Authenticity verification method and system based on educational certificate picture
By obtaining the characteristics and content information of the academic certificate picture and using deep learning models for automated verification, the inefficiency problem in traditional methods is solved, and efficient and accurate certificate authenticity verification is achieved.
Patent Information
- Application Number
- CN202510446929.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-10
- Publication Date
- 2025-07-18
AI Technical Summary
The authenticity verification method of traditional academic qualification certificates is inefficient, and it is impossible to effectively determine the authenticity, and the accuracy cannot be guaranteed by relying on manual verification.
By obtaining the feature information and content information of the academic certificate picture, using deep learning models for classification, remake and tampering pre-checking, combined with multi-factor rule verification, the authenticity of the certificate is automatically determined.
It realizes efficient and accurate verification of authenticity and falsehood of academic certificate pictures, avoids manual verification errors, and improves discrimination efficiency and accuracy.
Smart Images
Figure CN120340046A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a method and system for verifying the authenticity of diploma pictures, belonging to the field of Internet technology. Background Art
[0002] With the rapid development and application of information technology, diploma serves as an important basis for evaluating an individual's academic level and ability. In the processes of recruitment, talent selection, social networking, etc., it is often necessary to upload scanned copies or photos of diplomas through Internet platforms or institutional online service systems and conduct authenticity verification. The traditional methods for verifying the authenticity of diplomas mainly rely on manual inspection and querying official databases. The method of querying official databases is relatively inefficient, requiring multiple steps of inspection, and the paper-based manual verification method cannot effectively distinguish authenticity. Therefore, it is very meaningful to develop a method and system for verifying the authenticity of diploma pictures. Summary of the Invention
[0003] The object of the present invention is to enable the authenticity verification of diploma pictures to be completed in a simpler, more efficient, and secure manner.
[0004] To achieve the above object, on the one hand, the present invention discloses a method for verifying the authenticity of diploma pictures, which is characterized by including the following steps:
[0005] Step 1: Obtain a complete target picture, and parse to obtain the feature information and content information of the target picture;
[0006] Step 2: Perform pre-verification based on the parsed feature information to obtain a pre-verification result of the feature information;
[0007] Perform rule verification based on the parsed content information to obtain a rule verification result;
[0008] Step 3: Determine the authenticity verification result of the target picture based on the pre-verification result of the feature information of the target picture and the rule verification result of the content information of the target picture.
[0009] Preferably, in Step 1, the target picture includes a complete and clear diploma.
[0010] Preferably, in Step 1, when the target subject in the target picture is detected, the feature information and the content information of the target picture are parsed.
[0011] Preferably, the target subject includes the feature information and the content information unique to the diploma. The feature information includes certificate format, institutional seal, font, and the content information includes the photo of the certificate holder, name, gender, date of birth, school name, graduation year, education level, and diploma code.
[0012] Preferably, in step 2, the pre-verification includes classification pre-verification, reshooting pre-verification, and tampering pre-verification:
[0013] In the classification pre-verification: Process the picture, input the classification model trained according to the format of the certificate picture, and determine whether the format of the target picture is correct;
[0014] In the reshooting pre-verification: By extracting 10 statistical features including the mean value of the R channel, the mean value of the G channel, the mean value of the B channel, the variance of the R channel, the variance of the G channel, the variance of the B channel, the gray variance, the gray mean, the 32-bit gradient histogram, and the 32-bit gray histogram in the feature area of the target picture, input the extracted statistical features into the deep learning model, and judge the input features according to the discriminant model based on 10 statistical features pre-trained to confirm whether the picture has been reshot;
[0015] In the tampering pre-verification: Construct a data set composed of tampering with key areas, and a binary classification model of tampering and non-tampering trained based on the ResNet18 model. By extracting the image in the feature area of the target picture and inputting it into the binary classification model, judge whether the picture has been tampered with.
[0016] Preferably, the rule verification in step 2 includes 8 items: student name verification, student gender verification, student date of birth verification, school name verification, graduation year verification, education level verification, education code verification, and portrait photo verification.
[0017] Preferably, in step 2, when performing pre-verification based on the parsed feature information, adjust the corresponding threshold and the weight ratio of the verification according to the maturity of the basic ability components;
[0018] When performing rule verification based on the parsed content information, all verifications need to pass. If one verification fails, the verification fails.
[0019] Another aspect of the present invention discloses a authenticity verification system based on a diploma picture for implementing the above authenticity verification method, which is characterized in that it includes:
[0020] Diploma feature information parsing module: When the target picture uploaded by the user is complete, it is used to parse the target picture uploaded by the user to obtain the feature information of the target picture;
[0021] Diploma content information parsing module: When the target picture uploaded by the user is complete, it is used to parse the target picture uploaded by the user to obtain the content information of the target picture;
[0022] Diploma pre-verification module: Used to perform three-layer pre-verification of classification, reshooting, and tampering on the feature information extracted from the target picture;
[0023] Academic certificate rule verification module: used to perform 8 - item rule verification on the content information extracted from the target picture;
[0024] Academic certificate authenticity verification result output module: based on the pre - verification result of the feature information of the target picture and the verification result of the content information of the target picture, determine the authenticity verification result of the target picture. Among them, the pre - verification based on the feature information of the target picture needs to adjust the corresponding threshold and the weight ratio of verification according to the maturity of the basic ability component. The rule verification based on the content information of the target picture needs to pass all verifications. If any one verification fails, the verification fails.
[0025] Preferably, when the verification is passed, output "verification passed"; if the verification fails, output "verification failed".
[0026] Compared with the existing technical solutions, the present invention has the following beneficial effects:
[0027] (1) It can determine whether the target certificate picture is a real and legal academic certificate;
[0028] (2) It is based on the self - feature information of the target picture. The self - feature information of the picture can be used to confirm that the target picture has not been tampered with and there is no obvious situation that does not conform to the style, ensuring the rationality of the picture;
[0029] (3) It is based on the content information contained in the target picture. The content information contained in the picture can be used to analyze the content of the picture itself. Through multi - factor cross - verification, the picture is judged to ensure the legality of the picture;
[0030] (4) The present invention obtains the target picture, and when the target subject is detected in the target picture, it identifies the feature information in the target subject or extracts the content information in the target subject. Based on the feature information or the content information in the target subject, it determines the authenticity verification result of the target picture; the target subject needs to include the unique features of the academic certificate, including but not limited to information such as certificate format, institutional seal, and font. In addition, it also needs to include the unique content information of the academic certificate, including but not limited to information such as school, major, and graduation date; by using the feature information and content information of the target subject to determine the authenticity of the target subject, there is no need for manual verification to identify authenticity, which can improve the efficiency of authenticity identification; at the same time, through the authenticity identification method disclosed in the present invention, the situation of easy errors when identifying authenticity through manual verification can be avoided, and the identification accuracy can be improved. Brief Description of the Drawings
[0031] Figure 1 It is a schematic flowchart of the authenticity verification method for academic certificate pictures provided by one or more embodiments of this specification;
[0032] Figure 2 Schematic diagram of the module composition of the authenticity verification system for academic certificate pictures provided by one or more embodiments of this specification. Specific implementation manners
[0033] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and not to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms also fall within the scope defined by the appended claims of this application.
[0034] Reference Figure 1 , one aspect of the embodiment of the present invention discloses an authenticity verification method for academic certificate pictures, including the following steps:
[0035] Step 1, obtain a target picture, where the target picture needs to contain a complete academic certificate.
[0036] Step 2, obtain the feature information of the target picture, where the obtained feature information includes but is not limited to information such as certificate format, institutional seal, and font. The target picture is pre-verified in three layers: classification, reshooting, and tampering through the extracted feature information:
[0037] The first layer, process the picture, input the classification model trained according to the certificate picture format, and determine whether the target picture format is correct;
[0038] The second layer, extract 10 statistical features of the feature area of the target picture, including the mean value of the R channel, the mean value of the G channel, the mean value of the B channel, the variance of the R channel, the variance of the G channel, the variance of the B channel, the gray variance, the gray mean value, the 32-bit gradient histogram, and the 32-bit gray histogram. Input the extracted statistical features into the deep learning model, and judge the input features according to the discriminant model based on 10 statistical features pre-trained to confirm whether the picture has been reshot.
[0039] The third layer, construct a data set composed of tampering with key areas, and a two-classification model for tampering and non-tampering trained based on the ResNet18 model (a deep convolutional neural network). Extract the image of the feature area of the target picture and input it into the two-classification model to judge whether the picture has been tampered with.
[0040] Step 3: Obtain the content information of the target image, where the obtained academic certificate content information includes but is not limited to the certificate holder's photo, name, gender, date of birth, school name, graduation year, education level, graduation certificate code, etc. The extracted content information is checked for 8 rules, based on the official restrictions on certificate specifications and multi-factor comparison of the extracted content with the user's real personal information, to determine whether the image content information is legal.
[0041] Step 4: Determine the authenticity verification result of the target image based on the pre-verification result of the feature information of the target image and the verification result of the content information of the target image.
[0042] As a feature example, classification pre-verification is performed by base64 encoding the target image, decoding the system to obtain an RGB array, scaling the target image to 224*224 size according to the RGB array, and performing classification pre-verification on the target image feature information through a classification model trained based on a sample of academic certificate and the ResNet model (a deep convolutional neural network).
[0043] As a feature example, the pre-check of the copy is carried out by base64 encoding the target image. After the system decodes it, 10 feature data are extracted, including R channel mean, G channel mean, B channel mean, R channel variance, G channel variance, B channel variance, grayscale variance, grayscale mean, 32-bit gradient histogram, and 32-bit grayscale histogram, a total of 72 dimensions of features. The feature information of the target image is pre-checked through the LBP features extracted based on the open source moiré dataset and the classification model trained by the ResNet101 model (a deep convolutional neural network).
[0044] As a feature example, tampering pre-verification is achieved by constructing a pre-trained tampering and non-tampering binary classification model, where the binary classification model is based on the ResNet18 model (a deep convolutional neural network), and the dataset consists of normal academic certificate images and tampered images of normal images; the input target image is pre-verified for tampering through the constructed tampering and non-tampering binary classification model.
[0045] As an example of a feature, the content information of the target image is obtained. The content information can be completed by a manufacturer that specializes in providing universal text recognition capabilities for academic certificate images, and the content information includes but is not limited to the certificate holder's photo, name, gender, date of birth, school name, graduation year, education level, graduation certificate code, and other information.
[0046] As a feature example, rule verification is performed on the content information of the obtained target image. The rule verification includes two categories of a total of 8 rule verifications. The two categories are: verification based on the certificate specification restrictions announced by the official, and multi-factor comparison verification based on the extracted content and the user's personal real information. The total of 8 rule verifications include: student name verification, student gender verification, student date of birth verification, school name verification, graduation year verification, education level verification, education code verification, and portrait photo verification.
[0047] As a feature example, the verification based on the certificate specification restrictions announced by the official includes graduation year verification, school name verification, education level verification, and education code verification:
[0048] The graduation year verification is to compare whether the 7th to 10th digits of the certificate number are consistent with the graduation year on the certificate;
[0049] The school name verification is to compare whether the first 5 digits of the certificate number are consistent with the school name on the certificate, and the corresponding relationship between the code and the name can be verified through the list information of higher education institutions announced by the Ministry of Education;
[0050] The education level verification is to compare whether the 11th to 12th digits of the certificate number are consistent with the education level on the certificate, and the corresponding relationship between the number and the education level is shown in Table 1;
[0051] The education code verification is based on the publicly announced coding rules. The first 5 digits are the national standard codes of schools or other educational institutions; the 6th digit is the school-running type code; the 7th to 10th digits are the year; the 11th to 12th digits are the training level codes; the 13th to 17th digits are the serial numbers arranged by the school for graduation (completion) certificates, and the rules are
[0052] Table 1:
[0053]
[0054]
[0055] As a feature example, the verification based on multi-factor comparison of the extracted content and the user's personal real information includes student name verification, student gender verification, student date of birth verification, and portrait photo verification:
[0056] The student name verification is to compare whether the student name on the certificate is consistent with the name of the person who uploaded the certificate photo;
[0057] The student gender verification is to compare whether the student gender on the certificate is consistent with the gender of the person who uploaded the certificate photo;
[0058] The student date of birth verification is to compare whether the student date of birth on the certificate is consistent with the date of birth of the person who uploaded the certificate photo;
[0059] The verification of the portrait photo is to compare whether the portrait on the certificate is consistent with the portrait photo of the uploader.
[0060] As an example of features, the authenticity verification result is based on the pre-verification result of the feature information of the target picture and the verification result of the content information of the target picture. The pre-verification based on the feature information of the target picture needs to adjust the corresponding threshold and the weight ratio of the verification according to the maturity of the basic ability components. The rule verification based on the content information of the target picture needs to pass all verifications. If any one verification fails, the verification fails.
[0061] Reference Figure 2 , Another aspect of the embodiment of the present invention is to provide a system for verifying the authenticity of an academic certificate picture, including a feature information parsing module, a content information parsing module, a pre-verification module, a rule verification model module, and an output module.
[0062] Academic certificate feature information parsing module: When the target picture uploaded by the user is complete, it is used to parse the target picture uploaded by the user to obtain the feature information of the target picture, including but not limited to picture format, institutional seal, font and other information;
[0063] Academic certificate content information parsing module: When the target picture uploaded by the user is complete, it is used to parse the target picture uploaded by the user to obtain the content information of the target picture, including but not limited to the photo of the certificate holder, name, gender, date of birth, school name, graduation year, education level, graduation certificate code and other information;
[0064] Academic certificate pre-verification module: It is used to perform three-layer pre-verification of classification, reshooting, and tampering on the feature information extracted from the target picture;
[0065] Academic certificate rule verification module: It is used to perform 8-item rule verification on the content information extracted from the target picture;
[0066] Academic certificate authenticity verification result output module: Based on the pre-verification result of the feature information of the target picture and the verification result of the content information of the target picture, determine the authenticity verification result of the target picture. Among them, the pre-verification based on the feature information of the target picture needs to adjust the corresponding threshold and the weight ratio of the verification according to the maturity of the basic ability components. The rule verification based on the content information of the target picture needs to pass all verifications. If any one verification fails, the verification fails.
[0067] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the drawings are only examples and do not limit the present invention. The advantages of the present invention have been fully and effectively realized. The function and structural principle of the present invention have been shown and described in the embodiments. Without departing from the above principle, the embodiments of the present invention can have any deformation or modification.
Claims
1. A method for verifying the authenticity of academic certificate pictures, characterized in that, It includes the following steps: Step 1: Obtain a complete target picture, and parse to obtain the feature information and content information of the target picture; Step 2: Perform pre-verification based on the parsed feature information to obtain the pre-verification result of the feature information; Perform rule verification based on the parsed content information to obtain the rule verification result; Step 3: Determine the authenticity verification result of the target picture based on the pre-verification result of the feature information of the target picture and the rule verification result of the content information of the target picture.
2. The method for verifying the authenticity of an academic certificate picture according to claim 1, wherein In Step 1, the target picture contains a complete and clear academic certificate.
3. The method for verifying the authenticity of an academic certificate picture according to claim 2, wherein, In Step 1, when the target subject in the target picture is detected, the feature information and the content information of the target picture are parsed.
4. The method for verifying the authenticity of an academic certificate picture according to claim 3, characterized in that, The target subject includes the specific feature information and the content information of the academic certificate. The feature information includes the certificate layout, institutional seal, and font. The content information includes the photo of the certificate holder, name, gender, date of birth, school name, graduation year, education level, and graduation certificate code.
5. The method for verifying the authenticity of a diploma certificate picture according to claim 1, characterized in that, In Step 2, the pre-verification includes classification pre-verification, reshooting pre-verification, and tampering pre-verification: In the classification pre-verification: Process the picture, input the classification model trained according to the certificate picture layout, and determine whether the target picture layout is correct; In the reshooting pre-verification: Extract 10 statistical features including the mean value of the R channel, the mean value of the G channel, the mean value of the B channel, the variance of the R channel, the variance of the G channel, the variance of the B channel, the gray variance, the gray mean, the 32-bit gradient histogram, and the 32-bit gray histogram in the feature area of the target picture. Input the extracted statistical features into the deep learning model, and judge the input features according to the pre-trained discriminant model based on the 10 statistical features to confirm whether the picture has been reshot; In the tampering pre-verification: Construct a data set composed of tampering with key areas, and a binary classification model of tampering and non-tampering trained based on the ResNet18 model. Extract the image in the feature area of the target picture and input it into the binary classification model to judge whether the picture has been tampered with.
6. The method for verifying the authenticity of an academic certificate picture according to claim 1, wherein The rule verification in Step 2 includes 8 items: student name verification, student gender verification, student date of birth verification, school name verification, graduation year verification, education level verification, education code verification, and portrait photo verification.
7. The method for verifying the authenticity of an academic certificate picture according to claim 1, wherein, In Step 2, when performing pre-verification based on the parsed feature information, adjust the corresponding threshold and the weight ratio of the verification according to the maturity of the basic ability components; When performing rule verification based on the parsed content information, all verifications need to pass. If one verification fails, the verification fails.
8. A system for verifying the authenticity of academic certificate pictures, which is used to implement the authenticity verification method described in claim 1, is characterized in that, It includes: Academic certificate feature information parsing module: When the target picture uploaded by the user is complete, it is used to parse the target picture uploaded by the user to obtain the feature information of the target picture; Academic certificate content information parsing module: When the target picture uploaded by the user is complete, it is used to parse the target picture uploaded by the user to obtain the content information of the target picture; Academic certificate pre-verification module: It is used to perform three-layer pre-verification of classification, reshooting, and tampering on the feature information extracted from the target picture; Academic certificate rule verification module: used to perform 8-rule verification on the content information extracted from the target picture; Genuine or fake verification result output module for academic certificates: based on the pre-verification result of the feature information of the target picture and the verification result of the content information of the target picture, determine the genuine or fake verification result of the target picture. Among them, the pre-verification based on the feature information of the target picture needs to adjust the corresponding threshold and the weight ratio of verification according to the maturity of the basic ability components. The rule verification based on the content information of the target picture needs to pass all verifications. If one verification fails, the verification fails.
9. The authenticity verification system for academic certificate pictures according to claim 8, characterized in that, When the verification is passed, output "verification passed"; if the verification fails, output "verification failed".