CN111178203A - Signature verification method and device, computer equipment and storage medium - Google Patents
Signature verification method and device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN111178203A CN111178203A CN201911323852.3A CN201911323852A CN111178203A CN 111178203 A CN111178203 A CN 111178203A CN 201911323852 A CN201911323852 A CN 201911323852A CN 111178203 A CN111178203 A CN 111178203A
- Authority
- CN
- China
- Prior art keywords
- signature
- features
- image
- historical
- signature image
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/30—Writer recognition; Reading and verifying signatures
- G06V40/33—Writer recognition; Reading and verifying signatures based only on signature image, e.g. static signature recognition
-
- 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
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Library & Information Science (AREA)
- Human Computer Interaction (AREA)
- Databases & Information Systems (AREA)
- Collating Specific Patterns (AREA)
- Image Analysis (AREA)
Abstract
The method comprises the steps of obtaining a signature image to be checked corresponding to a user identifier, extracting edge features and handwriting features of the signature image to obtain combined current signature features, inquiring historical signature features corresponding to the user identifier, further detecting whether the current signature features are matched with the inquired historical signature features by adopting a deep learning neural network model, and determining that the signature image is checked to be passed if the current signature features are matched with the inquired historical signature features, so that automatic checking of the signature image is completed, the checking speed of the signature image is improved, and the accuracy is high.
Description
Technical Field
The present application relates to the field of image recognition technologies, and in particular, to a signature verification method and apparatus, a computer device, and a storage medium.
Background
With the continuous development of social economy, the daily business development of banks is also increased gradually. However, the amount of handwritten signatures in the accompanying business is more and more, so the auditing work of handwritten signatures is more and more complicated, and the traditional manual auditing can not meet the requirements.
Based on this, a handwritten signature recognition and verification system is proposed, and at present, recognition and verification of handwritten signatures mainly focuses on two directions of handwritten single character recognition and signature whole character recognition. While single character recognition mainly takes numeric characters and single Chinese characters as main characters, the method cannot recognize a plurality of written continuous characters and needs to recognize the continuous characters after being segmented. Although the whole signature character recognition can effectively recognize continuous characters, the traditional whole character recognition has low accuracy.
Disclosure of Invention
Therefore, it is necessary to provide a signature verification method, an apparatus, a computer device, and a storage medium, which can effectively improve the verification accuracy of handwritten signatures, in order to solve the problem that the verification accuracy of the conventional handwritten signatures is not high.
In order to achieve the above object, in one aspect, an embodiment of the present application provides a signature auditing method, where the method includes:
acquiring a signature image to be checked corresponding to the user identifier;
extracting edge features and handwriting features of the signature image to obtain combined current signature features;
querying historical signature characteristics corresponding to the user identification;
detecting whether the current signature features are matched with the inquired historical signature features by adopting a deep learning neural network model;
and if the images are matched, determining that the signature image is approved.
In one embodiment, extracting edge features and handwriting features of the signature image to obtain a merged current signature feature includes: extracting edge features of the signature image by adopting a first neural network; extracting handwriting characteristics of the signature image by adopting a second neural network; and combining the edge characteristics and the handwriting characteristics in a column-by-column combination mode to obtain the current signature characteristics.
In one embodiment, before querying the historical signature feature corresponding to the user identifier, the method further includes: acquiring a historical signature image, wherein the historical signature image has a corresponding user identifier; extracting historical signature characteristics of the historical signature image; and storing the historical signature characteristics in a distributed database according to the user identification corresponding to the historical signature image.
In one embodiment, querying the historical signature features corresponding to the user identification comprises: searching historical signature characteristics corresponding to the user identification in a distributed database by adopting a word counting command; and if the number of the searched historical signature features reaches the target number, extracting the historical signature features of the target number.
In one embodiment, the generation method of the deep learning neural network model comprises the following steps: acquiring a sample signature image dataset, wherein the sample signature image dataset comprises a positive sample data set and a negative sample data set; and training a support vector machine through the positive sample data set and the negative sample data set respectively until the parameters are converged to obtain the deep learning neural network model.
In one embodiment, training a support vector machine with a positive sample data set and a negative sample data set, respectively, comprises: extracting sample signature characteristics corresponding to each sample signature image in the positive sample data set and the negative sample data set; and training a support vector machine through sample signature characteristics corresponding to each sample signature image in the positive sample data set and the negative sample data set respectively.
In one embodiment, after determining that the signature image is approved, the method further includes: and storing the current signature characteristics corresponding to the signature image in a distributed database according to the user identification.
On the other hand, an embodiment of the present application further provides a signature verifying apparatus, where the apparatus includes:
the image acquisition module is used for acquiring a signature image to be audited corresponding to the user identifier;
the characteristic extraction module is used for extracting the edge characteristic and the handwriting characteristic of the signature image to obtain the combined current signature characteristic;
a query module for querying the historical signature characteristics corresponding to the user identification,
the characteristic detection module is used for detecting whether the current signature characteristic is matched with the inquired historical signature characteristic by adopting a deep learning neural network model; and if the images are matched, determining that the signature image is approved.
In yet another aspect, the present application further provides a computer device, which includes a memory and a processor, where the memory stores a computer program, and the processor implements the steps of the method when executing the computer program.
In yet another aspect, the present application further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the steps of the method described above.
According to the signature auditing method, the signature auditing device, the computer equipment and the storage medium, the signature image to be audited corresponding to the user identification is obtained, the edge characteristic and the handwriting characteristic of the signature image are extracted to obtain the combined current signature characteristic, the historical signature characteristic corresponding to the user identification is inquired, then the deep learning neural network model is adopted to detect whether the current signature characteristic is matched with the inquired historical signature characteristic, if so, the signature image is determined to be audited to be passed, so that the signature image is audited automatically, the auditing speed of the signature image is improved, and the accuracy is high.
Drawings
FIG. 1 is a diagram of an application environment of a signature review method in one embodiment;
FIG. 2 is a schematic flow diagram of a signature review method in one embodiment;
FIG. 3 is a schematic flow chart of the steps of extracting signature features in one embodiment;
FIG. 4 is a flowchart illustrating a signature review method according to another embodiment;
FIG. 5 is a schematic flow chart diagram illustrating the steps for generating a deep learning neural network model in one embodiment;
FIG. 6 is a flowchart illustrating a signature review method according to yet another embodiment;
FIG. 7 is a block diagram of an arrangement of a signature review device in one embodiment;
FIG. 8 is a diagram illustrating an internal structure of a computer device according to an embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
The signature auditing method provided by the application can be applied to the application environment shown in FIG. 1. The terminal 102 and the server 104 communicate with each other through a network, the terminal 102 may be various devices having an image capturing or storing function, such as but not limited to various smart phones, tablet computers, cameras, and portable image capturing devices, and the server 104 may be implemented by an independent server or a server cluster formed by a plurality of servers. Specifically, the terminal 102 is configured to collect or store a signature image, and send the collected signature image to the server 104 through a network, but the signature image may also be stored in the server 104 in advance. After the server 104 obtains the signature image to be checked corresponding to the user identifier, the edge features and the handwriting features of the signature image are extracted to obtain the combined current signature features, the historical signature features corresponding to the user identifier are inquired, whether the current signature features are matched with the inquired historical signature features or not is further detected by adopting the deep learning neural network model, and if the current signature features are matched with the inquired historical signature features, the signature image is determined to be checked to be passed, so that the automatic checking of the signature image is completed, the checking speed of the signature image is increased, and the accuracy is high. And can be widely applied to intelligent counters, counter cleaning systems, paperless transaction services of mobile halls and the like.
In one embodiment, as shown in fig. 2, a signature auditing method is provided, which is described by taking the method as an example applied to the server in fig. 1, and includes the following steps:
The signature image to be verified is the signature image needing authenticity verification, namely whether the signature handwriting in the image is the personal handwriting is verified. The user identifier is a unique identifier that can distinguish different users, such as an identification number of a user. Specifically, when a user transacts financial business or other business with real name and needing a user signature online or offline, the user is usually associated with a user identifier, the authenticity of the user signature needs to be verified, and the next process of the business is only entered after the authenticity verification is passed. In this embodiment, the signature image to be checked may be a vector file of a signature written by the user, which is collected during the transaction process of the user.
And step 204, extracting edge features and handwriting features of the signature image to obtain combined current signature features.
The edge features refer to edges with obvious changes or discontinuous areas in an image, generally the edges are boundary lines between different areas in an image, and the purpose of extracting the edge features is to capture the areas with sharp changes of brightness in the image. The handwriting characteristics refer to the specific image that one person handwriting is different from the other person handwriting. Therefore, in the embodiment, the edge features and the handwriting features of the signature image are respectively extracted and combined, so that the combined current signature features are obtained for subsequent handwriting inspection.
The historical signature characteristics are obtained by processing effective historical signature images of the user and are stored in an associated mode according to the user identification. Therefore, in this embodiment, the historical signature feature corresponding to the user identifier of the signature image to be checked can be searched in the stored historical signature features, so as to be used as a basis for judging the authenticity of the signature image to be checked.
And step 208, detecting whether the current signature features are matched with the inquired historical signature features by using a deep learning neural network model.
The deep learning neural network model is obtained after a support vector machine is trained and is used for detecting whether the current signature characteristics are matched with the inquired historical signature characteristics or not, namely judging whether the current signature characteristics and the inquired historical signature characteristics belong to the handwriting of the same person or not.
And step 210, if the images are matched, determining that the signature image is approved.
Specifically, after the deep learning neural network model is detected, if the current signature characteristic is determined to be matched with the inquired historical signature characteristic, the current signature characteristic and the inquired historical signature characteristic can be judged to belong to the handwriting of the same person, so that the signature image can be determined to be approved, and the next process of the business can be automatically entered.
According to the signature auditing method, after the signature image to be audited corresponding to the user identification is obtained, the edge features and the handwriting features of the signature image are extracted to obtain the combined current signature features, the historical signature features corresponding to the user identification are inquired, then the deep learning neural network model is adopted to detect whether the current signature features are matched with the inquired historical signature features, if the current signature features are matched with the inquired historical signature features, the signature image is determined to be approved, and therefore automatic audit on the signature image is completed, the auditing speed of the signature image is improved, and high accuracy is achieved.
In one embodiment, after deep learning neural network model detection, if it is determined that the current signature feature does not match the queried historical signature feature, it may be determined that the current signature feature and the queried historical signature feature do not belong to the same person's handwriting, and therefore, an alarm may be issued to prompt a worker to further perform manual checking on the signature image, i.e., checking the identity of the user currently submitting the signature image, so as to ensure that the service can be performed normally.
In an embodiment, as shown in fig. 3, extracting edge features and handwriting features of the signature image to obtain a merged current signature feature may specifically include the following steps:
The first neural network may be specifically implemented by a deep neural network vgg-16(Visual Geometry group network). Specifically, the deep neural network vgg-16 extracts 244 × 224 × 3 dimensional feature data of the signature image through 5 rounds of convolution and pooling layers, and converts the extracted feature data into 1000 dimensional edge features, thereby obtaining the edge features of the signature image.
And step 304, extracting handwriting features of the signature image by adopting a second neural network.
The second neural network may specifically be implemented by a Convolutional Neural Network (CNN). Specifically, the CNN extracts feature data of 224 × 1 dimension of the signature image through the convolution layer, the pooling layer, and the full-connection layer connected in sequence, converts the extracted feature data into 1000-dimensional handwriting edge features, and then forms the 1000-dimensional handwriting features through the long and short memory neural network, thereby obtaining the handwriting features of the signature image.
And step 306, combining the edge features and the handwriting features in a column combination mode to obtain current signature features.
Specifically, in this embodiment, the edge features and the handwriting features of the signature image extracted in the above steps are combined in a column-by-column combination manner, so as to obtain a current signature feature with a length of 1000 and a height of 2.
In the embodiment, the edge features of the signature image are extracted by adopting the first neural network, the handwriting features of the signature image are extracted by adopting the second neural network, and the edge features and the handwriting features are combined in a column-combining manner to obtain the current signature features, so that the accuracy and robustness of the subsequent detection of the current signature features are effectively improved.
In one embodiment, as shown in fig. 4, before querying the historical signature feature corresponding to the user identifier, the method may further include the following steps:
The historical signature image is a valid signature record of a user when transacting financial business or other real-name business online or offline in the past, and is usually associated with a user identifier.
And step 404, extracting historical signature characteristics of the historical signature image.
Specifically, the historical signature image may be processed by the method shown in fig. 3, so as to extract the historical signature features corresponding to the historical signature image.
And 406, storing the historical signature characteristics in a distributed database according to the user identification corresponding to the historical signature image.
The distributed database may be implemented by a Hadoop (Hadoop, which is a distributed system infrastructure developed by Apache foundation) distributed database based on a spark (computing engine) platform. In this embodiment, the historical signature features are stored in the distributed database according to the user identifier, that is, the user identifier is associated with the corresponding historical signature features during storage.
And further, when the historical signature characteristics corresponding to the user identification are inquired, the historical signature characteristics corresponding to the user identification can be searched in the distributed database by adopting a word counting command. The word statistic command can specifically adopt a spark-based word statistic command, that is, historical signature features corresponding to the user identifier are counted in the distributed database through the spark-based word statistic command, and if the counted number of the historical signature features corresponding to the user identifier reaches a target number, the historical signature features of the target number are extracted for detection of a subsequent deep learning neural network model. Specifically, the target number can be set according to an actual situation, and can be set to be 3-10 generally, the higher the target number is set, the higher the accuracy of the final detection result is, and the too low the target number is set, the accuracy of the detection result is not high due to the fact that the handwriting characteristics of the historical signature cannot be reflected.
In one embodiment, it is understood that the historical signature image may further include a corresponding service identifier, that is, the user identifier, the historical signature image, and the corresponding service identifier are associated when being stored. Similarly, the signature image to be checked also has a corresponding service identifier, so that when statistics is performed through a spark word statistic command in the distributed database, records with a service identifier field word frequency of 4, namely 4 target numbers, of user identifiers and signatures can be counted, and accordingly, corresponding number of historical signature features are extracted for detection of a subsequent deep learning neural network model.
In one embodiment, as shown in fig. 5, the method for generating the deep learning neural network model may include the following steps:
Since the deep learning neural network model is obtained by training a support vector machine, a sample signature image dataset for training the support vector machine needs to be acquired first. In this embodiment, the sample signature image data set includes a positive sample data set and a negative sample data set, where the positive sample data set includes a plurality of sample pairs, and each sample pair includes a personal signature feature of a same person signature and a plurality of corresponding historical signature features; the negative sample data set also includes a number of sample pairs, but the signature feature and the plurality of historical signature features in each sample pair do not belong to the same person signature. It can be understood that the signature feature and the historical signature feature are obtained by processing the corresponding signature image by the method shown in fig. 3, and are not described in detail in this embodiment.
And step 504, training a support vector machine through the positive sample data set and the negative sample data set respectively until the parameters are converged to obtain a deep learning neural network model.
In this embodiment, the support vector machine is trained through the acquired positive sample data set and the acquired negative sample data set, so that the model parameters are updated in continuous iterative learning until the parameters converge, and the support vector machine can learn the signature characteristics of the positive sample data set and the signature characteristics of the negative sample data set to obtain a deep learning neural network model, thereby improving the capability of distinguishing whether the signature characteristics are of the same person.
Specifically, the objective function of the support vector machine is
Wherein the constraint condition of corresponding characteristic is that y is equal to w' phi (x)i+b+εi) Further, a feature determination function of
Wherein the weight value thetai＝CsiεiC is a penalty factor, which is an adjustable parameter in the range of 1 to 100, epsiloniAs an error in the function, siThen the signature feature similarity distance. The signature feature similarity distance may adopt a Person correlation coefficient distance calculation method, where dist (X, Y) is 1- ρ X, Y, ρ X, and Y is Cov (X, Y)/σXσY＝E((X-μX)(Y-μY))/σXσYX, Y refer to a set of positive and negative sample data, uX,uYIs the mean, σ, of signature features in the set of positive and negative sample dataX,σYIs the variance of the signature features in the positive and negative sample data sets.
By increasing the similarity distance parameter of the signature features, the classification capability of a support vector machine for distinguishing the signature features of the person and the signature features of the person who does not belong to the person can be effectively increasedAnd the accuracy of the model is improved. The kernel function phi (x) of the alignment features used thereini) Is min (x (i), x)s(i) Wherein x (i), xs(i) Is the signature feature extracted from any two signature images, if x (i) and x (x) in model trainings(i) The signature label of the same person is set to be 1, the signature label of the different person is not set to be-1, and the weight theta is obtained through trainingiAnd the offset b, so that training is completed, a deep learning neural network model is obtained, and the accuracy of the model for predicting whether the current signature image is the self is effectively improved.
In one embodiment, detecting whether the current signature features match the queried historical signature features by using a deep learning neural network model specifically includes: combining the current signature characteristic with the historical signature characteristic with the length of 1000 and the height of 2 to form a signature characteristic independent variable with 8 x 1000 dimensions, taking the signature characteristic independent variable as an independent variable input by a trained deep learning neural network model, predicting whether the handwriting of the current signature characteristic is the signature of the user through the signature characteristic independent variable based on the reason that the handwriting characteristic has strong dependence on the space position, namely judging through the characteristic judgment function, if the value of the characteristic judgment function is 1, indicating that the signature of the user is the signature of the user, namely, the current signature characteristic is matched with the historical signature characteristic, then passing the audit, and if the value of the characteristic judgment function is-1, indicating that the signature of the user is not the signature of the user, namely, the current signature characteristic is not matched with the historical signature characteristic, then transferring the manual work for further audit.
In one embodiment, as shown in fig. 6, after determining that the signature image audit is passed, the method may further include the following steps:
and step 212, storing the current signature characteristics corresponding to the signature image in a distributed database according to the user identification.
Specifically, in this embodiment, after it is determined through the above steps that the signature image is approved, it may be determined that the signature image is valid, and therefore, the current signature feature corresponding to the signature image may be stored in the distributed database according to the user identifier, so as to be used as the historical signature feature.
It should be understood that although the various steps in the flow charts of fig. 1-6 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least some of the steps in fig. 1-6 may include multiple sub-steps or multiple stages that are not necessarily performed at the same time, but may be performed at different times, and the order of performance of the sub-steps or stages is not necessarily sequential, but may be performed in turn or alternating with other steps or at least some of the sub-steps or stages of other steps.
In one embodiment, as shown in fig. 7, there is provided a signature auditing apparatus including: an image acquisition module 701, a feature extraction module 702, a query module 703, and a feature detection module 704, wherein:
an image obtaining module 701, configured to obtain a signature image to be audited corresponding to the user identifier;
a feature extraction module 702, configured to extract edge features and handwriting features of the signature image to obtain a merged current signature feature;
a query module 703 for querying the historical signature characteristics corresponding to the user identifier,
a feature detection module 704, configured to detect whether a current signature feature matches the queried historical signature feature by using a deep learning neural network model; and if the images are matched, determining that the signature image is approved.
In one embodiment, the feature extraction module 702 is specifically configured to: extracting edge features of the signature image by adopting a first neural network; extracting handwriting characteristics of the signature image by adopting a second neural network; and combining the edge characteristics and the handwriting characteristics in a column-by-column combination mode to obtain the current signature characteristics.
In an embodiment, the signature auditing apparatus further includes a historical signature feature obtaining module, configured to obtain a historical signature image, where the historical signature image has a corresponding user identifier; extracting historical signature characteristics of the historical signature image; and storing the historical signature characteristics in a distributed database according to the user identification corresponding to the historical signature image.
In one embodiment, the query module 703 is specifically configured to: searching historical signature characteristics corresponding to the user identification in a distributed database by adopting a word counting command; and if the number of the searched historical signature features reaches the target number, extracting the historical signature features of the target number.
In one embodiment, the deep learning neural network model includes: the system comprises a sample acquisition module, a data processing module and a data processing module, wherein the sample acquisition module is used for acquiring a sample signature image dataset, and the sample signature image dataset comprises a positive sample data set and a negative sample data set; and the training module is used for training the support vector machine through the positive sample data set and the negative sample data set respectively until the parameters are converged to obtain the deep learning neural network model.
In one embodiment, the training module is specifically configured to: extracting sample signature characteristics corresponding to each sample signature image in the positive sample data set and the negative sample data set; and training a support vector machine through sample signature characteristics corresponding to each sample signature image in the positive sample data set and the negative sample data set respectively.
In an embodiment, the signature verification apparatus further includes a storage module, configured to store, after it is determined that the signature image is verified to be passed, a current signature feature corresponding to the signature image in a distributed database according to the user identifier.
For the specific definition of the signature verification apparatus, reference may be made to the above definition of the signature verification method, which is not described herein again. The modules in the signature verification device can be wholly or partially implemented by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in fig. 8. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer device is for a signature image to be reviewed and a sample signature image dataset. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a signature auditing method.
Those skilled in the art will appreciate that the architecture shown in fig. 8 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having a computer program stored therein, the processor implementing the following steps when executing the computer program:
acquiring a signature image to be checked corresponding to the user identifier;
extracting edge features and handwriting features of the signature image to obtain combined current signature features;
querying historical signature characteristics corresponding to the user identification;
detecting whether the current signature features are matched with the inquired historical signature features by adopting a deep learning neural network model;
and if the images are matched, determining that the signature image is approved.
In one embodiment, the processor, when executing the computer program, further performs the steps of: extracting edge features of the signature image by adopting a first neural network; extracting handwriting characteristics of the signature image by adopting a second neural network; and combining the edge characteristics and the handwriting characteristics in a column-by-column combination mode to obtain the current signature characteristics.
In one embodiment, the processor, when executing the computer program, further performs the steps of: acquiring a historical signature image before querying a historical signature characteristic corresponding to a user identifier, wherein the historical signature image has the corresponding user identifier; extracting historical signature characteristics of the historical signature image; and storing the historical signature characteristics in a distributed database according to the user identification corresponding to the historical signature image.
In one embodiment, the processor, when executing the computer program, further performs the steps of: searching historical signature characteristics corresponding to the user identification in a distributed database by adopting a word counting command; and if the number of the searched historical signature features reaches the target number, extracting the historical signature features of the target number.
In one embodiment, the processor, when executing the computer program, further performs the steps of: acquiring a sample signature image dataset, wherein the sample signature image dataset comprises a positive sample data set and a negative sample data set; and training a support vector machine through the positive sample data set and the negative sample data set respectively until the parameters are converged to obtain the deep learning neural network model.
In one embodiment, the processor, when executing the computer program, further performs the steps of: extracting sample signature characteristics corresponding to each sample signature image in the positive sample data set and the negative sample data set; and training a support vector machine through sample signature characteristics corresponding to each sample signature image in the positive sample data set and the negative sample data set respectively.
In one embodiment, the processor, when executing the computer program, further performs the steps of: and after the signature image is confirmed to be approved, storing the current signature characteristics corresponding to the signature image in a distributed database according to the user identification.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of:
acquiring a signature image to be checked corresponding to the user identifier;
extracting edge features and handwriting features of the signature image to obtain combined current signature features;
querying historical signature characteristics corresponding to the user identification;
detecting whether the current signature features are matched with the inquired historical signature features by adopting a deep learning neural network model;
and if the images are matched, determining that the signature image is approved.
In one embodiment, the computer program when executed by the processor further performs the steps of: extracting edge features of the signature image by adopting a first neural network; extracting handwriting characteristics of the signature image by adopting a second neural network; and combining the edge characteristics and the handwriting characteristics in a column-by-column combination mode to obtain the current signature characteristics.
In one embodiment, the computer program when executed by the processor further performs the steps of: acquiring a historical signature image before querying a historical signature characteristic corresponding to a user identifier, wherein the historical signature image has the corresponding user identifier; extracting historical signature characteristics of the historical signature image; and storing the historical signature characteristics in a distributed database according to the user identification corresponding to the historical signature image.
In one embodiment, the computer program when executed by the processor further performs the steps of: searching historical signature characteristics corresponding to the user identification in a distributed database by adopting a word counting command; and if the number of the searched historical signature features reaches the target number, extracting the historical signature features of the target number.
In one embodiment, the computer program when executed by the processor further performs the steps of: acquiring a sample signature image dataset, wherein the sample signature image dataset comprises a positive sample data set and a negative sample data set; and training a support vector machine through the positive sample data set and the negative sample data set respectively until the parameters are converged to obtain the deep learning neural network model.
In one embodiment, the computer program when executed by the processor further performs the steps of: extracting sample signature characteristics corresponding to each sample signature image in the positive sample data set and the negative sample data set; and training a support vector machine through sample signature characteristics corresponding to each sample signature image in the positive sample data set and the negative sample data set respectively.
In one embodiment, the computer program when executed by the processor further performs the steps of: and after the signature image is confirmed to be approved, storing the current signature characteristics corresponding to the signature image in a distributed database according to the user identification.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (10)
1. A signature review method, the method comprising:
acquiring a signature image to be checked corresponding to the user identifier;
extracting edge features and handwriting features of the signature image to obtain combined current signature features;
querying historical signature characteristics corresponding to the user identification;
detecting whether the current signature features are matched with the inquired historical signature features by adopting a deep learning neural network model;
and if the images are matched, determining that the signature image is approved.
2. The signature review method according to claim 1, wherein the extracting edge features and handwriting features of the signature image to obtain combined current signature features comprises:
extracting edge features of the signature image by adopting a first neural network;
extracting handwriting characteristics of the signature image by adopting a second neural network;
and combining the edge features and the handwriting features in a column-by-column combination mode to obtain the current signature features.
3. The signature review method according to claim 1, wherein before querying the historical signature features corresponding to the user identifier, the method further comprises:
acquiring a historical signature image, wherein the historical signature image has a corresponding user identifier;
extracting historical signature characteristics of the historical signature image;
and storing the historical signature characteristics in a distributed database according to the user identification corresponding to the historical signature image.
4. The signature review method according to claim 3, wherein the querying for the historical signature features corresponding to the user identifier includes:
searching historical signature characteristics corresponding to the user identification in the distributed database by adopting a word counting command;
and if the number of the searched historical signature features reaches the target number, extracting the historical signature features of the target number.
5. The signature auditing method according to claim 1, where the method of generating the deep-learning neural network model comprises:
acquiring a sample signature image dataset, wherein the sample signature image dataset comprises a positive sample data set and a negative sample data set;
and training a support vector machine through the positive sample data set and the negative sample data set respectively until parameters are converged to obtain the deep learning neural network model.
6. The signature auditing method according to claim 5, wherein said training a support vector machine with said positive and negative sample data sets, respectively, comprises:
extracting sample signature characteristics corresponding to each sample signature image in the positive sample data set and the negative sample data set;
and training the support vector machine respectively through the sample signature characteristics corresponding to each sample signature image in the positive sample data set and the negative sample data set.
7. The signature review method according to claim 3, wherein after determining that the signature image review passes, the method further comprises:
and storing the current signature characteristics corresponding to the signature image in the distributed database according to the user identification.
8. A signature review apparatus, characterized in that the apparatus comprises:
the image acquisition module is used for acquiring a signature image to be audited corresponding to the user identifier;
the feature extraction module is used for extracting edge features and handwriting features of the signature image to obtain combined current signature features;
a query module for querying the historical signature characteristics corresponding to the user identification,
the characteristic detection module is used for detecting whether the current signature characteristic is matched with the inquired historical signature characteristic by adopting a deep learning neural network model; and if the images are matched, determining that the signature image is approved.
9. A computer device comprising a memory and a processor, the memory storing a computer program, wherein the processor implements the steps of the method of any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911323852.3A CN111178203B (en) | 2019-12-20 | 2019-12-20 | Signature verification method and device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911323852.3A CN111178203B (en) | 2019-12-20 | 2019-12-20 | Signature verification method and device, computer equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111178203A true CN111178203A (en) | 2020-05-19 |
| CN111178203B CN111178203B (en) | 2021-01-29 |
Family
ID=70652000
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911323852.3A Active CN111178203B (en) | 2019-12-20 | 2019-12-20 | Signature verification method and device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111178203B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112215087A (en) * | 2020-09-21 | 2021-01-12 | 浙江数链科技有限公司 | Picture auditing method and device, computer equipment and storage medium |
| CN112766082A (en) * | 2020-12-30 | 2021-05-07 | 大连海事大学 | Chinese text handwriting identification method and device based on macro-micro characteristics and storage medium |
| CN116630033A (en) * | 2023-07-20 | 2023-08-22 | 杭银消费金融股份有限公司 | Information auditing method, system and storage medium |
Citations (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090220166A1 (en) * | 2008-02-28 | 2009-09-03 | Yahoo! Inc. | Filter for blocking image-based spam |
| CN102629322A (en) * | 2012-03-12 | 2012-08-08 | 华中科技大学 | Character feature extraction method based on stroke shape of boundary point and application thereof |
| CN105787503A (en) * | 2016-01-15 | 2016-07-20 | 四川德马克机器人科技有限公司 | Method for identifying plant leaf based on early state feature fusion |
| CN106529255A (en) * | 2016-11-14 | 2017-03-22 | 北京爱知之星科技股份有限公司 | Handwriting fingerprint based user identity identification method and device |
| CN106599941A (en) * | 2016-12-12 | 2017-04-26 | 西安电子科技大学 | Method for identifying handwritten numbers based on convolutional neural network and support vector machine |
| CN106778586A (en) * | 2016-12-08 | 2017-05-31 | 武汉理工大学 | Offline handwriting signature verification method and system |
| US20170249536A1 (en) * | 2016-02-29 | 2017-08-31 | Christopher J. Hillar | Self-Organizing Discrete Recurrent Network Digital Image Codec |
| CN107357815A (en) * | 2017-06-07 | 2017-11-17 | 上海斐讯数据通信技术有限公司 | A kind of recognition methods of image content and system |
| CN107766809A (en) * | 2017-10-09 | 2018-03-06 | 平安科技（深圳）有限公司 | Electronic installation, billing information recognition methods and computer-readable recording medium |
| CN108121991A (en) * | 2018-01-06 | 2018-06-05 | 北京航空航天大学 | A kind of deep learning Ship Target Detection method based on the extraction of edge candidate region |
| CN108875811A (en) * | 2018-06-01 | 2018-11-23 | 平安科技（深圳）有限公司 | Classification method, device, computer equipment and the storage medium of animal species |
| US20180373925A1 (en) * | 2017-06-22 | 2018-12-27 | Koninklijke Philips N.V. | Subject identification systems and methods |
| CN109284675A (en) * | 2018-08-13 | 2019-01-29 | 阿里巴巴集团控股有限公司 | A kind of recognition methods of user, device and equipment |
| CN109446905A (en) * | 2018-09-26 | 2019-03-08 | 深圳壹账通智能科技有限公司 | Sign electronically checking method, device, computer equipment and storage medium |
| CN109583227A (en) * | 2018-10-30 | 2019-04-05 | 中国科学院信息工程研究所 | A kind of method for protecting privacy, apparatus and system |
| CN109858927A (en) * | 2019-01-16 | 2019-06-07 | 深圳壹账通智能科技有限公司 | A kind of trade company's checking method, device, computer readable storage medium and server |
| CN109934809A (en) * | 2019-03-08 | 2019-06-25 | 深慧视(深圳)科技有限公司 | A kind of paper labels character defect inspection method |
-
2019
- 2019-12-20 CN CN201911323852.3A patent/CN111178203B/en active Active
Patent Citations (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090220166A1 (en) * | 2008-02-28 | 2009-09-03 | Yahoo! Inc. | Filter for blocking image-based spam |
| CN102629322A (en) * | 2012-03-12 | 2012-08-08 | 华中科技大学 | Character feature extraction method based on stroke shape of boundary point and application thereof |
| CN105787503A (en) * | 2016-01-15 | 2016-07-20 | 四川德马克机器人科技有限公司 | Method for identifying plant leaf based on early state feature fusion |
| US20170249536A1 (en) * | 2016-02-29 | 2017-08-31 | Christopher J. Hillar | Self-Organizing Discrete Recurrent Network Digital Image Codec |
| CN106529255A (en) * | 2016-11-14 | 2017-03-22 | 北京爱知之星科技股份有限公司 | Handwriting fingerprint based user identity identification method and device |
| CN106778586A (en) * | 2016-12-08 | 2017-05-31 | 武汉理工大学 | Offline handwriting signature verification method and system |
| CN106599941A (en) * | 2016-12-12 | 2017-04-26 | 西安电子科技大学 | Method for identifying handwritten numbers based on convolutional neural network and support vector machine |
| CN107357815A (en) * | 2017-06-07 | 2017-11-17 | 上海斐讯数据通信技术有限公司 | A kind of recognition methods of image content and system |
| US20180373925A1 (en) * | 2017-06-22 | 2018-12-27 | Koninklijke Philips N.V. | Subject identification systems and methods |
| CN107766809A (en) * | 2017-10-09 | 2018-03-06 | 平安科技（深圳）有限公司 | Electronic installation, billing information recognition methods and computer-readable recording medium |
| CN108121991A (en) * | 2018-01-06 | 2018-06-05 | 北京航空航天大学 | A kind of deep learning Ship Target Detection method based on the extraction of edge candidate region |
| CN108875811A (en) * | 2018-06-01 | 2018-11-23 | 平安科技（深圳）有限公司 | Classification method, device, computer equipment and the storage medium of animal species |
| CN109284675A (en) * | 2018-08-13 | 2019-01-29 | 阿里巴巴集团控股有限公司 | A kind of recognition methods of user, device and equipment |
| CN109446905A (en) * | 2018-09-26 | 2019-03-08 | 深圳壹账通智能科技有限公司 | Sign electronically checking method, device, computer equipment and storage medium |
| CN109583227A (en) * | 2018-10-30 | 2019-04-05 | 中国科学院信息工程研究所 | A kind of method for protecting privacy, apparatus and system |
| CN109858927A (en) * | 2019-01-16 | 2019-06-07 | 深圳壹账通智能科技有限公司 | A kind of trade company's checking method, device, computer readable storage medium and server |
| CN109934809A (en) * | 2019-03-08 | 2019-06-25 | 深慧视(深圳)科技有限公司 | A kind of paper labels character defect inspection method |
Non-Patent Citations (6)
| Title |
|---|
| 吴振华: "离线中文签名验证技术研究", 《中国博士学位论文全文数据库 信息科技辑》 * |
| 张保生等: "《中国证据法治发展报告2011》", 30 June 2013, 《中国政法大学出版社》 * |
| 曹军等: "基于多特征和神经网络的离线签名鉴别", 《2010年全国模式识别学术会议(CCPR2010)》 * |
| 李航: "《伪造数字图像盲检测技术研究》", 31 January 2016, 《吉林大学出版社》 * |
| 杨丹凤: "基于方向特征的离线签名鉴别", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 胥玉龙等: "基于深度学习的离线手写签名真伪识别方法", 《电脑知识与技术》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112215087A (en) * | 2020-09-21 | 2021-01-12 | 浙江数链科技有限公司 | Picture auditing method and device, computer equipment and storage medium |
| CN112766082A (en) * | 2020-12-30 | 2021-05-07 | 大连海事大学 | Chinese text handwriting identification method and device based on macro-micro characteristics and storage medium |
| CN112766082B (en) * | 2020-12-30 | 2024-04-23 | 大连海事大学 | Chinese text handwriting identification method and device based on macro-micro characteristics and storage medium |
| CN116630033A (en) * | 2023-07-20 | 2023-08-22 | 杭银消费金融股份有限公司 | Information auditing method, system and storage medium |
| CN116630033B (en) * | 2023-07-20 | 2023-11-07 | 杭银消费金融股份有限公司 | Information auditing method, system and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111178203B (en) | 2021-01-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109784170B (en) | Vehicle risk assessment method, device, equipment and storage medium based on image recognition | |
| CN111178203B (en) | Signature verification method and device, computer equipment and storage medium | |
| WO2019218699A1 (en) | Fraud transaction determining method and apparatus, computer device, and storage medium | |
| CN109658584B (en) | Bill information identification method and device | |
| CN110738236B (en) | Image matching method and device, computer equipment and storage medium | |
| CN111191568B (en) | Method, device, equipment and medium for identifying flip image | |
| CN109426785B (en) | Human body target identity recognition method and device | |
| CN109840524B (en) | Text type recognition method, device, equipment and storage medium | |
| CN108053545B (en) | Certificate verification method and device, server and storage medium | |
| CN112001932B (en) | Face recognition method, device, computer equipment and storage medium | |
| CN109934114B (en) | Finger vein template generation and updating algorithm and system | |
| EP2360619A1 (en) | Fast fingerprint searching method and fast fingerprint searching system | |
| CN110738222B (en) | Image matching method and device, computer equipment and storage medium | |
| CN111353491A (en) | Character direction determining method, device, equipment and storage medium | |
| CN111324802A (en) | Data auditing method and device based on identity recognition and computer equipment | |
| CN105740808A (en) | Human face identification method and device | |
| CN109670931B (en) | Loan user behavior detection method, loan user behavior detection device, loan user behavior detection equipment and loan user behavior detection storage medium | |
| CN111489262A (en) | Policy information detection method and device, computer equipment and storage medium | |
| CN112163110B (en) | Image classification method and device, electronic equipment and computer-readable storage medium | |
| CN114022926A (en) | Face recognition method, device, equipment and storage medium | |
| CN116151965B (en) | Risk feature extraction method and device, electronic equipment and storage medium | |
| Harfiya et al. | Offline signature verification based on pyramid histogram of oriented gradient features | |
| CN111444362A (en) | Malicious picture intercepting method, device, equipment and storage medium | |
| CN113378609A (en) | Method and device for identifying agent signature | |
| CN108364024B (en) | Image matching method and device, computer equipment and storage medium |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |