Search

CN-122001661-A - Competition account retrieval system and method based on contest record verification and multi-mode face verification

CN122001661ACN 122001661 ACN122001661 ACN 122001661ACN-122001661-A

Abstract

The invention relates to the technical field of internet account safety verification, in particular to a competition account recovery system and method based on competition record verification and multimode face verification, wherein the method realizes safe and efficient recovery of competition accounts through primary mobile phone number verification and risk initial judgment, personalized match of competition records, accurate verification of real name information, multimode face verification, new mobile phone number binding and log verification five-order progressive process, fusion of a multidimensional risk scoring model, a competition record matching degree calculation model and a multimode face feature fusion model equivalent model; meanwhile, a three-dimensional index mechanism of an account number-mobile phone number-competition ID is constructed, and the defects that the traditional account number recovery method depends on a single verification dimension, is weak in anti-attack capability and low in flow fault tolerance rate are overcome by combining technologies of quadruple cross comparison, dynamic threshold verification, full-link data encryption and the like. The quantitative verification and risk self-adaptive adjustment are realized through the mathematical model, the anti-attack recognition rate of living body detection reaches more than 99.5%, the flow passing rate is improved to more than 95%, the method is not only suitable for competition platforms, but also can be expanded to account number recovery scenes needing real-name verification such as electronic commerce, education and finance, and has the remarkable advantages of high safety, wide applicability and convenience in operation.

Inventors

  • REN LIFENG
  • SONG RUIFANG
  • LI FANGFANG
  • WANG ZHENG
  • WANG SHENGLI
  • ZHI DA

Assignees

  • 环球赛乐(北京)科技有限公司

Dates

Publication Date
20260508
Application Date
20260224

Claims (9)

  1. 1. A competition account retrieval system and method based on competition record verification and multi-mode face verification is characterized by comprising the following steps: Step 1, original mobile phone number verification and risk preliminary judgment, wherein a user inputs an original binding mobile phone number, front-end execution format and number segment verification is carried out, a background queries a registration state and invokes a multidimensional risk scoring model to calculate a user risk grade R, and the formula is as follows: Wherein, the 、 、 Is a weight coefficient, satisfies Default value =0.4、 =0.3、 =0.3, D is the device matching degree, I is the IP address matching degree, B is the operation behavior matching degree, if Directly enter the next link if Triggering the secondary verification of the verification code of the equipment, and calculating the input error rate by the front end in real time Wherein For the cumulative number of input errors for the current form, For the total number of input fields of the current form, when Popup step-by-step filling guide popup window; Step 2, personalized matching of the contest records, namely extracting the contest records of the user by the background based on the three-dimensional index of account number-mobile phone number-contest ID, sequencing according to the priority of the contest time interval within 1 year, the contest depth and the user memory frequency, displaying 3 high-priority records at the front end, supporting the screening or random replacement of the user according to the year and the contest type, answering 1-3 detailed questions after the user selects the contest, and calling a contest record matching degree calculation model to calculate the matching degree M by the background, wherein the formula is as follows: Wherein, the 、 、 Is a weight coefficient, satisfies + + =1; For the racing subject match score, n is the number of detail questions, For the answer score of the i-th question, As a time-decay factor, T is the time length (unit: year) of the racing hour interval, lambda is the attenuation coefficient, when Judging that the matching is passed; And 3, accurately checking real-name information, namely inputting a name and an identity card number by a user, and executing identity card check code compliance verification at the front end, wherein the formula is as follows: Wherein, the Is the digital value of the ith bit of the identification card number, The background completes the double comparison of name, ID card number and competition registration data and public security authority data; Step 4, multi-mode face nuclear body, namely after a user authorizes to call the camera, the background judges living bodies through a living body detection anti-attack discrimination function, and the formula is as follows: Wherein sigma (-) is a Sigmoid function, ; For the dynamic feature score to be a score, For the 3D depth feature score, Scoring texture details; 、 、 B is a bias term, when L is more than or equal to 0.7, judging that the living body is real, and then calling a multi-mode face feature fusion model to extract a fusion feature vector, wherein the formula is as follows: Wherein, the 、 、 For the fusion weight, satisfy The fusion weight is adaptively adjusted according to the acquisition environment: is a 2D face texture feature vector, Is a 3D face structure feature vector, Is an optical flow dynamic characteristic vector. Calculating the matching degree through a cosine similarity formula: Meanwhile, the risk level R is combined, and the face matching threshold value is dynamically adjusted When (when) Judging that the human face passes through the nucleus; And 5, binding the new mobile phone number and storing the log, namely inputting the new mobile phone number by a user, finishing verification of the verification code, updating the account binding relation in the background, and storing the operation log by adopting a blockchain technology, wherein the log comprises R, M, sim and other core parameters.
  2. 2. The competition account retrieval system and method based on competition record verification and multi-mode face verification according to claim 1, wherein the operation behavior matching degree B in step 1 is obtained by calculating the edit distance normalization of the current operation sequence and the historical behavior sequence, and the value range is In the equipment matching degree D, the common equipment D=1 and the strange equipment D=0.2, in the IP address matching degree I, the common IP section I=1 and the remote IP I=0.3, and the weight coefficient defaults to take values =0.4、 =0.3、 =0.3。
  3. 3. The system and method for recovering a competition account based on a competition record verification and a multi-modal face verification as set forth in claim 1, wherein the competition subject matching score S in step 2 is s=1 when the matching is accurate, s=0.6 when the matching is fuzzy, s=0 when the matching is not accurate, and the i-th question answer score Answer pair =1, Error-answering Weight coefficient default value =0.5、 =0.4、 =0.1, And the attenuation coefficient defaults to λ=0.3.
  4. 4. The system and method for recovering racing account based on race record verification and multi-modal face verification as set forth in claim 1, wherein the step 4 is characterized by fusing weights 、 、 Self-adaptive adjustment according to acquisition environment, under normal illumination =0.4、 =0.3、 =0.3, Dim light/blocked scene =0.2、 =0.5、 =0.3, Feature weight default value =0.4、 =0.35、 Bias term default b=0.5, dynamic threshold value =0.25 The calculation formula of (2) is Wherein =0.95 As the base threshold value, =0.03 Is the threshold adjustment amplitude.
  5. 5. The competition account retrieval system and method based on competition record verification and multi-mode face verification according to claim 1, wherein the background in the step 3 completes quadruple cross comparison, including comparison with competition registration data, comparison with public security authority identity information, comparison with equipment image and comparison with behavior characteristics, if different words of name homophones exist, a secondary confirmation prompt is popped up, and the next link is entered after confirmation.
  6. 6. The method for recovering a racing account based on a race record verification and a multi-modal face verification as set forth in claim 1, wherein in step 4 Extracted through the ResNet network of ResNet, By means of 3D structured light scanning extraction, Extracting by Farneback optical flow algorithm, and allowing the user to retry for less than or equal to 2 times, and suspending the flow for 24 hours if the retry fails.
  7. 7. The method for retrieving the competition account number based on the competition record verification and the multi-mode face verification according to claim 1, wherein in the step 5, the validity period of the verification code of the new mobile phone number is 60 seconds, retransmission is supported, if the new mobile phone number is bound with other accounts, a confirmation prompt is popped up, and the user releases the original binding and completes the new binding after confirmation.
  8. 8. A competition account recovery system and method based on competition record verification and multi-mode face verification as claimed in any one of claims 1-7, comprising four modules for realizing two-way communication through a real-time data synchronization bus: The front-end interaction module is used for receiving user input, executing format/check code/input error rate calculation, realizing environment self-adaptive face acquisition, process progress visualization, cross-link lossless backtracking and displaying verification guide and results; the background data processing module is used for storing user account numbers, competition records and real-name information, extracting data based on three-dimensional indexes of account numbers, mobile phone numbers and competition IDs, managing flow states, realizing breakpoint recovery and archiving operation logs; The identity verification module is used for calling a multidimensional risk scoring model, a contest record matching degree calculation model and a multi-mode face feature fusion model equalization model to complete full-flow identity verification and risk judgment; And the data security module is used for transmitting sensitive data by SSL/TLS1.3+SM4 double encryption, and storing the face feature vector and the desensitization identity card number by AES-256 encryption so as to realize real-time monitoring of behavior risk and operation log blockchain storage.
  9. 9. The competition account retrieval system and method based on the competition record verification and the multi-mode face verification as claimed in any one of claims 8, wherein the front-end interaction module provides 7 core function pages to support cross-device adaptation, and the background data processing module achieves a data preloading function and improves data extraction efficiency.

Description

Competition account retrieval system and method based on contest record verification and multi-mode face verification Technical Field The invention relates to the technical field of internet account safety verification, in particular to a competition account recovery system and method based on competition record verification and multi-mode face verification. Background With popularization of the internet competition platform, the problem that the user cannot log in the account due to reasons such as mobile phone number replacement, original mobile phone number logout, fuzzy account memory and the like is increasingly highlighted, and competition account retrieval becomes a high-frequency just-needed scene. The competition account number is usually related to real-name competition information and competition history data of a user, and has extremely high requirements on safety, accuracy and fault tolerance of the recovery process. The existing competition account recovery method mainly has the following defects: 1) The verification is single, the coverage scene is limited, the traditional method is used for excessively binding the original mobile phone number, when a user cannot receive verification information due to the reasons of number changing, mobile phone number cancellation, operator recovery number and the like, the account is always in permanent disconnection, the account cannot be recovered up to 63% due to the situation of statistics in the industry, and part of platform supplementary mailbox verification is carried out, but the mailbox also has the problems of long-term unused, password forgetting and the like, and the actual recovery success rate is less than 30%. 2) The verification dimension is single, and the risk of identity exploitation is easily generated by verifying the general identity information such as an identity card number, a mailbox and the like without combining the exclusive competition record in the user platform; 3) The human face nuclear body has weak anti-attack capability and poor user experience, namely the acquisition interface of the human face nuclear body link is unreasonable in design, the living body detection only depends on single 2D image characteristics, the anti-attack capability of photos, videos and 3D models is weak, and the recognition failure rate is high; 4) The method has the advantages of no quantitative model support, strong decision subjectivity, low accuracy, high misjudgment rate, low flow fault tolerance, no accurate guidance for input errors, repeated data filling in midway withdrawal, and poor user operation experience, and the verification result depends on manual judgment or simple comparison; 5) The data security compliance is insufficient, namely the sensitive data transmission and storage lack of full-link encryption, the risk of information leakage exists, and the compliance requirement of the personal information protection law, the network security law is difficult to meet. At present, a competition account recovery system and a competition account recovery method which are integrated with personalized verification of a competition record and a multi-mode face verification and realize quantitative risk judgment and accurate matching verification through a mathematical model are not yet developed, so that the research and development of the competition account recovery system and the method which are high in safety, wide in applicability, convenient to operate and data compliance have important practical application values. Therefore, aiming at the situation, the invention provides a personalized verification and multi-mode face verification system and method for fusing the contest records. Disclosure of Invention In order to achieve the above purpose, the invention provides a competition account retrieval method based on competition record verification and multi-mode face verification, which comprises the following steps: 1. original mobile phone number checking and risk initial judgment Firstly, a user inputs an original account number to bind a mobile phone number at the front end, the front end firstly executes 11-bit digital format verification and number segment validity pre-verification, and after the verification is passed, mobile phone number information is sent to the background; Secondly, after the front end passes verification, the background rapidly inquires the platform registration state of the mobile phone number through the three-dimensional index of the account number-mobile phone number-competition ID, if registration records are not inquired, the mobile phone number unbound platform account number is popped up to prompt and terminate the flow; Thirdly, a multi-dimensional risk scoring model is called by the background to calculate a user risk grade R, and a calculation formula is as follows: Wherein, the 、、Is a weight coefficient, satisfiesDefault value=0.4、=0.3、D is the matching degree of the equipment, D is 1 of com