group. The facial parts include eyebrows, eyes, nose, lips and the lower edge of the outer contour of the face. The aging index is the percentage increase or decrease in the area of ​​the facial parts in the following year compared to the previous year within the age group.

[0096] In the filtering module, if all first similarities are less than the similarity threshold, the N first similarities closest to the similarity threshold are obtained, and the percentage difference between the N first similarities and the similarity threshold is calculated for each; where N is a preset number.

[0097] The geometric correction module, if the M percentage differences are less than a preset percentage, corrects the registered face image feature data corresponding to the M percentage differences based on the tilt angle threshold between the image acquisition device and the face to be identified; where 0 <M≤N。

[0098] In the geometric correction module, based on the pitch angle threshold between the image acquisition device and the face to be identified, the registered face image feature data corresponding to M difference percentages are corrected, including: calculating the cosine value of the pitch angle threshold between the image acquisition device and the face to be identified; and using the cosine value to correct the registered face image feature data corresponding to the M difference percentages.

[0099] The second similarity module calculates the second similarity between the feature data of the face image to be identified and the feature data of the corrected registered face image.

[0100] If any second similarity is not less than the similarity threshold, the face image to be identified matches the corresponding corrected registered face image feature data, and the face recognition is successful.

[0101] The pre-storage module is used to pre-store facial image feature data into the base database. The pre-storage process includes:

[0102] 1) Acquire registered facial images from image acquisition devices and extract feature data from the registered facial images;

[0103] 2) If the brightness value in the registered face image feature data is not within the preset brightness range, or the clarity is within the preset poor clarity range, discard the feature data of the registered face image.

[0104] If the brightness value in the registered face image feature data is within the preset brightness range and the clarity is within the preset high-resolution range, then the registered face image feature data will be stored in the base database.

[0105] If the brightness value in the registered face image feature data is within a preset brightness range and the clarity is within a preset medium clarity range, the contrast of the registered face image is enhanced based on histogram equalization until the clarity is within a preset high clarity range, and then the registered face image feature data is stored in the base database.

[0106] Based on the same technical solution, the present invention also discloses a computer-readable storage medium for storing one or more programs, wherein the one or more programs include instructions that, when executed by a computing device, cause the computing device to perform a face recognition method.

[0107] Based on the same technical solution, the present invention also discloses a computing device, including one or more processors, one or more memories, and one or more programs, wherein the one or more programs are stored in the one or more memories and configured to be executed by the one or more processors, and the one or more programs include instructions for performing a face recognition method.

[0108] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0109] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0110] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0111] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0112] The above are merely embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of the claims of the present invention pending approval.

Claims

1. A face recognition method, characterized in that, include: Acquire the face image to be identified from the image acquisition device, and extract the feature data of the face image to be identified; The first similarity between the feature data of the face image to be identified and the feature data of each registered face image is calculated respectively; wherein, the registered face image feature data are face image feature data pre-stored in the base database, and the registered face image feature data in the base database is periodically corrected according to a preset aging index; If all first similarities are less than the similarity threshold, obtain the N first similarities that are closest to the similarity threshold, and calculate the percentage difference between the N first similarities and the similarity threshold for each; where N is a preset number. If the M percentage differences are less than a preset percentage, calculate the cosine of the tilt angle threshold between the image acquisition device and the face to be identified, and use the cosine value to correct the registered face image feature data corresponding to the M percentage differences; where 0 <M≤N; Calculate the second similarity between the feature data of the face image to be identified and the feature data of the corrected registered face image; If any second similarity score is not less than the similarity threshold, the face image to be identified matches the corresponding corrected registered face image feature data, and the face recognition is successful.

2. The face recognition method according to claim 1, characterized in that, The pre-stored facial image feature data in the base database includes: Acquire registered facial images from image acquisition devices and extract feature data from the registered facial images; If the brightness value in the registered face image feature data is within the preset brightness range and the clarity is within the preset high-resolution range, then the registered face image feature data will be stored in the base database. If the brightness value in the registered face image feature data is within a preset brightness range and the clarity is within a preset medium clarity range, the contrast of the registered face image is enhanced based on histogram equalization until the clarity is within a preset high clarity range, and then the registered face image feature data is stored in the base database.

3. The face recognition method according to claim 1, characterized in that, The aging index includes age groups and age-matched facial features; the facial features include eyebrows, eyes, nose, lips, and the lower edge of the facial contour. The aging index is the percentage increase or decrease in the area of ​​the facial features in the following year compared to the previous year within the age group.

4. A face recognition system, characterized in that, include: The feature extraction module acquires the face image to be identified from the image acquisition device and extracts the feature data of the face image to be identified; The first similarity module calculates the first similarity between the feature data of the face image to be identified and the feature data of each registered face image; wherein, the registered face image feature data are face image feature data pre-stored in the base database, and the registered face image feature data in the base database are periodically corrected according to a preset aging index; In the filtering module, if all first similarities are less than the similarity threshold, the N first similarities closest to the similarity threshold are obtained, and the percentage difference between the N first similarities and the similarity threshold is calculated for each; where N is a preset number. The geometric correction module, if the M percentage differences are less than a preset percentage, calculates the cosine value of the tilt angle threshold between the image acquisition device and the face to be identified, and uses the cosine value to correct the registered face image feature data corresponding to the M percentage differences; where 0 <M≤N; The second similarity module calculates the second similarity between the feature data of the face image to be identified and the feature data of the corrected registered face image. If any second similarity is not less than the similarity threshold, the face image to be identified matches the corresponding corrected registered face image feature data, and the face recognition is successful.

5. A face recognition system according to claim 4, characterized in that, It also includes a pre-storage module for pre-storing facial image feature data into the base database. The pre-storage process includes: Acquire registered facial images from image acquisition devices and extract feature data from the registered facial images; If the brightness value in the registered face image feature data is within the preset brightness range and the clarity is within the preset high-resolution range, then the registered face image feature data will be stored in the base database. If the brightness value in the registered face image feature data is within a preset brightness range and the clarity is within a preset medium clarity range, the contrast of the registered face image is enhanced based on histogram equalization until the clarity is within a preset high clarity range, and then the registered face image feature data is stored in the base database.

6. A face recognition system according to claim 4, characterized in that, The aging index includes age groups and age-matched facial features; the facial features include eyebrows, eyes, nose, lips, and the lower edge of the facial contour. The aging index is the percentage increase or decrease in the area of ​​the facial features in the following year compared to the previous year within the age group.

7. A computer-readable storage medium for storing one or more programs, characterized in that, The one or more programs include instructions that, when executed by a computing device, cause the computing device to perform any of the methods according to claims 1 to 3.

8. A computing device, characterized in that, include: One or more processors, one or more memories, and one or more programs, wherein the one or more programs are stored in the one or more memories and configured to be executed by the one or more processors, the one or more programs including instructions for performing any of the methods according to claims 1 to 3.

Images