CN113378609B - Agent proxy signature identification method and device - Google Patents
Agent proxy signature identification method and device Download PDFInfo
- Publication number
- CN113378609B CN113378609B CN202010162874.2A CN202010162874A CN113378609B CN 113378609 B CN113378609 B CN 113378609B CN 202010162874 A CN202010162874 A CN 202010162874A CN 113378609 B CN113378609 B CN 113378609B
- Authority
- CN
- China
- Prior art keywords
- handwriting
- agent
- signature
- verified
- 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.)
- Active
Links
Classifications
-
- 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
- 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
- G06N3/045—Combinations of networks
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Collating Specific Patterns (AREA)
Abstract
The invention discloses a method and a device for identifying agent representative signatures, wherein the method comprises the steps of obtaining agent handwriting images and identifying handwriting content information contained in the agent handwriting images; classifying the agent handwriting images according to the handwriting content information and channel information of the agent, and forming an agent handwriting library according to the classification result; when receiving the signature handwriting image to be verified, identifying the content information of the signature to be verified contained in the signature handwriting image to be verified; extracting matched agent handwriting images from an agent handwriting library according to the content information of the signature to be verified and agent channel information corresponding to the signature handwriting images to be verified; training a first handwriting recognition model based on the matched agent handwriting images; and inputting the handwriting image of the signature to be verified into the first handwriting recognition model for calculation to obtain a first similarity result, and determining whether the signature to be verified is an agent code label according to the first similarity result.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a method and a device for identifying agent proxy signature.
Background
With the development of various services, the service risk is increased while the service volume and the user increase, and the phenomenon of completing the service index in an unknowing subscription manner is already generated at present. For online business or business handling in a self-service hall, prevention and control are performed by means of authentication comparison, portrait comparison, order secondary determination and the like at present; in addition to conventional identity card verification, it is also necessary to audit the customer signature at the agent level.
The auditing methods commonly used at present comprise the following two methods:
the first mode, for signature text recognition, includes: 1. judging whether a signature exists or not; 2. whether the signature is a Chinese character; 3. whether the content of the label is consistent with the name of the accepted customer or not; 4. whether the field signature is consistent with the original retained signature of the customer. The main technology is to identify the signature, and along with the evolution of handwriting identification technology, the identification rate of handwriting is improved, so that the signature content can be identified, and the judgment of whether the signature is compliant is made.
In a second way, for the identification of non-principal signatures, based on the comparison of the actual signature with the signature to be verified, the implementation of this way depends on the following two preconditions: the method has the advantages that the method is provided with a real client signature feature library for comparison, and enough real client handwriting is used as training data.
However, the inventors found in the course of implementing the present invention that: the first method can solve the problem of whether the signature is the name of the person, and ensure the consistency with the name of the person to be accepted, but the problem of whether the agent uses the code signature cannot be solved by identifying the signature text. In the second approach described above, first, for the new customer, there is no business transaction record previously, nor a signature available for comparison; second, in the case where the number of business processes is not large, there is not enough real customer handwriting as training data, and there is also a problem that the history signature of the customer is also a proxy signature, so that the above two preconditions are often difficult to satisfy.
Disclosure of Invention
The present invention has been made in view of the above problems, and it is an object of the present invention to provide a method and apparatus for identifying a proxy signature that overcomes or at least partially solves the above problems.
According to one aspect of the present invention, there is provided a method for identifying a proxy signature, comprising:
acquiring an agent handwriting image, and identifying handwriting content information contained in the agent handwriting image;
classifying the agent handwriting images according to the handwriting content information and channel information of the agent, and forming an agent handwriting library according to the classification result;
When receiving the signature handwriting image to be verified, identifying the content information of the signature to be verified contained in the signature handwriting image to be verified;
extracting matched agent handwriting images from an agent handwriting library according to the content information of the signature to be verified and agent channel information corresponding to the signature handwriting images to be verified;
training a first handwriting recognition model based on the matched agent handwriting images;
and inputting the handwriting image of the signature to be verified into the first handwriting recognition model for calculation to obtain a first similarity result, and determining whether the signature to be verified is an agent code label according to the first similarity result.
Optionally, after acquiring the agent handwriting image, the method further comprises:
preprocessing the handwriting image of the agent; wherein the pretreatment comprises one or more of the following: binarization processing, smoothing processing and text segmentation processing.
Optionally, classifying the agent handwriting image according to the handwriting content information and the channel information of the agent, and forming the agent handwriting library according to the classification result further includes:
if the word sense of the handwriting content information of the handwriting images of the agents is the same, establishing an index aiming at the word sense, and storing the handwriting images of the agents under the index of the word sense to form an agent handwriting library.
Optionally, the method further comprises:
acquiring client information according to the signature content information to be verified;
according to the client information, searching whether signature retention data of the client at other agents exists or not;
if yes, training a second handwriting recognition model according to the signature retention data;
inputting the signature handwriting image to be verified into a second handwriting recognition model for calculation to obtain a second similarity result;
and determining whether the signature handwriting to be verified is the true signature handwriting of the client according to the second similarity result.
Optionally, the first handwriting recognition model and the second handwriting recognition model are obtained based on convolutional neural network algorithm training.
Optionally, the method further comprises:
and under the condition that the signature to be verified is determined to be the agent code label according to the first similarity result, marking the signature handwriting image to be verified as an agent handwriting image sample, and incorporating the agent handwriting image sample into an agent handwriting library.
According to another aspect of the present invention, there is provided an agent proxy signature recognition apparatus including:
the acquisition module is suitable for acquiring the handwriting image of the agent;
the first identification module is suitable for identifying handwriting content information contained in the agent handwriting image;
The classifying and warehousing module is suitable for classifying the agent handwriting images according to the handwriting content information and the channel information of the agent, and forming an agent handwriting library according to the classifying result;
the second identification module is suitable for identifying the content information of the signature to be verified, which is contained in the signature handwriting image to be verified, when the signature handwriting image to be verified is received;
the matching module is suitable for extracting matched agent handwriting images from the agent handwriting library according to the content information of the signature to be verified and agent channel information corresponding to the handwriting images of the signature to be verified;
the model training module is suitable for training a first handwriting recognition model based on the matched agent handwriting images;
the computing module is suitable for inputting the signature handwriting image to be verified into the first handwriting recognition model for computing to obtain a first similarity result;
and the judging module is suitable for determining whether the signature to be verified is an agent code sign according to the first similarity result.
Optionally, the apparatus further comprises:
the preprocessing module is suitable for preprocessing the agent handwriting image after acquiring the agent handwriting image; wherein the pretreatment comprises one or more of the following: binarization processing, smoothing processing and text segmentation processing.
Optionally, the categorizing and warehousing module is further adapted to: if the word sense of the handwriting content information of the handwriting images of the agents is the same, establishing an index aiming at the word sense, and storing the handwriting images of the agents under the index of the word sense to form an agent handwriting library.
Optionally, the apparatus further comprises:
the client information module is suitable for acquiring client information according to the signature content information to be verified;
the data retrieval module is suitable for retrieving whether signature retention data of the client at other agents exists according to the client information;
the model training module is further adapted to: training a second handwriting recognition model according to the signature retention data;
the computing module is further adapted to: inputting the signature handwriting image to be verified into a second handwriting recognition model for calculation to obtain a second similarity result;
the judgment module is further adapted to: and determining whether the signature handwriting to be verified is the true signature handwriting of the client according to the second similarity result.
Optionally, the first handwriting recognition model and the second handwriting recognition model are obtained based on convolutional neural network algorithm training.
Optionally, the apparatus further comprises: marking and warehousing modules: and marking the signature handwriting image to be verified as an agent handwriting image sample and incorporating the agent handwriting image sample into an agent handwriting library under the condition that the signature to be verified is determined to be the agent code according to the first similarity result. .
According to yet another aspect of the present invention, there is provided a computing device comprising: the device comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface complete communication with each other through the communication bus;
the memory is used for storing at least one executable instruction, and the executable instruction enables the processor to execute the operation corresponding to the agent representative signature identification method.
According to still another aspect of the present invention, there is provided a computer storage medium having stored therein at least one executable instruction for causing a processor to perform operations corresponding to the above-described agent proxy signature identification method.
The invention relates to a method and a device for identifying agent representative signatures, wherein the method comprises the following steps: acquiring an agent handwriting image, and identifying handwriting content information contained in the agent handwriting image; classifying the agent handwriting images according to the handwriting content information and channel information of the agent, and forming an agent handwriting library according to the classification result; when receiving the signature handwriting image to be verified, identifying the content information of the signature to be verified contained in the signature handwriting image to be verified; extracting matched agent handwriting images from an agent handwriting library according to the content information of the signature to be verified and agent channel information corresponding to the signature handwriting images to be verified; training a first handwriting recognition model based on the matched agent handwriting images; and inputting the handwriting image of the signature to be verified into the first handwriting recognition model for calculation to obtain a first similarity result, and determining whether the signature to be verified is an agent code label according to the first similarity result. The agent code signature recognition method provided by the invention is opposite to the traditional method, the traditional method relies on the historical signature of the client as a sample to recognize the similarity between the signature handwriting to be verified and the historical signature handwriting of the client, and the method in the embodiment relies on the signature handwriting of the agent as the sample to construct a recognition model, and the recognition model is used for calculating the similarity between the signature handwriting to be verified and the handwriting of the agent, so that whether the signature handwriting to be verified is the agent code signature is recognized. Compared with the traditional mode, the method reduces the data collection difficulty, improves the technical application range, and is helpful for preventing the risk of unknowing custom-made business of clients.
The foregoing description is only an overview of the present invention, and is intended to be implemented in accordance with the teachings of the present invention in order that the same may be more clearly understood and to make the same and other objects, features and advantages of the present invention more readily apparent.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to designate like parts throughout the figures. In the drawings:
FIG. 1 illustrates a flow chart of an embodiment of a method of identifying a representative signature of an agent of the present invention;
FIG. 2 is a flow chart illustrating another embodiment of a method of identifying a proxy signature in accordance with the present invention;
FIG. 3 shows a schematic diagram of a convolutional network in an embodiment of the invention;
fig. 4 is a schematic structural diagram of an agent proxy signature recognition device according to an embodiment of the present invention;
FIG. 5 illustrates a schematic diagram of a computing device provided by an embodiment of the present invention.
Detailed Description
Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
The theory basis of handwriting recognition is introduced first, handwriting habit is a dynamic shaping expression, and a person with fixed handwriting habit is controlled by personal consciousness in the whole process of handwriting, but specific handwriting action is automatically realized by writing habit. This automated repeated reproduction keeps the individual's signature or handwriting relatively stable. The handwriting signature is inconsistent to a certain extent even if the same person is in different time periods and different writing environments, so that judgment cannot be made by means of certain fixed features, namely the handwriting signature cannot be identified by relying on manually preset handwriting feature points, and the handwriting signature is realized in a mode of combining key feature points with integral image features from the aspect of integrity.
In an actual business scenario, agents usually have territory, and business handling staff of the same agent are relatively fixed, which leads to that personnel having a signature behavior of a certain agent, possibly the same as the personnel doing business handling, if a signature is to be identified, whether the signature is signed by the agent business staff or not can be detected, so that the identification of the signature is converted from the authentication with a real client to the authentication with historical handwriting data of the agent, and both the authentication data and the training data can be ensured.
In the embodiment of the invention, the automatic identification is realized mainly by using a convolutional neural network, and the artificial neural network is a network system which is established by a large number of processing units through wide interconnection and manual mode and is called as an artificial neuron in order to simulate the structure and the function of a human brain nervous system. The artificial neural network is actually a directed network structure which uses artificial neurons as nodes and is connected by directed weights. The positive and negative of the weights correspond to excitability or inhibitivity of the synaptic connection. A feature of a fully connected network is that all incoming information will have an impact on the subsequent training, but this is not meaningful for image data.
Convolutional networks have two important concepts: 1. a local receptive field; 2. weight sharing. The local receptive field is characterized in that the relevance of points far away in one picture is not high, so that a part of parameters are saved from a full-connection mode, only local operation is performed, and the number of operation parameters is reduced. For the "weight sharing", the calculation mode is the same as that of a single image, namely, the feature extraction mode is the same; the method is characterized in that when the convolution is continuously performed on the input image, the data in all filters are calculated in the same way, namely parameter sharing is realized, and thus parameter values can be reduced as well.
Convolutional neural networks (CNN, convolutional Neural Network) are a class of models of deep neural networks and are guided by the ideas of deep learning architecture. It is particularly suited to be built to process and identify images. The structure of the convolutional neural network includes: input layer, convolution layer, downsampling layer (pooling layer), full connection layer.
Input layer: the input to the whole network, typically a matrix of pixels of an image, is seen in the above figures as a stereoscopic structure, since the typical image has a concept of depth, just like the RGB color image we typically see, in a form of a x b x c, where the first two dimensions specify the length and width of the image, the third dimension is depth, the depth of the color RGB is 3, and the depth of the black and white image in the MNIST we have seen before is 1.
Convolution layer (convolutional layer): and extracting the features from the bottom layer to the high layer from the image by utilizing convolution operation, and ensuring the local relevance and the spatial invariance of the picture. Filter: it is understood as a neuron that implements a defined convolution kernel. Step size Stride: for an area, sliding is performed after the calculation is completed, and the moving distance during sliding is the Stride. Filling Padding: considering the convolution calculation process, for pixels of the image near the middle, we can see that they are calculated "overlapping" many times, in contrast to pixels at the edges being calculated only once, in order to ensure that the number of calculations is relatively balanced, some values are filled in the edges to ensure that they are also calculated many times.
Depth: the depth here does not refer to the image, but refers to the number of neurons (filters) in a certain layer, at this time, the output is not a Feature Map, but a three-dimensional structure composed of several Feature maps, one Filter can process the input image into a Feature Map, the features of different Filter emphasis processes are different, and we want to obtain different Feature maps, multiple filters are needed to set up, so that each Filter process obtains a Feature Map, multiple filters obtain multiple Feature maps, and the Feature maps are stacked together to form the three-dimensional structure of the output, so it can be seen that the number of filters is the same as the number of Feature maps, and the number is the depth.
Downsampling layer (pooling layer): a downsampling operation is performed. The unimportant high-frequency information is filtered out by convolution output characteristic graphs, namely local maximum value (max-pooling) and average value (avg-pooling). The pooling layer has the following functions: 1. feature Map is extracted again, and the Feature Map is also an operation for reducing the data volume; 2. the more abstract features are obtained, the overfitting is prevented, and the generalization is improved; 3. with this process, small variations in the input are more tolerated, i.e. if the data has some noise, the effect of the noise is reduced to some extent through this feature extraction process.
Full tie layer (fully connected layer): the neurons of each layer in the network are all connected. The convolutional layer and the fully-connected layer are usually followed by a nonlinear change processing layer to enhance the expression capability of the network to the characteristics.
The image recognition model using convolutional neural network mainly consists of two important characteristics of the image: local correlation and spatial invariance. By repeatedly applying convolution and operation, the characteristics of the image can be reflected well. When the convolution layer extracts the features, the input and output data have relevance, the relative relation among the features is reserved, the features cannot be extracted in an isolated way in a linear relation, and the nonlinear combination of the internal information of the image is reflected in the local relevance. Meanwhile, in the feature mapping process, the number of parameters is greatly reduced due to the fact that weight values are shared among neurons of the convolution network, complexity is reduced, and information effectiveness is greatly improved.
FIG. 1 shows a flow chart of an embodiment of a method of identifying a representative signature of an agent of the present invention, as shown in FIG. 1, comprising the steps of:
step S101, acquiring a handwriting image of the agent, and recognizing handwriting content information contained in the handwriting image of the agent.
The method of the embodiment of the invention needs to archive and input the handwriting of the agent, and uniformly collects the handwriting fonts of the business documents which are all related to the business document transacting in the agent channel so as to form an agent handwriting library.
Firstly, handwriting images of all agents are obtained, handwriting comprises but is not limited to signature handwriting, and character contents such as Chinese characters or English contained in the agent handwriting images are identified by utilizing an identification technology. The method is not limited to the mode of acquiring the agent handwriting image, and the agent handwriting image can be acquired by scanning or photographing.
Step S102, classifying the agent handwriting images according to the handwriting content information and the channel information of the agent, and forming an agent handwriting library according to the classification result.
Wherein the channel information of the agent may refer to the agent number. And classifying the agent handwriting images of the same agent and consistent handwriting content information into one class according to the handwriting content information and the corresponding agent channel information contained in the agent handwriting images, and warehousing and storing. In the specific implementation, the first category may be established according to the agent number, and the second category subordinate to the first category may be established according to the handwritten content information under the first category, which is not limited by the invention.
Step S103, when receiving the signature handwriting image to be verified, identifying the content information of the signature to be verified contained in the signature handwriting image to be verified.
Step S101-step S102 are processes of creating a proxy handwriting library, and step S103-step S106 are processes of identifying whether the signature to be verified is a proxy code. Firstly, receiving a signature handwriting image to be verified, and identifying character contents such as Chinese or English contained in the signature handwriting image to be verified by utilizing an identification technology to obtain signature content information to be verified. The signature handwriting image to be verified, namely the electronic signature document, can be obtained by scanning or shooting a paper signature document, and some documents are electronic documents.
Step S104, extracting matched agent handwriting images from the agent handwriting library according to the content information of the signature to be verified and the agent channel information corresponding to the handwriting images of the signature to be verified.
And then, according to the content information of the signature to be verified and the agent number corresponding to the agent verification signature handwriting image, searching in an agent handwriting library to obtain a matched agent handwriting image.
Step S105, training a first handwriting recognition model based on the matched agent handwriting images.
And taking the extracted agent handwriting image as a training sample to train the first handwriting recognition model, for example, the first handwriting recognition model can be trained through a convolutional neural network algorithm.
And S106, inputting the handwriting image of the signature to be verified into a first handwriting recognition model for calculation, obtaining a first similarity result, and determining whether the signature to be verified is an agent code label according to the first similarity result.
And finally, inputting the signature handwriting image to be verified into a first handwriting recognition model for calculation, outputting a first similarity result, and judging that the signature to be verified is an agent code if the first similarity exceeds a preset threshold value, which indicates that the similarity between the signature handwriting to be verified and the agent code is higher. Otherwise, if the first similarity does not exceed the preset threshold, indicating that the signature handwriting to be verified is dissimilar to the agent handwriting, determining that the signature to be verified is not an agent proxy signature.
The embodiment of the invention combines the recognition comparison of Chinese characters and the verification of specific characteristics in a specific agent handwriting recognition scene. Firstly, recognizing characters consistent with a signature to be verified from the handwriting content of an agent, and then utilizing a neural network to verify whether the characters are similar in integrity and local characteristics, so as to determine whether the signature is signed by an agent service personnel.
Therefore, the agent code signature recognition method provided by the embodiment of the invention is opposite to the traditional method, the traditional method relies on the historical signature of the client as a sample to recognize the similarity between the signature handwriting to be verified and the historical signature handwriting of the client, and the method in the embodiment relies on the signature handwriting of the agent as a sample to construct a recognition model, and the recognition model is used for calculating the similarity between the signature handwriting to be verified and the agent handwriting, so that whether the signature handwriting to be verified is the agent code signature is recognized. Compared with the traditional mode, the method reduces the data collection difficulty, improves the technical application range, and is helpful for preventing the risk of unknowing custom-made business of clients.
FIG. 2 shows a flow chart of another embodiment of the method of identifying a proxy signature of the present invention, as shown in FIG. 2, comprising the steps of:
step S201, acquiring a handwriting image of the agent, and recognizing handwriting content information contained in the handwriting image of the agent.
Firstly, handwriting images of all agents are obtained, and Chinese characters contained in the handwriting images of the agents are identified by utilizing an identification technology.
Optionally, after the agent handwriting image is obtained, preprocessing is first performed to form a clear, independent single chinese character image that is convenient to compare and maintain. In this case, by recognizing the handwriting image after the preprocessing, handwritten content information contained in the handwriting image is obtained.
Wherein, the pretreatment mainly comprises the following three treatments:
binarization processing of handwriting images: the signed Chinese character image is processed into digital information of (0, 1), namely the gray value of the pixel point on the image is set to 0 or 255, namely the whole image is obviously black-white. The binarized image is obtained, so that the aggregate property of the image is only related to the position of the point with the pixel value of 0 or 255 when the image is further processed, the multi-level value of the pixel is not related any more, the processing is simple, and the processing and compression amount of data are small.
Smoothing of handwriting images: in order to reduce the edge noise of handwriting, the binarized signature image needs to be subjected to smooth noise reduction. The smoothing filtering in the spatial domain is generally performed by a simple averaging method, i.e. the average brightness value of adjacent pixel points is obtained. The size of the neighborhood is directly related to the smoothing effect, the larger the neighborhood is, the better the smoothing effect is, but the larger the neighborhood is, the larger the smoothing loss of edge information is, so that the output image becomes fuzzy, and the size of the neighborhood needs to be reasonably selected. In the embodiment of the invention, in order to ensure the definition of the handwriting after smoothing, the edges of the handwriting are required to be protected from blurring, so that a median filtering method is selected for carrying out image smoothing noise reduction treatment, namely, each pixel of an image has a good filtering effect on impulse noise by replacing median filtering with median of pixels in a neighborhood (square area taking the current pixel as the center), and particularly, the edges of signals can be protected from blurring while noise is filtered, but textures in a uniform medium area can be washed.
Chinese character segmentation: in order to improve the accuracy of verification, in the embodiment of the invention, matching verification is performed according to a single Chinese character. After the two steps are processed, the handwriting content is still in the form of sentences or phrases, and the requirement of matching verification according to single Chinese characters is not met, so that the handwriting content needs to be segmented, and finally is stored in the form of independent Chinese characters. The Chinese character segmentation is mainly completed by judging the space left blank width and the connecting handwriting of the Chinese characters. The width judgment of the gap margin is mainly based on the identification of the transverse width, because the left-right spacing is a main mode for identifying single Chinese characters structurally. The specific width judgment is obtained through self-learning of the historical handwriting, and can also be realized through a mode of setting artificial features.
Step S202, classifying the agent handwriting images according to the handwriting content information and the channel information of the agent, and forming an agent handwriting library according to the classification result.
In particular, a first category may be established according to the agent number, and a second category subordinate to the first category may be established according to the handwritten content information under the first category. For example, a plurality of agent handwriting images of the agent a are obtained, handwriting content information of the agent handwriting images is identified to include "Zhang san", "Li Si" and the like, the agent numbers are "agent a", a first category "agent a" and a second category "Zhang san", "Li Si" and the like are established, agent handwriting images of all agent a and text content of Zhang san are stored under category "agent a-Zhang san", and agent handwriting images of all agent a and text content of Li Si are stored under category "agent-Li Si". Of course, the manner of classifying the invention is not limited to this, and the invention can be classified into one type of agent handwriting images with identical handwriting content information for the same agent.
In an alternative embodiment, the agent handwriting images may be further classified in combination with time information and/or service information to form an agent handwriting library.
In an alternative embodiment, the agent handwriting images may also be categorized in combination with word senses to form an agent handwriting library. Specifically, if the word senses of the handwriting content information of the handwriting images of the plurality of agents are the same, an index is established for the word senses, and the handwriting images of the plurality of agents are stored under the index of the word senses to form an agent handwriting library. For example, in a plurality of electronic business documents, when different handwriting but the sense of the handwriting is the same, an index is established according to the sense of the handwriting, and a plurality of pens with all the sense of the handwriting are stored under the index. In the subsequent process, the handwriting to be compared can be quickly searched according to word sense.
Step S203, when receiving the signature handwriting image to be verified, identifying the content information of the signature to be verified contained in the signature handwriting image to be verified.
And receiving the signature handwriting image to be verified, and recognizing Chinese contained in the signature handwriting image to be verified by utilizing a character recognition technology to obtain the content information of the signature to be verified. The signature handwriting image to be verified, namely the electronic signature document, can be obtained by scanning or shooting a paper signature document, and some documents are electronic documents. The step is to identify the handwriting in the handwriting image of the signature to be verified and judge which Chinese characters the signature is.
Step S204, extracting matched agent handwriting images from the agent handwriting library according to the content information of the signature to be verified and the agent channel information corresponding to the handwriting images of the signature to be verified.
And then, according to the content information of the signature to be verified and the agent number corresponding to the handwriting image of the signature to be verified, searching in an agent handwriting library to obtain a matched agent handwriting image. Along with the above example, if the signature handwriting image to be verified of the agent a is received, the signature content information to be verified is identified as "li si", and the agent is encoded as "agent a", then the agent handwriting image under category "agent a-li si" is extracted from the agent handwriting library.
Step S205, training a first handwriting recognition model based on the matched agent handwriting images.
Training a first handwriting recognition model based on the extracted agent handwriting image. Dividing the matched agent handwriting image into training data and verification data, enabling the training data to enter a convolution network for iterative training, extracting and identifying image features in batches through training, detecting the verification data in a large quantity after complete iteration is obtained, and continuously reducing the identification error rate.
Fig. 3 shows a schematic diagram of a convolutional network in an embodiment of the present invention, where a neural network including two layers of convolutions is formed, a convolutional neural network algorithm extracts an input handwriting image, a certain number of feature maps (such as angles, textures, gray scales, etc.) are obtained in an A1 convolutional layer, then sampling operation calculation (such as weighting values, offset adding, etc.) is performed on the A1 feature maps, feature mapping information of A1 is obtained through a function, a sampling layer B1 is obtained, and then convolutions are repeated once again, namely A2 and B2, the obtained data is built in a full-connection layer and input to a classifier, and a conclusion is obtained through recognition, thereby ending an iterative training. And continuously adjusting the process parameters according to the output result to perform continuous iteration. 4 hidden layers are arranged, and learning and feature extraction are performed on the image by halving the hidden layers layer by layer in sequence.
Step S206, inputting the handwriting image of the signature to be verified into the first handwriting recognition model for calculation, obtaining a first similarity result, and determining whether the signature to be verified is an agent code label according to the first similarity result.
Inputting the signature handwriting image to be verified into a first handwriting recognition model for calculation, outputting a first similarity result, if the first similarity exceeds a preset threshold value, indicating that the similarity between the signature handwriting to be verified and the agent handwriting is high, determining that the signature to be verified is an agent code, marking the signature handwriting image to be verified as an agent handwriting image sample, and incorporating the agent handwriting image sample into an agent handwriting library. Otherwise, if the first similarity does not exceed the preset threshold, indicating that the signature handwriting to be verified is dissimilar to the agent handwriting, determining that the signature to be verified is not an agent proxy signature.
Step S207, obtaining client information according to the content information of the signature to be verified, and searching whether signature retention data of the client at other agents exists according to the client information.
Then, the client information is obtained according to the signature content information to be verified, in practical application, the client signs the bill with the name of the client, and the identified signature content information to be verified, namely the name of the client, can obtain the client identity information and the like according to the name of the client. And then searching the past business acceptance record and paperless authentication record of the client according to the client information, and judging whether the records for transacting business in other agents and signature retention exist or not, wherein the signature retention is the historical real handwriting image of the client without considering the condition of the other agents. For example, if the signature handwriting image to be verified belongs to agent a, then there is customer signature retention data associated with the customer information in the business records of other agents than agent a.
Step S208, if the second handwriting recognition model exists, training the second handwriting recognition model according to the signature retention data.
If signature retention data of the client at other agents exists, extracting the part of the signature retention data, and training a second handwriting recognition model according to the signature retention data.
Step S209, inputting the signature handwriting image to be verified to a second handwriting recognition model for calculation to obtain a second similarity result, and determining whether the signature handwriting to be verified is the customer real signature handwriting according to the second similarity result.
Inputting the signature handwriting image to be verified into a second handwriting recognition model for calculation, outputting a second similarity result, and if the second similarity exceeds a preset threshold value, indicating that the similarity between the signature handwriting to be verified and the client historical signature handwriting is higher, determining that the signature to be verified is the client real handwriting and is not the agent code signing handwriting. Otherwise, if the second similarity does not exceed the preset threshold, the to-be-verified signature handwriting is not similar to the client historical signature handwriting, and the agent signing risk is determined to exist.
When the method is implemented, a recognition mode can be selected according to the specific retrieved information, and if a matched agent handwriting image is retrieved, a model is built based on the matched agent handwriting image and used for recognizing whether the signature handwriting to be verified is similar to the agent handwriting; if the reserved data of the client in other agents is retrieved, a model is built based on the reserved data and used for identifying whether the signature handwriting to be verified is the real handwriting of the client; if both the signature handwriting and the code handwriting are searched, two recognition modes are adopted at the same time, the results of the two recognition modes are combined, and whether the signature handwriting to be verified is an agent code signature or not is judged.
In summary, in the embodiment of the present invention, on the one hand, a method for identifying a proxy to be signed based on a signature script of a proxy is provided, and under the condition that a matched proxy handwriting script can be retrieved, an identification model is constructed based on the proxy handwriting script as a sample, and a similarity relationship between the signature script to be verified and the proxy handwriting script is calculated by using the identification model, so that whether the signature script to be verified is a proxy code is identified. On the other hand, the method for identifying the signature handwriting to be verified based on the client reserved handwriting data is provided, under the condition that the client reserved handwriting can be retrieved, an identification model is constructed based on the client reserved handwriting as a sample, and the model is used for identifying the similarity relationship between the signature handwriting to be verified and the historical real handwriting of the client, so that whether the signature handwriting to be verified is the real handwriting of the client is identified. The two modes are combined, so that the identification accuracy of the agent signature can be improved, corresponding identification modes can be selected according to different data retrieval results, the flexibility is higher, the agent with the agent signature behavior can be effectively identified, and the risk of unknowingly customizing the service by the client can be prevented.
Fig. 4 shows a schematic structural diagram of an embodiment of the identification device of the agent representative signature of the present invention. As shown in fig. 4, the apparatus includes:
an acquisition module 41 adapted to acquire an agent handwriting image;
a first recognition module 42 adapted to recognize handwritten content information contained in the agent handwriting image;
the classifying and warehousing module 43 is suitable for classifying the agent handwriting images according to the handwriting content information and the channel information of the agent, and forming an agent handwriting library according to the classifying result;
the second identifying module 44 is adapted to identify the to-be-verified signature content information contained in the to-be-verified signature handwriting image when the to-be-verified signature handwriting image is received;
the matching module 45 is suitable for extracting matched agent handwriting images from the agent handwriting library according to the content information of the signature to be verified and the agent channel information corresponding to the signature handwriting images to be verified;
a model training module 46 adapted to train a first handwriting recognition model based on the matched agent handwriting images;
a calculation module 47, adapted to input the signature handwriting image to be verified into the first handwriting recognition model for calculation, so as to obtain a first similarity result;
The determining module 48 is adapted to determine whether the signature to be verified is a proxy signature according to the first similarity result.
Optionally, the apparatus further comprises:
the preprocessing module is suitable for preprocessing the agent handwriting image after acquiring the agent handwriting image; wherein the pretreatment comprises one or more of the following: binarization processing, smoothing processing and text segmentation processing.
Optionally, the categorization binning module 43 is further adapted to: if the word sense of the handwriting content information of the handwriting images of the agents is the same, establishing an index aiming at the word sense, and storing the handwriting images of the agents under the index of the word sense to form an agent handwriting library.
Optionally, the apparatus further comprises:
the client information module is suitable for acquiring client information according to the signature content information to be verified;
the data retrieval module is suitable for retrieving whether signature retention data of the client at other agents exists according to the client information;
model training module 46 is further adapted to: training a second handwriting recognition model according to the signature retention data;
the calculation module 47 is further adapted to: inputting the signature handwriting image to be verified into a second handwriting recognition model for calculation to obtain a second similarity result;
The judgment module 48 is further adapted to: and determining whether the signature handwriting to be verified is the true signature handwriting of the client according to the second similarity result.
Optionally, the first handwriting recognition model and the second handwriting recognition model are obtained based on convolutional neural network algorithm training.
Optionally, the apparatus further comprises:
marking and warehousing modules: and marking the signature handwriting image to be verified as an agent handwriting image sample and incorporating the agent handwriting image sample into an agent handwriting library under the condition that the signature to be verified is determined to be the agent code according to the first similarity result.
Embodiments of the present invention provide a non-volatile computer storage medium storing at least one executable instruction that is capable of performing the method for identifying a proxy signature in any of the method embodiments described above.
The executable instructions may be particularly useful for causing a processor to:
acquiring an agent handwriting image, and identifying handwriting content information contained in the agent handwriting image;
classifying the agent handwriting images according to the handwriting content information and channel information of the agent, and forming an agent handwriting library according to the classification result;
When receiving the signature handwriting image to be verified, identifying the content information of the signature to be verified contained in the signature handwriting image to be verified;
extracting matched agent handwriting images from an agent handwriting library according to the content information of the signature to be verified and agent channel information corresponding to the signature handwriting images to be verified;
training a first handwriting recognition model based on the matched agent handwriting images;
and inputting the handwriting image of the signature to be verified into the first handwriting recognition model for calculation to obtain a first similarity result, and determining whether the signature to be verified is an agent code label according to the first similarity result.
In one alternative, the executable instructions cause the processor to:
after the agent handwriting image is obtained, preprocessing the agent handwriting image; wherein the pretreatment comprises one or more of the following: binarization processing, smoothing processing and text segmentation processing.
In one alternative, the executable instructions cause the processor to:
if the word sense of the handwriting content information of the handwriting images of the agents is the same, establishing an index aiming at the word sense, and storing the handwriting images of the agents under the index of the word sense to form an agent handwriting library.
In one alternative, the executable instructions cause the processor to:
acquiring client information according to the signature content information to be verified;
according to the client information, searching whether signature retention data of the client at other agents exists or not;
if yes, training a second handwriting recognition model according to the signature retention data;
inputting the signature handwriting image to be verified into a second handwriting recognition model for calculation to obtain a second similarity result;
and determining whether the signature handwriting to be verified is the true signature handwriting of the client according to the second similarity result.
In an alternative manner, the first handwriting recognition model and the second handwriting recognition model are trained based on a convolutional neural network algorithm.
In one alternative, the executable instructions cause the processor to:
and under the condition that the signature to be verified is determined to be the agent code label according to the first similarity result, marking the signature handwriting image to be verified as an agent handwriting image sample, and incorporating the agent handwriting image sample into an agent handwriting library.
FIG. 5 illustrates a schematic diagram of an embodiment of a computing device of the present invention, and the embodiments of the present invention are not limited to a particular implementation of the computing device.
As shown in fig. 5, the computing device may include: a processor 502, a communication interface (Communications Interface) 504, a memory 506, and a communication bus 508.
Wherein: processor 502, communication interface 504, and memory 506 communicate with each other via communication bus 508. A communication interface 504 for communicating with network elements of other devices, such as clients or other servers. The processor 502 is configured to execute the program 510, and may specifically perform relevant steps in the above-described embodiment of a method for identifying an agent proxy signature for a computing device.
In particular, program 510 may include program code including computer-operating instructions.
The processor 502 may be a central processing unit CPU, or a specific integrated circuit ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement embodiments of the present invention. The one or more processors included by the computing device may be the same type of processor, such as one or more CPUs; but may also be different types of processors such as one or more CPUs and one or more ASICs.
A memory 506 for storing a program 510. Memory 506 may comprise high-speed RAM memory or may also include non-volatile memory (non-volatile memory), such as at least one disk memory.
The program 510 may be specifically operable to cause the processor 502 to:
acquiring an agent handwriting image, and identifying handwriting content information contained in the agent handwriting image;
classifying the agent handwriting images according to the handwriting content information and channel information of the agent, and forming an agent handwriting library according to the classification result;
when receiving the signature handwriting image to be verified, identifying the content information of the signature to be verified contained in the signature handwriting image to be verified;
extracting matched agent handwriting images from an agent handwriting library according to the content information of the signature to be verified and agent channel information corresponding to the signature handwriting images to be verified;
training a first handwriting recognition model based on the matched agent handwriting images;
and inputting the handwriting image of the signature to be verified into the first handwriting recognition model for calculation to obtain a first similarity result, and determining whether the signature to be verified is an agent code label according to the first similarity result.
In an alternative, the program 510 causes the processor 502 to:
preprocessing the handwriting image of the agent; wherein the pretreatment comprises one or more of the following: binarization processing, smoothing processing and text segmentation processing.
In an alternative, the program 510 causes the processor 502 to:
if the word sense of the handwriting content information of the handwriting images of the agents is the same, establishing an index aiming at the word sense, and storing the handwriting images of the agents under the index of the word sense to form an agent handwriting library.
In an alternative, the program 510 causes the processor 502 to:
acquiring client information according to the signature content information to be verified;
according to the client information, searching whether signature retention data of the client at other agents exists or not;
if yes, training a second handwriting recognition model according to the signature retention data;
inputting the signature handwriting image to be verified into a second handwriting recognition model for calculation to obtain a second similarity result;
and determining whether the signature handwriting to be verified is the true signature handwriting of the client according to the second similarity result.
In an alternative manner, the first handwriting recognition model and the second handwriting recognition model are trained based on a convolutional neural network algorithm.
In an alternative, the program 510 causes the processor 502 to:
And under the condition that the signature to be verified is determined to be the agent code label according to the first similarity result, marking the signature handwriting image to be verified as an agent handwriting image sample, and incorporating the agent handwriting image sample into an agent handwriting library.
The algorithms or displays presented herein are not inherently related to any particular computer, virtual system, or other apparatus. Various general-purpose systems may also be used with the teachings herein. The required structure for a construction of such a system is apparent from the description above. In addition, embodiments of the present invention are not directed to any particular programming language. It will be appreciated that the teachings of the present invention described herein may be implemented in a variety of programming languages, and the above description of specific languages is provided for disclosure of enablement and best mode of the present invention.
In the description provided herein, numerous specific details are set forth. However, it is understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures and techniques have not been shown in detail in order not to obscure an understanding of this description.
Similarly, it should be appreciated that in the above description of exemplary embodiments of the invention, various features of the embodiments of the invention are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure and aiding in the understanding of one or more of the various inventive aspects. However, the disclosed method should not be construed as reflecting the intention that: i.e., the claimed invention requires more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects lie in less than all features of a single foregoing disclosed embodiment. Thus, the claims following the detailed description are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate embodiment of this invention.
Those skilled in the art will appreciate that the modules in the apparatus of the embodiments may be adaptively changed and disposed in one or more apparatuses different from the embodiments. The modules or units or components of the embodiments may be combined into one module or unit or component and, furthermore, they may be divided into a plurality of sub-modules or sub-units or sub-components. Any combination of all features disclosed in this specification (including any accompanying claims, abstract and drawings), and all of the processes or units of any method or apparatus so disclosed, may be used in combination, except insofar as at least some of such features and/or processes or units are mutually exclusive. Each feature disclosed in this specification (including any accompanying claims, abstract and drawings), may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise.
Furthermore, those skilled in the art will appreciate that while some embodiments herein include some features but not others included in other embodiments, combinations of features of different embodiments are meant to be within the scope of the invention and form different embodiments. For example, in the following claims, any of the claimed embodiments can be used in any combination.
Various component embodiments of the invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that some or all of the functionality of some or all of the components according to embodiments of the present invention may be implemented in practice using a microprocessor or Digital Signal Processor (DSP). The present invention can also be implemented as an apparatus or device program (e.g., a computer program and a computer program product) for performing a portion or all of the methods described herein. Such a program embodying the present invention may be stored on a computer readable medium, or may have the form of one or more signals. Such signals may be downloaded from an internet website, provided on a carrier signal, or provided in any other form.
It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The use of the words first, second, third, etc. do not denote any order. These words may be interpreted as names. The steps in the above embodiments should not be construed as limiting the order of execution unless specifically stated.
Claims (10)
1. A method of identifying a proxy signature, comprising:
acquiring an agent handwriting image, and identifying handwriting content information contained in the agent handwriting image;
classifying the agent handwriting images according to the handwriting content information and channel information of the agent, and forming an agent handwriting library according to classification results;
when receiving a signature handwriting image to be verified, identifying the content information of the signature to be verified contained in the signature handwriting image to be verified;
extracting matched agent handwriting images from the agent handwriting library according to the content information of the signature to be verified and agent channel information corresponding to the handwriting images of the signature to be verified;
training a first handwriting recognition model based on the matched agent handwriting images;
and inputting the handwriting image of the signature to be verified into the first handwriting recognition model for calculation to obtain a first similarity result, and determining whether the signature to be verified is an agent code according to the first similarity result.
2. The method of claim 1, wherein after the acquiring the agent handwriting image, the method further comprises:
Preprocessing the handwriting image of the agent; wherein the pretreatment comprises one or more of the following: binarization processing, smoothing processing and text segmentation processing.
3. The method of claim 1, wherein classifying the agent handwriting image according to the handwriting content information and channel information of the agent, and forming an agent handwriting library according to the classification result further comprises:
if the word sense of the handwriting content information of the handwriting images of the agents is the same, establishing an index aiming at the word sense, and storing the handwriting images of the agents under the index of the word sense to form an agent handwriting library.
4. The method according to claim 1, wherein the method further comprises:
acquiring client information according to the signature content information to be verified;
retrieving, based on the customer information, whether there is signature retention data for the customer at other agents;
if yes, training a second handwriting recognition model according to the signature retention data;
inputting the signature handwriting image to be verified to the second handwriting recognition model for calculation to obtain a second similarity result;
And determining whether the signature handwriting to be verified is the true signature handwriting of the client according to the second similarity result.
5. A method as claimed in claim 1 or 4, wherein the first handwriting recognition model and the second handwriting recognition model are trained based on a convolutional neural network algorithm.
6. The method according to claim 1, wherein the method further comprises:
and under the condition that the signature to be verified is determined to be the agent code label according to the first similarity result, marking the signature handwriting image to be verified as an agent handwriting image sample, and incorporating the agent handwriting image sample into an agent handwriting library.
7. An agent proxy signature recognition device, comprising:
the acquisition module is suitable for acquiring the handwriting image of the agent;
the first identification module is suitable for identifying handwriting content information contained in the agent handwriting image;
the classifying and warehousing module is suitable for classifying the agent handwriting images according to the handwriting content information and channel information of the agent, and forming an agent handwriting library according to the classifying result;
the second identification module is suitable for identifying the content information of the signature to be verified, which is contained in the handwriting image of the signature to be verified, when the handwriting image of the signature to be verified is received;
The matching module is suitable for extracting matched agent handwriting images from the agent handwriting library according to the content information of the signature to be verified and agent channel information corresponding to the signature handwriting images to be verified;
the model training module is suitable for training a first handwriting recognition model based on the matched agent handwriting images;
the computing module is suitable for inputting the signature handwriting image to be verified into the first handwriting recognition model for computing to obtain a first similarity result;
and the judging module is suitable for determining whether the signature to be verified is an agent code sign according to the first similarity result.
8. The apparatus of claim 7, wherein the apparatus further comprises:
the preprocessing module is suitable for preprocessing the agent handwriting image after acquiring the agent handwriting image; wherein the pretreatment comprises one or more of the following: binarization processing, smoothing processing and text segmentation processing.
9. A computing device, comprising: the device comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface complete communication with each other through the communication bus;
The memory is configured to store at least one executable instruction that causes the processor to perform operations corresponding to the method for identifying an agent proxy signature as claimed in any one of claims 1 to 6.
10. A computer storage medium having stored therein at least one executable instruction for causing a processor to perform operations corresponding to the method of identifying an agent proxy signature as claimed in any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010162874.2A CN113378609B (en) | 2020-03-10 | 2020-03-10 | Agent proxy signature identification method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010162874.2A CN113378609B (en) | 2020-03-10 | 2020-03-10 | Agent proxy signature identification method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113378609A CN113378609A (en) | 2021-09-10 |
| CN113378609B true CN113378609B (en) | 2023-07-21 |
Family
ID=77568851
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010162874.2A Active CN113378609B (en) | 2020-03-10 | 2020-03-10 | Agent proxy signature identification method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113378609B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113610064B (en) * | 2021-10-09 | 2022-02-08 | 北京世纪好未来教育科技有限公司 | Handwriting recognition method and device |
| CN114241463A (en) * | 2021-11-12 | 2022-03-25 | 中国南方电网有限责任公司 | Signature verification method and device, computer equipment and storage medium |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6424728B1 (en) * | 1999-12-02 | 2002-07-23 | Maan Ammar | Method and apparatus for verification of signatures |
| CN1701323A (en) * | 2001-10-15 | 2005-11-23 | 西尔弗布鲁克研究有限公司 | Digital ink database searching using handwriting feature synthesis |
| CN104820924A (en) * | 2015-05-13 | 2015-08-05 | 重庆邮电大学 | Online safe payment system based on handwriting authentication |
| CN106803082A (en) * | 2017-01-23 | 2017-06-06 | 重庆邮电大学 | A kind of online handwriting recognition methods based on conditional generation confrontation network |
| CN108985297A (en) * | 2018-06-04 | 2018-12-11 | 平安科技（深圳）有限公司 | Handwriting model training, hand-written image recognition methods, device, equipment and medium |
| 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 |
| CN110096977A (en) * | 2019-04-18 | 2019-08-06 | 中金金融认证中心有限公司 | The training method and handwriting verification method, equipment and medium of handwriting verification model |
| TW201942803A (en) * | 2018-03-31 | 2019-11-01 | 華南商業銀行股份有限公司 | Transaction system based on face recognition for verification and method thereof |
| WO2019232853A1 (en) * | 2018-06-04 | 2019-12-12 | 平安科技（深圳）有限公司 | Chinese model training method, chinese image recognition method, device, apparatus and medium |
| CN114241463A (en) * | 2021-11-12 | 2022-03-25 | 中国南方电网有限责任公司 | Signature verification method and device, computer equipment and storage medium |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2003271967A (en) * | 2002-03-19 | 2003-09-26 | Fujitsu Prime Software Technologies Ltd | Program, method and device for authentication of hand- written signature |
| US7715629B2 (en) * | 2005-08-29 | 2010-05-11 | Microsoft Corporation | Style aware use of writing input |
| US9262676B2 (en) * | 2012-09-28 | 2016-02-16 | Intel Corporation | Handwritten signature detection, validation, and confirmation |
| US11556548B2 (en) * | 2017-08-08 | 2023-01-17 | Microsoft Technology Licensing, Llc | Intelligent query system for attachments |
-
2020
- 2020-03-10 CN CN202010162874.2A patent/CN113378609B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6424728B1 (en) * | 1999-12-02 | 2002-07-23 | Maan Ammar | Method and apparatus for verification of signatures |
| CN1701323A (en) * | 2001-10-15 | 2005-11-23 | 西尔弗布鲁克研究有限公司 | Digital ink database searching using handwriting feature synthesis |
| CN104820924A (en) * | 2015-05-13 | 2015-08-05 | 重庆邮电大学 | Online safe payment system based on handwriting authentication |
| CN106803082A (en) * | 2017-01-23 | 2017-06-06 | 重庆邮电大学 | A kind of online handwriting recognition methods based on conditional generation confrontation network |
| TW201942803A (en) * | 2018-03-31 | 2019-11-01 | 華南商業銀行股份有限公司 | Transaction system based on face recognition for verification and method thereof |
| CN108985297A (en) * | 2018-06-04 | 2018-12-11 | 平安科技（深圳）有限公司 | Handwriting model training, hand-written image recognition methods, device, equipment and medium |
| WO2019232853A1 (en) * | 2018-06-04 | 2019-12-12 | 平安科技（深圳）有限公司 | Chinese model training method, chinese image recognition method, device, apparatus and medium |
| 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 |
| CN110096977A (en) * | 2019-04-18 | 2019-08-06 | 中金金融认证中心有限公司 | The training method and handwriting verification method, equipment and medium of handwriting verification model |
| CN114241463A (en) * | 2021-11-12 | 2022-03-25 | 中国南方电网有限责任公司 | Signature verification method and device, computer equipment and storage medium |
Non-Patent Citations (3)
| Title |
|---|
| Personal identification based on iris texture analysis;Li Ma等;《IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE》;第1519-1533页 * |
| 基于纹理分析的笔迹识别系统;杨子华等;《湖南工程学院学报（自然科学版）》;第67-69页 * |
| 多因子融合的移动终端用户身份识别机制研究;张云;《中国优秀硕士学位论文全文数据库 信息科技辑》;第I138-6617页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113378609A (en) | 2021-09-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Emeršič et al. | Evaluation and analysis of ear recognition models: performance, complexity and resource requirements | |
| CN109543690B (en) | Method and device for extracting information | |
| CN111488756B (en) | Face recognition-based living body detection method, electronic device, and storage medium | |
| De Souza et al. | Deep texture features for robust face spoofing detection | |
| CN112084917A (en) | Living body detection method and device | |
| CN111353491B (en) | Text direction determining method, device, equipment and storage medium | |
| CN111191568A (en) | Method, device, equipment and medium for identifying copied image | |
| CN112862024B (en) | Text recognition method and system | |
| CN113378609B (en) | Agent proxy signature identification method and device | |
| CN105335760A (en) | Image number character recognition method | |
| CN113111880A (en) | Certificate image correction method and device, electronic equipment and storage medium | |
| Roy et al. | Face sketch-photo recognition using local gradient checksum: LGCS | |
| CN110650108A (en) | Fishing page identification method based on icon and related equipment | |
| Uma et al. | Copy-move forgery detection of digital images using football game optimization | |
| CN117058723B (en) | Palmprint recognition method, palmprint recognition device and storage medium | |
| Lavanya et al. | Particle Swarm Optimization Ear Identification System | |
| US10657369B1 (en) | Unsupervised removal of text from images using linear programming for optimal filter design | |
| CN116798041A (en) | Image recognition method and device and electronic equipment | |
| CN112365451A (en) | Method, device and equipment for determining image quality grade and computer readable medium | |
| Li | Saliency prediction based on multi-channel models of visual processing | |
| CN116386117A (en) | Face recognition method, device, equipment and storage medium | |
| CN115880702A (en) | Data processing method, device, equipment, program product and storage medium | |
| CN112766082B (en) | Chinese text handwriting identification method and device based on macro-micro characteristics and storage medium | |
| CN114937280A (en) | Method and device for carrying out consistency comparison on document images | |
| Li et al. | Global attention network for collaborative saliency detection |
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 |