CN116543181A - Anti-partner fraud method and system based on image background feature recognition - Google Patents
Anti-partner fraud method and system based on image background feature recognition Download PDFInfo
- Publication number
- CN116543181A CN116543181A CN202310521449.1A CN202310521449A CN116543181A CN 116543181 A CN116543181 A CN 116543181A CN 202310521449 A CN202310521449 A CN 202310521449A CN 116543181 A CN116543181 A CN 116543181A
- Authority
- CN
- China
- Prior art keywords
- image
- partner
- registered user
- fraudulent
- fraud
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 26
- 239000013598 vector Substances 0.000 claims description 19
- 238000012216 screening Methods 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 3
- 238000004590 computer program Methods 0.000 claims description 2
- 238000011835 investigation Methods 0.000 claims description 2
- 238000013527 convolutional neural network Methods 0.000 description 7
- 238000013528 artificial neural network Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 238000013459 approach Methods 0.000 description 4
- 238000013136 deep learning model Methods 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 238000013475 authorization Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005065 mining Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000000149 penetrating effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/761—Proximity, similarity or dissimilarity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/10—Image acquisition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Business, Economics & Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Medical Informatics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Finance (AREA)
- General Business, Economics & Management (AREA)
- Technology Law (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Accounting & Taxation (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the invention provides an anti-group partner fraud method based on image background feature identification, which comprises the following steps: constructing a fraud group image library; collecting a new registered user image; and judging whether the newly registered user belongs to the fraudulent party or not through the comparison between the image of the newly registered user and each image of the fraudulent party in the image library of the fraudulent party. According to the technical scheme, the image background information is mined to extract the feature points and the feature description, and the similarity of the image of the new registered user and each image of the fraudulent party can be effectively identified in a mode of matching the feature points of the image background library of the predetermined fraudulent party, so that whether the new registered user belongs to the fraudulent party or not is identified.
Description
Technical Field
The invention relates to the technical field of network security, in particular to an anti-group partner fraud method and system based on image background feature recognition.
Background
The rapid development of financial technology is penetrating all aspects of society and bringing great convenience to our lives. However, fraud risk is also accompanied. Moreover, the fraud means is also continuously evolved and upgraded, and the characteristics of concealment, partner, intellectualization and the like are reflected. How to accurately identify fraud by using AI technology is a great difficulty facing various financial institutions.
The traditional anti-fraud technology mainly comprises the following two types: depending on expert experience. By analyzing a large number of fraud cases, the expert aggregates a series of fraud means and patterns and runs online in a regular and strategic manner. This approach is labor intensive and has hysteresis. And after the fraud mode is upgraded, time is required to accumulate fraud cases, and the latest anti-fraud strategy cannot be iterated quickly. Second, a traditional machine learning model. The method mainly comprises the step of constructing a classification model through personal attribute characteristics, behavior characteristics and the like of users to identify fraudulent users. This approach suffers from the same dilemma as expert experience: the model effects against the fraudulent means of upgrades may be reduced or even disabled. Moreover, for new users of the platform, more accuracy problems may occur due to missing features.
For this reason, the prior art also presents a better solution, namely employing deep learning models to identify fraud. The current deep learning model mainly performs verification on a portrait acquired image and a certificate image. With the maturity of face recognition technology, the image verification accuracy can generally reach the industry.
In the process of implementing the present invention, the inventor finds that at least the following problems exist in the prior art:
the method adopting the deep learning model is only effective for the fraud mode of single person crimes, but can not identify the problem of the group fraud in the group registration mode. Therefore, how to accurately and effectively identify the partner fraud is a problem to be solved.
Disclosure of Invention
The embodiment of the invention provides an anti-partner fraud method and system based on image background feature identification, which are used for solving the problem that partner fraud is difficult to identify in the prior art.
In order to achieve the above object, in one aspect, an embodiment of the present invention provides an anti-group fraud method based on image background feature identification, including: constructing a fraud group image library; collecting a new registered user image; and judging whether the new registered user belongs to the fraudulent party or not through the comparison between the image of the new registered user and each fraudulent party image in the fraudulent party image library.
On the other hand, the embodiment of the invention provides an anti-partner fraud system based on image background feature identification, which comprises the following steps: the image library construction module is used for constructing a fraud group image library; the image acquisition module is used for acquiring images of the new registered users; and the comparison judging module is used for judging whether the new registered user belongs to the fraudulent party or not through the comparison between the image of the new registered user and each fraudulent party image in the fraudulent party image library.
Meanwhile, the embodiment of the invention also provides a computer readable storage medium, on which a computer program is stored, which when being executed by a processor, realizes the anti-group fraud method based on image background feature identification.
In addition, an embodiment of the present invention further provides a computer device, which includes: one or more processors; a storage means for storing one or more programs; the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the anti-group fraud method described above based on image background feature identification.
The technical scheme has the following beneficial effects:
according to the technical scheme, the characteristic points and the characteristic description are extracted from the image background part by mining the image background information, and the similarity of the images of the new registered user and the images of the fraudulent party can be effectively identified by matching the characteristic points of the extracted image background area with the characteristic points in the image background library of the determined fraudulent party, so that whether the new registered user is a normal user or not can be identified.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of an anti-partner fraud method based on image background feature recognition in an embodiment of the invention;
FIG. 2 is a block diagram of an anti-partner fraud system based on image background feature recognition in accordance with an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
As shown in fig. 1, an embodiment of the present invention provides an anti-group fraud method based on image background feature identification, including:
s101, constructing a fraud group image library;
s102, acquiring a new registered user image;
s103, judging whether the new registered user belongs to the fraudulent party or not through comparison between the new registered user image and each fraudulent party image in the fraudulent party image library.
Currently, images acquired by a user registering with a financial platform are mainly used for authentication. This approach plays an important role in protecting against fraud by other people, but there is also a certain limitation to this approach: i.e. the case of partner registration cannot be judged. This is because the face verification link focuses more on the foreground (i.e., face) portion of the image and filters the background portion of the image. However, the scene represented by the image background is just one of the available tools for mining fraudulent parties (or intermediary parties), because the fraudulent parties are usually centrally registered, and through analysis by the inventor, a major feature of the centralized registration is the centralization of the scene, that is, the background parts of the images of the members in the fraudulent parties are similar or even identical when registering. Therefore, when the problem that the partner fraud cannot be judged is solved, the idea of solving the problem is how to further mine background information in the image, and further predict whether the current user belongs to a member of the fraudulent partner by using the similarity in the background image.
Therefore, in the technical scheme, when a new user is registered, firstly, an image of the new registered user is acquired, after personal authorization of the new user is obtained, the image is uploaded to perform background area image feature recognition, namely, image features (including feature points and corresponding feature vectors) of a background area of the image are acquired, then the image features of the image are compared with image features of all images of a predetermined fraudulent party, so that the similarity with the fraudulent party is found, and when the similarity meets a preset requirement, the user is suspected to belong to a member of the predetermined fraudulent party. The new registered user image is a single image acquired during user registration, and after the image is compared with images in the fraudulent image library for identification, the identity of the user can be judged, and the image is not acquired again for comparison during subsequent login. The adoption of the technical scheme also solves the problem that the background part information is ignored and part information is lost commonly existing in the prior art. In the comparison process, the technical scheme is not a mode of simply comparing the current user with all acquired user images one by one, because different users possibly have common backgrounds, as the number of users increases, the backgrounds among the users show cross similarity to influence the accuracy of a judgment result, the running efficiency is very low, and the application is difficult to realize; the technical scheme is limited between the current new user and the predetermined fraud partner, has strong pertinence and high accuracy, greatly improves the working efficiency and has very high practicability.
Further, the step S103 includes:
s1031, calculating a similarity value between a background area of the new registered user image and a background area of each rogue partner image in the rogue partner image library;
s1032, selecting a maximum similarity value from the similarity values;
s1033, if the maximum similarity value is larger than a preset threshold value, judging that the new registered user belongs to the fraud group.
Since there are multiple images in the rogue group, the current user image needs to be compared with the multiple images respectively, so as to obtain a set of similarity values. If the maximum value in the similarity set is greater than the set threshold, the user is considered to be a suspected fraud partner member, and the location of the user is the same as the location of the image corresponding to the maximum similarity value (or called an intermediary point, namely a certain offline point of the fraud partner); otherwise, the user is considered to be a normal user.
Further, the step S1031 specifically includes:
for each rogue partner image in the rogue partner image library, the following steps are sequentially executed:
s10311, carrying out background area image feature recognition on the current fraud partner image to obtain background area image features of the current fraud partner image, wherein the image features comprise feature points and feature vectors corresponding to the feature points;
s10312, carrying out background area image feature recognition on the new registered user image to obtain new registered user background area image features;
s10313, calculating a similarity value between the background area image characteristics of the new registered user and the background area image characteristics of the current fraud partner image.
As described above, in the present solution, the entire image is not compared, but the background area of the new user image is compared with the background area of the fraudulent party image. Only the image background areas with rich features are focused, common public backgrounds can be filtered, interference is avoided, and meanwhile, the working efficiency can be improved.
Further, the specific steps of performing background area image feature recognition are as follows:
splitting a current image into a foreground region and a background region; disposing a mask over the foreground region; identifying all image features of the current image; and determining the background area image characteristics of the current image according to all the image characteristics of the current image and the mask.
An image can be seen as being made up of two parts: foreground and background. The current image can be split into a foreground region and a background region through image matting, namely the technology of separating the foreground region and the background region of the image through controlling transparency through an alpha channel. In the field of image matting techniques, many very successful algorithms have been developed: traditional algorithms such as bayesian matting and more commonly used deep learning methods now. At present, the deep learning-based matting algorithm also achieves good effect in practical engineering application.
The image features include feature points and corresponding feature vectors. Feature points refer to places in an image or in the field of vision that are clearly distinguished from their surrounding areas. These feature points should be unique to ensure that they can be better distinguished from other areas. Meanwhile, the characteristic points are also provided with rotation invariance and scale invariance to ensure the stability of subsequent application. The matching task cannot be performed well by using only the position information of the feature points, and thus more detailed information, i.e., feature description of the feature points, is required. The feature description is usually a vector, and is used for describing the features of the neighborhood around the feature points, so as to facilitate subsequent operations such as feature point matching.
In the technical scheme, two convolutional neural networks can be adopted to acquire the image characteristics of the image background area: firstly, inputting an image into a convolutional neural network A, and carrying out matting by applying a first algorithm corresponding to the neural network A in the system to respectively obtain a foreground region (namely a face region) and a background region. Then mask the face area to generate a mask, wherein the mask mainly marks the position information of the background area for later steps. And then, respectively inputting the images in the predetermined fraud image library into a convolutional neural network B, and extracting feature points of each fraud image and feature descriptions corresponding to the feature points by a second algorithm corresponding to the neural network B in an application system, wherein the feature points refer to points with identifiability and distinguishability, and the corresponding feature descriptions are high-dimensional feature vectors corresponding to the feature points. And then, combining the determined characteristic points and the output mask, and screening out the characteristic points and the characteristic description (namely the image characteristics of the background area of the rogue group image) of each rogue group image in the background area.
The above steps may also be employed for acquisition of background area image features of a newly registered user.
Further, the step S10313 specifically includes:
s103131, determining common feature points of the background area of the new registered user and the background area of the current fraud partner image by matching all feature points of the background area of the new registered user with all feature points of the background area of the current fraud partner image;
s103132, determining a common feature vector corresponding to the common feature point;
s103133, calculating a similarity value between the background area image features of the newly registered user and the background area image features of the current fraud group image according to the number of the common feature points and the common feature vector.
Feature point matching is a task of establishing a corresponding relationship between two images through feature points. Once the feature points and the feature descriptions (feature vectors) thereof in the images are extracted, the matching relation of the feature points can be obtained through the feature points and the feature descriptions thereof in the two images, the common feature points of the two images are determined, and whether the two images are derived from the same scene is further known.
In the process, feature points and feature vectors of two images to be compared are input into a convolutional neural network C together, and the matching quantity of the feature points in the background areas of the two images can be obtained by applying a third algorithm corresponding to the neural network C in the system.
Further, the step S101 specifically includes:
s1011, determining suspected fraud partners through preliminary screening;
s1012, determining a fraudulent party from the suspected fraudulent party according to the manual investigation result;
s1013, invoking an image of a partner member belonging to the fraudulent partner as a fraudulent partner image;
s1014, constructing a rogue partner image library through all the rogue partner images.
In the technical scheme, the fraudulent group partner is screened initially through the modes of geographic information, IP information, equipment information and the like, and then the fraudulent group partner is confirmed by combining manual proposal adjustment, so that the accuracy is ensured. In the subsequent operation, the predetermined image library of fraudulent parties is used as a benchmark for determining whether the new user is a member of the fraudulent party.
As shown in fig. 2, the embodiment of the present invention further provides an anti-group fraud system based on image background feature identification, including:
an image library construction module 21 for constructing a fraudulent party image library;
an image acquisition module 22 for acquiring an image of a newly registered user;
a comparison and judgment module 23, configured to judge whether the new registered user belongs to the rogue group through comparison between the new registered user image and each rogue group image in the rogue group image library.
Further, the comparison and judgment module 23 is specifically configured to: calculating a similarity value between the background area of the new registered user image and the background area of each fraudulent partner image in the fraudulent partner image library; selecting a maximum similarity value from among the similarity values; and if the maximum similarity value is larger than a preset threshold value, judging that the new registered user belongs to the fraudulent party.
The following describes the above technical solution of the embodiment of the present invention in detail with reference to a specific application example:
step 1, constructing a fraud partner image library, and extracting feature points and feature descriptions of an image background area in the fraud partner image library.
Step 1.1, primarily screening fraud partners through geographic information, IP information, equipment information and the like. And combining manual proposal to construct a rogue group image library for the images which are confirmed to be rogue group and authorized.
And 1.2, inputting image information of a fraud partner image library into a convolutional neural network A, and respectively obtaining a foreground region (namely a face region) and a background region by applying an algorithm 1 corresponding to the neural network A in the system. Then mask the face area to generate a mask, wherein the mask mainly marks the position information of the background area for later steps.
Step 1.3, inputting images of the rogue partner image library, such as the K-th image, into the convolutional neural network B, and extracting mk feature points { (x) of the rogue partner image K by using the algorithm 2 corresponding to the neural network B in the system i ,y i ) And its corresponding d-dimensional feature vector
And step 1.4, combining the characteristic points obtained in the step 1.3 and the mask outputted in the step 1.2, and further screening out the characteristic points and the characteristic vectors of the fraud group image K in the background area. Let n be the sum of intermediate images K after filtering through a mask k Feature points { (x) i ,y i ) And its corresponding d-dimensional feature vector
And step 2, uploading the information of the new user registration image for acquiring the personal authorization of the new user into the system.
And 3, extracting characteristic points and characteristic vectors of the background area of the image of the new registered user according to the same method according to the steps 1.2 to 1.4 in the step 1. Assume that the background area of the image of the newly registered user has q feature points in totalAnd its corresponding feature vector->
And 4, matching the characteristic points of the image background area of the new registered user in the step 3 with the characteristic points of each image background area in the image library of the fraudulent party, and further calculating the similarity between the image background area of the new registered user and the background area of each image of the fraudulent party on the basis of the matching. When the maximum similarity with the rogue group image background area is greater than a certain set threshold, then the new registered user image background is considered to be similar to a certain rogue group image background.
Step 4.1, feature points of the background area of the new registered user image obtained in the step 3 are obtainedFeature vectorFeature points { (x) of background area of image K for fraudulent party i ,y i ) ' feature vectorThe characteristic points are input into the convolutional neural network C together, and the matching quantity of the characteristic points in the background area between the two characteristic points can be obtained by applying an algorithm 3 corresponding to the neural network C in the system;
and 4.2, combining the matching quantity obtained in the step 4.1 with the feature vectors respectively corresponding to the feature points in the two images to obtain the similarity between the image background area of the new registered user and the image background area of the fraudulent party.
Step 4.3, all similarity sets obtained through steps 4.1 to 4.2. If the maximum value in the similarity set is greater than a certain set threshold value, the new registered user is considered to be a suspected fraud partner member, and is subordinate to an intermediary point (a physical place where fraud partner performs fraud) corresponding to the maximum similarity value; otherwise, the user is considered to be a normal user.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. As will be apparent to those skilled in the art; various modifications to these embodiments will be readily apparent, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the disclosure. Thus, the present disclosure is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The foregoing description of the embodiments has been provided for the purpose of illustrating the general principles of the invention, and is not meant to limit the scope of the invention, but to limit the invention to the particular embodiments, and any modifications, equivalents, improvements, etc. that fall within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (10)
1. An anti-partner fraud method based on image background feature recognition, comprising the steps of:
constructing a fraud group image library;
collecting a new registered user image;
and judging whether the new registered user belongs to the fraudulent party or not through the comparison between the image of the new registered user and each fraudulent party image in the fraudulent party image library.
2. An anti-fraud method based on image context feature recognition as defined in claim 1, wherein said determining whether said new registered user belongs to said fraudulent party by comparing said new registered user image with each fraudulent party image in said library of fraudulent party images, in particular comprises:
calculating a similarity value between the background area of the new registered user image and the background area of each fraudulent partner image in the fraudulent partner image library;
selecting a maximum similarity value from among the similarity values;
and if the maximum similarity value is larger than a preset threshold value, judging that the new registered user belongs to the fraudulent party.
3. An anti-fraud method based on image background feature recognition as defined in claim 2, wherein said calculating a similarity value between a background region of said new registered user image and a background region of each fraudulent party image in said fraudulent party image library, in particular comprises:
for each rogue partner image in the rogue partner image library, the following steps are sequentially executed:
performing background area image feature recognition on the current fraud partner image to obtain background area image features of the current fraud partner image, wherein the image features comprise feature points and feature vectors corresponding to the feature points;
performing background area image feature recognition on the new registered user image to obtain new registered user background area image features;
and calculating a similarity value between the background area image characteristics of the new registered user and the background area image characteristics of the current fraud partner image.
4. An anti-partner fraud method based on image background feature recognition as defined in claim 3, characterized by the specific steps of performing background region image feature recognition:
splitting a current image into a foreground region and a background region;
disposing a mask over the foreground region;
identifying all image features of the current image;
and determining the background area image characteristics of the current image according to all the image characteristics of the current image and the mask.
5. An anti-group joining fraud method based on image background feature recognition according to claim 3, characterized in that said calculating a similarity value between the new registered user background region image feature and the background region image feature of the current fraudulent group joining image specifically comprises:
the common characteristic points of the background area of the new registered user and the background area of the current fraud partner image are determined by matching all characteristic points of the background area of the new registered user with all characteristic points of the background area of the current fraud partner image;
determining a common feature vector corresponding to the common feature point;
and calculating the similarity value between the background area image features of the new registered user and the background area image features of the current fraud partner image according to the number of the common feature points and the common feature vector.
6. An anti-partner fraud method based on image background feature recognition as defined in claim 1, wherein said constructing a fraudulent partner image library specifically comprises:
the suspected fraud group partner is determined through preliminary screening;
according to the manual investigation result, determining a fraudulent partner from the suspected fraudulent partner;
invoking an image of a partner member belonging to the fraudulent partner as a fraudulent partner image;
and constructing the rogue partner image library through all the rogue partner images.
7. An anti-partner fraud system based on image background feature recognition, comprising:
the image library construction module is used for constructing a fraud group image library;
the image acquisition module is used for acquiring images of the new registered users;
and the comparison judging module is used for judging whether the new registered user belongs to the fraudulent party or not through the comparison between the image of the new registered user and each fraudulent party image in the fraudulent party image library.
8. The anti-partner fraud system based on image background feature identification of claim 7, wherein the comparison and judgment module is specifically configured to: calculating a similarity value between the background area of the new registered user image and the background area of each fraudulent partner image in the fraudulent partner image library; selecting a maximum similarity value from among the similarity values; and if the maximum similarity value is larger than a preset threshold value, judging that the new registered user belongs to the fraudulent party.
9. A computer readable storage medium having stored thereon a computer program which when executed by a processor implements an anti-partner fraud method based on image background feature identification as defined in any of claims 1-6.
10. A computer device, comprising:
one or more processors;
a storage means for storing one or more programs;
the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the anti-group partner fraud method of any of claims 1-6 based on image background feature identification.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310521449.1A CN116543181A (en) | 2023-05-10 | 2023-05-10 | Anti-partner fraud method and system based on image background feature recognition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310521449.1A CN116543181A (en) | 2023-05-10 | 2023-05-10 | Anti-partner fraud method and system based on image background feature recognition |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116543181A true CN116543181A (en) | 2023-08-04 |
Family
ID=87455558
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310521449.1A Pending CN116543181A (en) | 2023-05-10 | 2023-05-10 | Anti-partner fraud method and system based on image background feature recognition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116543181A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160086015A1 (en) * | 2007-01-09 | 2016-03-24 | Si Corporation | Method and system for automated face detection and recognition |
| CN109145139A (en) * | 2018-09-25 | 2019-01-04 | 北京市商汤科技开发有限公司 | A kind of image search method and device, equipment and storage medium |
| CN110348519A (en) * | 2019-07-12 | 2019-10-18 | 深圳众赢维融科技有限公司 | Financial product cheats recognition methods and the device of clique |
| WO2021169631A1 (en) * | 2020-02-29 | 2021-09-02 | 深圳壹账通智能科技有限公司 | Fraudster identification method, apparatus and device, and storage medium |
| CN114462553A (en) * | 2022-04-12 | 2022-05-10 | 之江实验室 | Image labeling and element extraction method and system for car insurance fraud prevention |
| CN114926725A (en) * | 2022-07-18 | 2022-08-19 | 中邮消费金融有限公司 | Online financial group partner fraud identification method based on image analysis |
-
2023
- 2023-05-10 CN CN202310521449.1A patent/CN116543181A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160086015A1 (en) * | 2007-01-09 | 2016-03-24 | Si Corporation | Method and system for automated face detection and recognition |
| CN109145139A (en) * | 2018-09-25 | 2019-01-04 | 北京市商汤科技开发有限公司 | A kind of image search method and device, equipment and storage medium |
| CN110348519A (en) * | 2019-07-12 | 2019-10-18 | 深圳众赢维融科技有限公司 | Financial product cheats recognition methods and the device of clique |
| WO2021169631A1 (en) * | 2020-02-29 | 2021-09-02 | 深圳壹账通智能科技有限公司 | Fraudster identification method, apparatus and device, and storage medium |
| CN114462553A (en) * | 2022-04-12 | 2022-05-10 | 之江实验室 | Image labeling and element extraction method and system for car insurance fraud prevention |
| CN114926725A (en) * | 2022-07-18 | 2022-08-19 | 中邮消费金融有限公司 | Online financial group partner fraud identification method based on image analysis |
Non-Patent Citations (1)
| Title |
|---|
| 周晓楠等: "图数据库在识别重大疾病保险团伙式欺诈中的应用研究", 《保险研究》, no. 09, pages 92 - 104 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111861240A (en) | Suspicious user identification method, device, equipment and readable storage medium | |
| WO2018082011A1 (en) | Living fingerprint recognition method and device | |
| CN111160313B (en) | Face representation attack detection method based on LBP-VAE anomaly detection model | |
| CN109784015B (en) | Identity authentication method and device | |
| CN111191568A (en) | Method, device, equipment and medium for identifying copied image | |
| CN108206813B (en) | Security audit method and device based on k-means clustering algorithm and server | |
| CN112329811A (en) | Abnormal account identification method and device, computer equipment and storage medium | |
| WO2020093303A1 (en) | Processing method and apparatus based on facial recognition, and device and readable storage medium | |
| CN111339897B (en) | Living body identification method, living body identification device, computer device, and storage medium | |
| CN102637255A (en) | Method and device for processing faces contained in images | |
| CN112418167A (en) | Image clustering method, device, equipment and storage medium | |
| CN112199530B (en) | Multi-dimensional face library picture automatic updating method, system, equipment and medium | |
| CN111488798B (en) | Fingerprint identification method, fingerprint identification device, electronic equipment and storage medium | |
| CN111553241B (en) | Palm print mismatching point eliminating method, device, equipment and storage medium | |
| CN113591603A (en) | Certificate verification method and device, electronic equipment and storage medium | |
| CN111259894B (en) | Certificate information identification method and device and computer equipment | |
| CN114067381A (en) | Deep forgery identification method and device based on multi-feature fusion | |
| WO2018137226A1 (en) | Fingerprint extraction method and device | |
| CN110290101B (en) | Deep trust network-based associated attack behavior identification method in smart grid environment | |
| CN116543181A (en) | Anti-partner fraud method and system based on image background feature recognition | |
| CN114841705B (en) | Anti-fraud monitoring method based on scene recognition | |
| Raskar et al. | VFDHSOG: copy-move video forgery detection using histogram of second order gradients | |
| CN114373213A (en) | Juvenile identity recognition method and device based on face recognition | |
| CN114005184A (en) | Handwritten signature authenticity identification method and device based on small amount of samples | |
| CN104615922B (en) | A kind of identification authentication system and method based on user software behavior pattern |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |