WO2021027336A1 - 基于印章和签名的身份验证方法、装置和计算机设备 - Google Patents

基于印章和签名的身份验证方法、装置和计算机设备 Download PDF

Info

Publication number
WO2021027336A1
WO2021027336A1 PCT/CN2020/088000 CN2020088000W WO2021027336A1 WO 2021027336 A1 WO2021027336 A1 WO 2021027336A1 CN 2020088000 W CN2020088000 W CN 2020088000W WO 2021027336 A1 WO2021027336 A1 WO 2021027336A1
Authority
WO
WIPO (PCT)
Prior art keywords
signature
handwritten
designated
pattern
preset
Prior art date
Application number
PCT/CN2020/088000
Other languages
English (en)
French (fr)
Inventor
吴静平
Original Assignee
深圳壹账通智能科技有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 深圳壹账通智能科技有限公司 filed Critical 深圳壹账通智能科技有限公司
Publication of WO2021027336A1 publication Critical patent/WO2021027336A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/24Classification techniques
    • G06F18/241Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2411Classification 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
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/04Architecture, e.g. interconnection topology
    • G06N3/045Combinations of networks
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/08Learning methods
    • G06N3/084Backpropagation, e.g. using gradient descent
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/40Extraction of image or video features
    • G06V10/56Extraction of image or video features relating to colour
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/30Writer recognition; Reading and verifying signatures
    • G06V40/33Writer recognition; Reading and verifying signatures based only on signature image, e.g. static signature recognition

Definitions

  • This application relates to the field of artificial intelligence technology, in particular to an identity verification method, device, computer equipment, and storage medium based on seals and signatures.
  • Identity verification is an indispensable part of modern production and life.
  • signatures, seals and other methods are used for identity verification.
  • the inventor realizes that the traditional technology has relatively simple verification methods for signatures or seals, and it is easy to be deceived by criminals by forging signatures or seals, causing verification errors. Therefore, the verification accuracy of the traditional technology for identity verification technical solutions has defects.
  • the main purpose of this application is to provide an identity verification method, device, computer equipment and storage medium based on seals and signatures, aiming to improve the accuracy of identity verification.
  • this application proposes an identity verification method based on seal and signature, which includes the following steps:
  • the signature pattern is the projection pattern of a pre-stored virtual three-dimensional seal
  • the handwriting features of the handwritten signature are extracted, and the handwriting features are input into the handwriting recognition model based on neural network model training for calculation, so as to obtain the recognition result output by the handwriting recognition model, where the handwriting recognition model Training based on pre-collected handwritten text and sample data composed of writers corresponding to the pre-collected handwritten text;
  • the recognition result is the same as the signature text, obtain the designated projection pattern of the pre-stored virtual three-dimensional seal in the designated projection direction, and use the preset image similarity judgment method to determine whether the designated projection pattern is the same as the signature pattern;
  • This application provides a seal and signature-based identity verification device, including:
  • the acquiring unit is used to acquire the handwritten signature and the signature pattern input by the user, where the signature pattern is a projection pattern of a pre-stored virtual three-dimensional seal;
  • the name judging unit is used to recognize the handwritten signature using a preset text recognition technology to obtain the signature text, and determine whether the signature text is the same as the preset name;
  • the handwriting recognition unit is used to extract the handwriting features of the handwritten signature if the signature text is the same as the preset name, and input the handwriting features into the handwriting recognition model based on neural network model training for calculation, thereby obtaining the recognition output by the handwriting recognition model
  • the handwriting recognition model is trained based on pre-collected handwritten characters and sample data composed of writers corresponding to the pre-collected handwritten characters;
  • the signature text identical judging unit is used to judge whether the recognition result is the same as the signature text
  • the signature pattern judgment unit is used to obtain the designated projection pattern of the pre-stored virtual three-dimensional seal in the designated projection direction if the recognition result is the same as the signature text, and use a preset image similarity judgment method to determine whether the designated projection pattern and the signature pattern are the same;
  • the identity verification passing determining unit is configured to determine that the user's identity verification is passed if the designated projection pattern is the same as the signature pattern.
  • the present application provides a computer device, including: one or more processors; memory; one or more computer programs, wherein one or more computer programs are stored in the memory and configured to be executed by one or more processors , One or more computer programs are configured to execute an identity verification method based on seal and signature: wherein, the identity verification method based on seal and signature includes:
  • the signature pattern is the projection pattern of a pre-stored virtual three-dimensional seal
  • the handwriting features of the handwritten signature are extracted, and the handwriting features are input into the handwriting recognition model based on neural network model training for calculation, so as to obtain the recognition result output by the handwriting recognition model, where the handwriting recognition model Training based on pre-collected handwritten text and sample data composed of writers corresponding to the pre-collected handwritten text;
  • the recognition result is the same as the signature text, obtain the designated projection pattern of the pre-stored virtual three-dimensional seal in the designated projection direction, and use the preset image similarity judgment method to determine whether the designated projection pattern is the same as the signature pattern;
  • the present application provides a computer-readable storage medium with a computer program stored on the computer-readable storage medium.
  • a seal and signature-based identity verification method is implemented, wherein the seal and signature-based identity verification method It includes the following steps:
  • the signature pattern is the projection pattern of a pre-stored virtual three-dimensional seal
  • the handwriting features of the handwritten signature are extracted, and the handwriting features are input into the handwriting recognition model based on neural network model training for calculation, so as to obtain the recognition result output by the handwriting recognition model, where the handwriting recognition model Training based on pre-collected handwritten text and sample data composed of writers corresponding to the pre-collected handwritten text;
  • the recognition result is the same as the signature text, obtain the designated projection pattern of the pre-stored virtual three-dimensional seal in the designated projection direction, and use the preset image similarity judgment method to determine whether the designated projection pattern is the same as the signature pattern;
  • the seal and signature-based identity verification method, device, computer equipment, and storage medium of this application obtain the handwritten signature and signature pattern input by the user; recognize the handwritten signature to obtain the signature text; if the signature text is the same as the preset name, then Extract the handwriting features of the handwritten signature, input the handwriting features into the handwriting recognition model based on neural network model training, and then get the recognition result output by the handwriting recognition model; if the recognition result is the same as the signature text, get the pre-stored virtual three-dimensional seal
  • FIG. 1 is a schematic flowchart of an identity verification method based on a seal and a signature according to an embodiment of the application;
  • FIG. 2 is a schematic block diagram of the structure of a seal and signature-based identity verification device according to an embodiment of the application;
  • FIG. 3 is a schematic block diagram of the structure of a computer device according to an embodiment of the application.
  • an embodiment of the present application provides a seal and signature-based identity verification method, including the following steps:
  • the handwritten signature and the signature pattern input by the user are acquired, wherein the signature pattern is a projection pattern of a pre-stored virtual three-dimensional seal.
  • the handwritten signature and signature pattern of this application are used to cross-verify the identity of the user.
  • the signature pattern is the projection pattern of a pre-stored virtual three-dimensional seal, thereby increasing the risk of inverting the front shape and pattern of the seal only through the seal pattern. This increases the security of the seal and the accuracy of identity verification.
  • the handwritten signature is recognized by the preset text recognition technology to obtain the signature text, and it is determined whether the signature text is the same as the preset name.
  • the preset text recognition technology is, for example, OCR (Optical Character Recognition) technology, in which one or more of the following technical means can be used in the recognition process: Grayscale: RGB model is used to represent each image For each pixel, take the average value of R, G, and B for each pixel instead of the original R, G, and B values to obtain the gray value of the image; binarization: divide the pixels of the image into black and white Part, to distinguish the handwritten signature; noise reduction: use median filter, mean filter, adaptive Wiener filter, etc.
  • tilt correction use Hough transform and other methods to process the image to correct the photo
  • the image is tilted
  • text segmentation use projection operations to split text, project a single line of text or multiple lines of text on the X axis, and accumulate the value.
  • the text area must have a relatively large value, and the interval area must have no value (or a value Small) to segment a single text
  • feature extraction extract the special points of these pixels, such as extreme points, isolated points, etc., as the feature points of the image
  • classification use SVM (Support Vector Machine, support vector The machine) classifier performs classification and obtains the initial recognition result.
  • the characteristic data is, for example, the position of the repetitive strokes and the number of repetitive strokes in the handwritten text.
  • the pen of the handwritten text is divided into multiple points for data collection and analysis, and the pressure value of each point is obtained by identifying the data change trend of the pixel point, The sharpness of the sequence during writing, etc., and then the feature data including the position of the heavy pen and the number of the heavy pen are obtained, where the heavy pen refers to the stroke with the greatest force in the handwritten text.
  • the handwriting features of the handwritten signature are extracted, and the handwriting features are input into the handwriting recognition model based on neural network model training for calculation, so as to obtain the recognition result output by the handwriting recognition model ,
  • the handwriting recognition model is trained based on pre-collected handwritten characters and sample data composed of writers corresponding to the pre-collected handwritten characters.
  • the neural network model can be any model, such as VGG-F model, VGG16 model, ResNet152 model, ResNet50 model, DPN131 model, AlexNet model, DenseNet model, etc., and the DPN model is preferred.
  • DPN Deep Path Network
  • step S4 it is judged whether the recognition result is the same as the signature text. If the recognition result is the same as the signature text, it means that the handwritten signature is indeed genuine and not a forged signature, and the subsequent verification process can be carried out accordingly.
  • the designated projection pattern of the pre-stored virtual three-dimensional seal in the designated projection direction is acquired, and the predetermined image similarity determination method is used to determine whether the designated projection pattern and the signature pattern are the same. Since the virtual three-dimensional seal is not known to outsiders, the designated projection pattern of the pre-stored virtual three-dimensional seal in the designated projection direction is also not known to outsiders, so it can be used for identity verification. Moreover, the virtual three-dimensional seal cannot be deduced from the known projection pattern, thereby ensuring the safety of the seal.
  • the preset image similarity judging method can be any method, for example, comparing the corresponding pixels in the two pictures sequentially. If the number of the same pixels or the proportion of the number is greater than a predetermined threshold, the judgment is the same; if the same pixels If the number of points or the proportion of the number is not greater than a predetermined threshold, the determination is different.
  • step S6 if the designated projection pattern is the same as the signature pattern, it is determined that the user's identity verification is passed. If the designated projection pattern is the same as the signature pattern, the signature verification is passed, and the aforementioned signature verification is combined to cross-verify the user's identity, and the user's identity verification is determined based on this.
  • the handwritten signature is located in a designated picture
  • the step S2 of using a preset text recognition technology to recognize the handwritten signature to obtain the signature text includes:
  • S202 Obtain pixels whose value of reference value F1 is not equal to A, record them as handwritten signature pixels, and record the graphics formed by handwritten signature pixels as handwritten signature graphics;
  • S203 Extract the text features of the handwritten signature graphics, and input them into a preset support vector machine for classification, so as to obtain recognized handwritten text and printed text.
  • the preset text recognition technology is used to recognize the handwritten signature to obtain the signature text.
  • support vector machine is a kind of generalized linear classifier that binary classification of data according to supervised learning method. It is suitable for comparing the recognized text with the pre-stored text to output the most similar text.
  • the character features are, for example, special points in the pixel points corresponding to the character, such as extreme points and isolated points.
  • step S3 includes:
  • S21 Call the pre-collected sample data and divide it into a training set and a test set; where the sample data includes pre-collected handwritten characters and writers corresponding to the pre-collected handwritten characters;
  • the initial handwriting recognition model is recorded as the handwriting recognition model.
  • the set handwriting recognition model is realized.
  • This application is based on a neural network model to train a handwriting recognition model.
  • the neural network model can be VGG16 model, VGG-F model, AlexNet model, ResNet152 model, ResNet50 model, DPN131 model and DenseNet model.
  • the stochastic gradient descent method is to randomly sample some training data to replace the entire training set. If the sample size is large, only some of the samples are used to iterate to the optimal solution, which can increase the training speed. Further, the training can also use the reverse conduction law to update the parameters of each layer of the neural network.
  • the reverse conduction law is based on the gradient descent method, and its input-output relationship is essentially a mapping relationship: the function of a neural network with n inputs and m outputs is from n-dimensional Euclidean space to m-dimensional Ou A continuous mapping of a finite field in the space. This mapping is highly non-linear, which is beneficial to the update of the parameters of each layer of the neural network model.
  • the initial handwriting recognition model In order to obtain the initial handwriting recognition model. Then use the sample data of the test set to verify the initial handwriting recognition model. If the verification passes, the initial handwriting recognition model is recorded as the handwriting recognition model.
  • the handwritten signature is located in a designated picture, and the step S3 of extracting the handwriting features of the handwritten signature includes:
  • this application obtains the color value of handwritten signature pixels, and records adjacent pixels with color values in the same preset range as the detail unit, and the color value of the detail unit Recorded as the average value of the color values of adjacent pixels; obtain the color value change trend of the adjacent detail unit, use the detail unit, the color value and the color value change trend of the detail unit as the handwriting feature of the handwritten signature, and extract the handwriting feature
  • the method uses the detail unit, the color value of the detail unit and the change trend of the color value as the basis for subsequent identification of the writer of the handwritten signature. Among them, due to the different writing habits of different writers, there will be subtle differences in the shape, color, and color value trend of the detail unit when the strength is different, and the correct writer can be identified based on this.
  • the step S1 of acquiring the handwritten signature and signature pattern input by the user includes:
  • the step S5 of obtaining the designated projection pattern of the pre-stored virtual three-dimensional seal in the designated projection direction includes:
  • S501 According to the preset correspondence between the signature and the virtual three-dimensional seal, retrieve the designated virtual three-dimensional seal corresponding to the handwritten signature;
  • S503 Record the direction from the designated coordinate point to the origin as the designated projection direction, and project the designated virtual three-dimensional seal from the designated projection direction to obtain the designated projection pattern.
  • the designated virtual three-dimensional seal is projected from the designated projection direction to obtain the designated projection pattern.
  • this application adopts a virtual three-dimensional seal to prevent the seal from being forged.
  • This application uses the preset corresponding relationship between time and space coordinate points to obtain the designated coordinate point of the generation time of the signature pattern, the direction of the designated coordinate point to the origin is recorded as the designated projection direction, and the designated virtual The three-dimensional seal is projected to obtain a flat projection image, which ensures the safety of the signature (planar projection images at different times, so it is impossible to reverse the flat projection image or specify a virtual three-dimensional seal).
  • the front side of the designated virtual three-dimensional seal can be any one of the preset sides of the designated virtual three-dimensional seal, preferably a side of the designated virtual three-dimensional seal with a specific pattern, where the specific pattern is, for example, the same as the signature of the physical seal or the same as that of the physical seal
  • the embossing corresponding to the signature (the signature of the physical seal is in negative).
  • the step S502 of obtaining the designated coordinate point corresponding to the generation time of the signature pattern according to the preset correspondence relationship between time and space coordinate point includes:
  • the designated coordinate point corresponding to the generation time of the signature pattern is obtained according to the preset correspondence between time and space coordinate points.
  • the designated projection pattern can provide a certain degree of information feedback, which is on the premise of ensuring information security. This will help improve the utilization of information.
  • the step S5 of judging whether the designated projection pattern and the signature pattern are the same using a preset image similarity judgment method includes:
  • S501 Perform gray-scale processing on the designated projection pattern and the signature pattern respectively to obtain a first gray-scale picture and a second gray-scale picture;
  • S502 Calculate the average value Am of the gray values of all pixels in the m-th column or the m-th row of the gray-scale picture, and calculate the average value B of the gray values of all the pixels in the gray-scale picture;
  • grayscale refers to the color representing a grayscale color.
  • the color represents a grayscale color
  • N is the total number of columns or rows in the grayscale image.
  • the overall variance is used to measure the difference between the average value Am of the gray values of the pixels in the mth column or the mth row of the grayscale image and the average value B of all pixels in the grayscale image.
  • the difference According to the formula: Get the difference between the overall variance of the mth column or mth row of two grayscale images
  • Difference of population variance Reflects the difference in the gray value of the mth column or mth row of the two grayscale pictures.
  • the gray value of the mth column or row of the first grayscale image is the same or approximately the same as the gray value of the mth column or row of the second grayscale image (approximate judgment to save computing power , And because the overall variance of the two different pictures is generally not equal, the accuracy of the judgment is very high), on the contrary, the gray value of the mth column or mth row of the first grayscale image is the same as the second grayscale value.
  • the gray value of the mth column or mth row of the picture is different. judgment Whether it is less than the preset variance error threshold. among them The return value is The maximum value in. If If it is less than the preset variance error threshold, it is determined that the designated projection pattern is the same as the signature pattern.
  • the identity verification method based on seal and signature of this application obtains the handwritten signature and signature pattern input by the user; recognizes the handwritten signature to obtain the signature text; if the signature text is the same as the preset name, the handwriting features of the handwritten signature are extracted, Handwriting feature input is calculated based on the handwriting recognition model trained by the neural network model to obtain the recognition result output by the handwriting recognition model; if the recognition result is the same as the signature text, the designated projection pattern of the pre-stored virtual three-dimensional seal in the designated projection direction is obtained ; If the designated projection pattern is the same as the signature pattern, the user's identity verification is determined to pass, thereby improving the accuracy of identity verification.
  • an embodiment of the present application provides an identity verification device based on a seal and a signature, including:
  • the acquiring unit 10 is configured to acquire a handwritten signature and a signature pattern input by a user, wherein the signature pattern is a projection pattern of a pre-stored virtual three-dimensional seal;
  • the name judging unit 20 is used for recognizing the handwritten signature using a preset text recognition technology to obtain the signature text, and judging whether the signature text is the same as the preset name;
  • the handwriting recognition unit 30 is configured to extract the handwriting features of the handwritten signature if the signature text is the same as the preset name, and input the handwriting features into the handwriting recognition model based on neural network model training for calculation, thereby obtaining the output of the handwriting recognition model Recognition results, where the handwriting recognition model is trained based on pre-collected handwritten text and sample data composed of writers corresponding to the pre-collected handwritten text;
  • the signature text identity judgment unit 40 is used to judge whether the recognition result is the same as the signature text
  • the signature pattern judging unit 50 is configured to, if the recognition result is the same as the signature text, obtain the designated projection pattern of the pre-stored virtual three-dimensional seal in the designated projection direction, and use a preset image similarity determination method to determine the designated projection pattern and the signature pattern Are they the same
  • the identity verification passing determining unit 60 is configured to determine that the user's identity verification is passed if the designated projection pattern is the same as the signature pattern.
  • the handwritten signature is located in the designated picture
  • the name judgment unit 20 includes:
  • the handwritten signature pixel acquisition subunit is used to acquire the pixels whose reference value F1 is not equal to A, record them as handwritten signature pixels, and record the graphics formed by handwritten signature pixels as handwritten signature graphics;
  • the classification subunit is used to extract the text characteristics of the handwritten signature graphics and input them into a preset support vector machine for classification, so as to obtain the recognized handwritten text and printed text.
  • the device includes:
  • the sample data calling unit is used to call pre-collected sample data and divide it into a training set and a test set; wherein the sample data includes pre-collected handwritten characters and writers corresponding to the pre-collected handwritten characters;
  • the training unit is used to input the sample data of the training set into the preset neural network model for training, thereby obtaining the initial handwriting recognition model, where the stochastic gradient descent method is used in the training process;
  • the verification unit is used to verify the initial handwriting recognition model with the sample data of the test set
  • the marking unit is used to record the initial handwriting recognition model as a handwriting recognition model if the verification of the initial handwriting recognition model is passed.
  • the handwritten signature is located in a designated picture
  • the handwriting recognition unit 30 includes:
  • the handwritten signature pixel mark subunit is used to obtain the pixels corresponding to the handwritten signature and record them as the handwritten signature pixels;
  • the detail unit marking sub-unit is used to obtain the color value of the handwritten signature pixel, and record the adjacent pixel with the color value within the same preset range as the detail unit, and record the color value of the detail unit as the adjacent pixel The average value of the color values;
  • the handwriting feature extraction subunit is used to obtain the color value change trend of the adjacent detail unit, use the detail unit, the color value and the color value change trend of the detail unit as the handwriting feature of the handwritten signature, and extract the handwriting feature.
  • the acquiring unit 10 includes:
  • the obtaining subunit is used to obtain the handwritten signature, the signature pattern and the generation time of the signature pattern input by the user;
  • the signature pattern judgment unit 50 includes:
  • the designated virtual three-dimensional seal retrieval subunit is used to retrieve the designated virtual three-dimensional seal corresponding to the handwritten signature according to the preset correspondence between the signature and the virtual three-dimensional seal;
  • the designated coordinate point acquisition subunit is used to take the front center of the designated virtual three-dimensional seal as the origin, and acquire the designated coordinate point corresponding to the generation time of the signature pattern according to the preset correspondence relationship between time and space coordinate points;
  • the designated projection pattern acquisition subunit is used to record the direction from the designated coordinate point to the origin as the designated projection direction, and project the designated virtual three-dimensional seal from the designated projection direction to obtain the designated projection pattern.
  • the designated coordinate point acquisition subunit includes:
  • a plane rectangular coordinate system module used to specify the front center of the virtual three-dimensional seal as the origin, the line between the origin and the preset point in the front as the x-axis, and the straight line perpendicular to the x-axis and passing through the origin in the front as the y-axis , The vertical line passing through the origin on the front is the z-axis, thus establishing a plane rectangular coordinate system;
  • the designated coordinate point calculation module is used to obtain the current time and according to the formula:
  • the signature pattern determining unit 50 includes:
  • the gray-scale processing sub-unit is used to perform gray-scale processing on the designated projection pattern and the signature pattern to obtain the first gray-scale picture and the second gray-scale picture;
  • Average value calculation subunit used to calculate the average value Am of the gray values of all pixels in the mth column or mth row of the grayscale image, and calculate the average value B of all pixels in the grayscale image ;
  • the overall variance calculation subunit is used according to the formula: Calculate the overall variance of the mth column or mth row of the grayscale image Where N is the total number of columns or rows in the grayscale image;
  • the difference of the population variance is calculated as a subunit, which is used according to the formula: Get the difference between the overall variance of the mth column or mth row of two grayscale images among them, Is the overall variance of the mth column or mth row of the first grayscale image, Is the overall variance of the mth column or mth row of the second grayscale image;
  • Variance error threshold judgment subunit used to judge Whether it is less than the preset variance error threshold
  • the seal and signature-based identity verification device of the present application obtains the handwritten signature and signature pattern input by the user; recognizes the handwritten signature to obtain the signature text; if the signature text is the same as the preset name, the handwriting features of the handwritten signature are extracted, and Handwriting feature input is calculated based on the handwriting recognition model trained by the neural network model to obtain the recognition result output by the handwriting recognition model; if the recognition result is the same as the signature text, the designated projection pattern of the pre-stored virtual three-dimensional seal in the designated projection direction is obtained ; If the designated projection pattern is the same as the signature pattern, the user's identity verification is determined to pass, thereby improving the accuracy of identity verification.
  • an embodiment of the present application also provides a computer device.
  • the computer device may be a server, and its internal structure may be as shown in the figure.
  • the computer equipment includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the computer designed processor is used to provide calculation and control capabilities.
  • the memory of the computer device includes a non-volatile storage medium and an internal memory.
  • the non-volatile storage medium stores an operating system, a computer program, and a database.
  • the memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium.
  • the database of the computer equipment is used to store the data used in the identity verification method based on the seal and signature.
  • the network interface of the computer device is used to communicate with an external terminal through a network connection.
  • the computer program is executed by the processor to realize an identity verification method based on seal and signature.
  • the above-mentioned processor executes the above-mentioned seal-and-signature-based identity verification method, wherein the steps included in the method respectively correspond to the steps of executing the seal-and-signature-based identity verification method in the foregoing embodiment, and will not be repeated here.
  • the computer device of this application obtains the handwritten signature and signature pattern input by the user; recognizes the handwritten signature to obtain the signature text; if the signature text is the same as the preset name, the handwriting feature of the handwritten signature is extracted, and the handwriting feature input is based on the neural network
  • the calculation is performed in the handwriting recognition model completed by the model training to obtain the recognition result output by the handwriting recognition model; if the recognition result is the same as the signature text, the designated projection pattern of the pre-stored virtual three-dimensional seal in the designated projection direction is obtained; if the designated projection pattern is the same If the signature pattern is the same, it is determined that the user's identity verification is passed, thereby improving the accuracy of identity verification.
  • An embodiment of the present application also provides a computer-readable storage medium.
  • the computer-readable storage medium is a volatile storage medium or a non-volatile storage medium, and a computer program is stored thereon.
  • the computer program is executed by a processor, A seal and signature-based identity verification method is implemented, and the steps included in the method respectively correspond to the steps of performing the seal and signature-based identity verification method of the foregoing embodiment, and will not be repeated here.
  • the computer-readable storage medium of this application obtains the handwritten signature and signature pattern input by the user; recognizes the handwritten signature to obtain the signature text; if the signature text is the same as the preset name, extract the handwriting characteristics of the handwritten signature and input the handwriting characteristics Perform calculations in the handwriting recognition model completed based on the neural network model training to obtain the recognition result output by the handwriting recognition model; if the recognition result is the same as the signature text, obtain the designated projection pattern of the pre-stored virtual three-dimensional seal in the designated projection direction; if designated If the projection pattern is the same as the signature pattern, it is determined that the user's identity verification is passed, thereby improving the accuracy of identity verification.
  • Non-volatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory.
  • Volatile memory may include random access memory (RAM) or external cache memory.
  • RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Molecular Biology (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Physics (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Multimedia (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Human Computer Interaction (AREA)
  • Collating Specific Patterns (AREA)

Abstract

本申请揭示了一种基于印章和签名的身份验证方法、装置、计算机设备和存储介质,所述方法包括:获取用户输入的手写签名与签章图案;采用预设的文字识别技术识别所述手写签名从而获取签名文本;若所述签名文本与预设的姓名相同,则提取所述手写签名的字迹特征,将所述字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到所述字迹识别模型输出的识别结果;若所述识别结果与所述签名文本相同,则获取预存的虚拟立体印章在指定投影方向的指定投影图案,并利用预设的图像相似判断方法判断所述指定投影图案与所述签章图案是否相同;若所述指定投影图案与所述签章图案相同,则判定所述用户的身份验证通过。

Description

基于印章和签名的身份验证方法、装置和计算机设备
本申请要求于2019年8月14日提交中国专利局、申请号为201910750602.1,发明名称为“基于印章和签名的身份验证方法、装置和计算机设备”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本申请涉及到人工智能技术领域,特别是涉及到一种基于印章和签名的身份验证方法、装置、计算机设备和存储介质。
背景技术
身份验证是现代生产生活中不可缺少的一部分,一般采用签名、签章等方式进行身份验证。发明人意识到传统技术对于签名或者签章的验证方式较为简单,容易被不法分子通过伪造签名或者签章的方式欺骗,造成验证错误。因此传统技术进行身份验证的技术方案的验证准确度存在缺陷。
技术问题
本申请的主要目的为提供一种基于印章和签名的身份验证方法、装置、计算机设备和存储介质,旨在提高身份验证的准确性。
技术解决方案
为了实现上述发明目的,本申请提出一种基于印章和签名的身份验证方法,包括以下步骤:
获取用户输入的手写签名与签章图案,其中签章图案是预存的虚拟立体印章的投影图案;
采用预设的文字识别技术识别手写签名从而获取签名文本,并判断签名文本是否与预设的姓名相同;
若签名文本与预设的姓名相同,则提取手写签名的字迹特征,将字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到字迹识别模型输出的识别结果,其中字迹识别模型基于预先采集的手写文字,以及与预先采集的手写文字对应的书写者组成的样本数据训练而成;
判断识别结果是否与签名文本相同;
若识别结果与签名文本相同,则获取预存的虚拟立体印章在指定投影方向的指定投影图案,并利用预设的图像相似判断方法判断指定投影图案与签章图案是否相同;
若指定投影图案与签章图案相同,则判定用户的身份验证通过。
本申请提供一种基于印章和签名的身份验证装置,包括:
获取单元,用于获取用户输入的手写签名与签章图案,其中签章图案是预存的虚拟立体印章的投影图案;
姓名判断单元,用于采用预设的文字识别技术识别手写签名从而获取签名文本,并判断签名文本是否与预设的姓名相同;
字迹识别单元,用于若签名文本与预设的姓名相同,则提取手写签名的字迹特征,将字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计 算,从而得到字迹识别模型输出的识别结果,其中字迹识别模型基于预先采集的手写文字,以及与预先采集的手写文字对应的书写者组成的样本数据训练而成;
签名文本相同判断单元,用于判断识别结果是否与签名文本相同;
签章图案判断单元,用于若识别结果与签名文本相同,则获取预存的虚拟立体印章在指定投影方向的指定投影图案,并利用预设的图像相似判断方法判断指定投影图案与签章图案是否相同;
身份验证通过判定单元,用于若指定投影图案与签章图案相同,则判定用户的身份验证通过。
本申请提供一种计算机设备,包括:一个或多个处理器;存储器;一个或多个计算机程序,其中一个或多个计算机程序被存储在存储器中并被配置为由一个或多个处理器执行,一个或多个计算机程序配置用于执行一种基于印章和签名的身份验证方法:其中,基于印章和签名的身份验证方法包括:
获取用户输入的手写签名与签章图案,其中签章图案是预存的虚拟立体印章的投影图案;
采用预设的文字识别技术识别手写签名从而获取签名文本,并判断签名文本是否与预设的姓名相同;
若签名文本与预设的姓名相同,则提取手写签名的字迹特征,将字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到字迹识别模型输出的识别结果,其中字迹识别模型基于预先采集的手写文字,以及与预先采集的手写文字对应的书写者组成的样本数据训练而成;
判断识别结果是否与签名文本相同;
若识别结果与签名文本相同,则获取预存的虚拟立体印章在指定投影方向的指定投影图案,并利用预设的图像相似判断方法判断指定投影图案与签章图案是否相同;
若指定投影图案与签章图案相同,则判定用户的身份验证通过。
本申请提供一种计算机可读存储介质,计算机可读存储介质上存储有计算机程序,该计算机程序被处理器执行时实现基于印章和签名的身份验证方法,其中,基于印章和签名的身份验证方法包括以下步骤:
获取用户输入的手写签名与签章图案,其中签章图案是预存的虚拟立体印章的投影图案;
采用预设的文字识别技术识别手写签名从而获取签名文本,并判断签名文本是否与预设的姓名相同;
若签名文本与预设的姓名相同,则提取手写签名的字迹特征,将字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到字迹识别模型输出的识别结果,其中字迹识别模型基于预先采集的手写文字,以及与预先采集的手写文字对应的书写者组成的样本数据训练而成;
判断识别结果是否与签名文本相同;
若识别结果与签名文本相同,则获取预存的虚拟立体印章在指定投影方向 的指定投影图案,并利用预设的图像相似判断方法判断指定投影图案与签章图案是否相同;
若指定投影图案与签章图案相同,则判定用户的身份验证通过。
有益效果
本申请的基于印章和签名的身份验证方法、装置、计算机设备和存储介质,获取用户输入的手写签名与签章图案;识别手写签名从而获取签名文本;若签名文本与预设的姓名相同,则提取手写签名的字迹特征,将字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到字迹识别模型输出的识别结果;若识别结果与签名文本相同,则获取预存的虚拟立体印章在指定投影方向的指定投影图案;若指定投影图案与签章图案相同,则判定用户的身份验证通过,从而提高了身份验证的准确性。
附图说明
图1为本申请一实施例的基于印章和签名的身份验证方法的流程示意图;
图2为本申请一实施例的基于印章和签名的身份验证装置的结构示意框图;
图3为本申请一实施例的计算机设备的结构示意框图。
本发明的最佳实施方式
参照图1,本申请实施例提供一种基于印章和签名的身份验证方法,包括以下步骤:
S1、获取用户输入的手写签名与签章图案,其中签章图案是预存的虚拟立体印章的投影图案;
S2、采用预设的文字识别技术识别手写签名从而获取签名文本,并判断签名文本是否与预设的姓名相同;
S3、若签名文本与预设的姓名相同,则提取手写签名的字迹特征,将字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到字迹识别模型输出的识别结果,其中字迹识别模型基于预先采集的手写文字,以及与预先采集的手写文字对应的书写者组成的样本数据训练而成;
S4、判断识别结果是否与签名文本相同;
S5、若识别结果与签名文本相同,则获取预存的虚拟立体印章在指定投影方向的指定投影图案,并利用预设的图像相似判断方法判断指定投影图案与签章图案是否相同;
S6、若指定投影图案与签章图案相同,则判定用户的身份验证通过。
如上述步骤S1,获取用户输入的手写签名与签章图案,其中签章图案是预存的虚拟立体印章的投影图案。本申请的手写签名与签章图案用于交叉验证用户的身份,其中签章图案是预存的虚拟立体印章的投影图案,从而提高了仅通过印章图案,逆推出印章的正面形状和图案的风险,进而增加了印章的安全性与身份验证的准确性。
如上述步骤S2,采用预设的文字识别技术识别手写签名从而获取签名文本,并判断签名文本是否与预设的姓名相同。其中预设的文字识别技术例如为 OCR(Optical Character Recognition,光学字符识别)技术,其中在识别过程中可以采中下述一种或者多种技术手段:灰度化:采用RGB模型表示图像的每个像素点,取每个像素点的R、G、B的平均值代替原来的R、G、B的值得到图像的灰度值;二值化:将图像的像素点分为黑色和白色两部分,以区分出手写签名;降噪:采用中值滤波、均值滤波、自适应维纳滤波等进行滤波,以处理图像噪声;倾斜矫正:采用霍夫变换等方法处理图像,以矫正拍照等导致的图像倾斜;文字分割:采用投影运算进行文字切分,将单行文字或者多行文字投影到X轴上,并将值累加,文字的区域必定值比较大,间隔区域必定没有值(或者值较小),以此分割出单个文字;特征提取:提取出这些像素点中的特殊的点如极值点,孤立点等,作为图像的特征点;分类:采用SVM(Support Vector Machine,采用支持向量机)分类器进行分类,得到初识别结。其中特征数据例如为手写文字中的重笔位置与重笔数量,例如通过将手写文字的笔划分解为多个点进行数据采集分析,通过识别像素点的数据变化趋势得到每个点的压力值、书写时顺序的清晰度等,进而获取包括重笔位置与重笔数量的特征数据,其中重笔指手写文字中用力最大的笔划。
如上述步骤S3,若签名文本与预设的姓名相同,则提取手写签名的字迹特征,将字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到字迹识别模型输出的识别结果,其中字迹识别模型基于预先采集的手写文字,以及与预先采集的手写文字对应的书写者组成的样本数据训练而成。其中神经网络模型可以为任意模型,例如VGG-F模型、VGG16模型、ResNet152模型、ResNet50模型、DPN131模型、AlexNet模型和DenseNet模型等,优选DPN模型。DPN(Dual Path Network)是神经网络结构,在ResNeXt的基础上引入了DenseNet的核心内容,使得模型对特征的利用更加充分。上述DPN、ResNeXt和DenseNet是现有的网络结构,在此不在赘述。从而识别出手写签名对应的书写者。
如上述步骤S4,判断识别结果是否与签名文本相同。若识别结果与签名文本相同,即表示手写签名的确是真的,并非伪造签名,据此可以进行后续的验证流程。
如上述步骤S5,若识别结果与签名文本相同,则获取预存的虚拟立体印章在指定投影方向的指定投影图案,并利用预设的图像相似判断方法判断指定投影图案与签章图案是否相同。由于虚拟立体印章不为外人所知,所以预存的虚拟立体印章在指定投影方向的指定投影图案也不为外人所知,因此可用于验证身份。并且由已知的投影图案无法反推出虚拟立体印章,从而保证了印章的安全性。更进一步地,由于下次签章采用的投影方向与本次签章的投影方向不同,因此本次签章的平面投影图像不能用于下次签章,从而杜绝了盗用平面投影图像用于下次伪造签章的可能性。其中预设的图像相似判断方法可以为任意方法,例如为依次对比两张图片中对应的像素点,若相同的像素点的数量或者数量的占比大于预定阈值,则判定相同;若相同的像素点的数量或者数量的占比不大于预定阈值,则判定不相同。
如上述步骤S6,若指定投影图案与签章图案相同,则判定用户的身份验证通过。若指定投影图案与签章图案相同,即签章验证通过,再结合前述的签名验证,从而交叉验证用户的身份,据此判定用户的身份验证通过。
在一个实施方式中,手写签名位于指定图片中,采用预设的文字识别技术识别手写签名从而获取签名文本的步骤S2,包括:
S201、采集指定图片中的像素点的RGB颜色模型中的R颜色通道的数值、G颜色通道的数值和B颜色通道的数值,并根据公式:F1=MIN{ROUND[(a1R+a2G+a3B)/L,0],A},获取参考数值F1,其中MIN为最小值函数,ROUND为四舍五入函数,a1、a2、a3和L均为预设参数,a1、a2、a3均为大于0且小于L的正数,L为大于0的整数,A为预设的取值在范围(0,255)之内阈值参数,R、G、B分别为指定图片中的指定像素点的RGB颜色模型中的R颜色通道的数值、G颜色通道的数值和B颜色通道的数值;
S202、获取参考数值F1的值不等于A的像素点,记为手写签名像素点,并将手写签名像素点构成的图形记为手写签名图形;
S203、提取手写签名图形的文字特征,并输入预设的支持向量机中进行分类,从而获得识别而得的手写文字文本和印刷体文字文本。
如上,实现了采用预设的文字识别技术识别手写签名从而获取签名文本。为了更准确地提取手写签名,本申请采用公式:F1=MIN{ROUND[(a1R+a2G+a3B)/L,0],A},将背景颜色与手写签名区分开来,从而保留了手写签名的特征细节,而参考数值F1的值不等于A的像素点即被视为黑色字体颜色,被视为手写签名的像素点,因此手写签名像素点构成的图形记为手写签名图形。再提取手写签名图形的文字特征,并输入预设的支持向量机中进行分类,从而获得识别而得的手写文字文本和印刷体文字文本。其中支持向量机是一类按监督学习方式对数据进行二元分类的广义线性分类器,适用于对待识别文字与预存的文字进行对比,以输出最相似的文字。其中文字特征例如为文字对应的像素点中的特殊的点如极值点,孤立点等。其中ROUND函数是四舍五入函数,ROUND(X,a)指对实数X按小数位为a进行四舍五入运算,其中a为大于等于0的整数,例如ROUND(2.1,0)=2。
在一个实施方式中,若签名文本与预设的姓名相同,则提取手写签名的字迹特征,将字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到字迹识别模型输出的识别结果,其中字迹识别模型基于预先采集的手写文字,以及与预先采集的手写文字对应的书写者组成的样本数据训练而成的步骤S3之前,包括:
S21、调用预先采集的样本数据,并分成训练集和测试集;其中,样本数据包括预先采集的手写文字,以及与预先采集的手写文字对应的书写者;
S22、将训练集的样本数据输入到预设的神经网络模型中进行训练,从而得到初始字迹识别模型,其中,训练的过程中采用随机梯度下降法;
S23、利用测试集的样本数据验证初始字迹识别模型;
S24、若初始字迹识别模型验证通过,则将初始字迹识别模型记为字迹识 别模型。
如上,实现了设置字迹识别模型。本申请基于神经网络模型以训练出字迹识别模型。其中神经网络模型可为VGG16模型、VGG-F模型、AlexNet模型、ResNet152模型、ResNet50模型、DPN131模型和DenseNet模型等。其中,随机梯度下降法就是随机取样一些训练数据,替代整个训练集,如果样本量很大的情况,只用其中部分的样本,就已经迭代到最优解了,可以提高训练速度。进一步地,训练还可以采用反向传导法则更新神经网络各层的参数。其中反向传导法则是建立在梯度下降法的基础上,其输入输出关系实质上是一种映射关系:一个n输入m输出的神经网络所完成的功能是从n维欧氏空间向m维欧氏空间中一有限域的连续映射,这一映射具有高度非线性,有利于神经网络模型各层的参数的更新。从而获得初始字迹识别模型。再利用测试集的样本数据验证初始字迹识别模型,若验证通过,则将初始字迹识别模型记为字迹识别模型。
在一个实施方式中,手写签名位于指定图片中,提取手写签名的字迹特征的步骤S3,包括:
S301、获取手写签名对应的像素点,并记为手写签名像素点;
S302、获取手写签名像素点的颜色数值,并将颜色数值处于同一预设范围内的相邻像素点记为细节单位,并将细节单位的颜色数值记为相邻像素点的颜色数值的平均值;
S303、获取相邻的细节单位的颜色数值变化趋势,将细节单位、细节单位的颜色数值和颜色数值变化趋势作为手写签名的字迹特征,并提取字迹特征。
如上,实现了提取手写签名的字迹特征。为了节省算力,并且有效利用特征细节,本申请通过获取手写签名像素点的颜色数值,并将颜色数值处于同一预设范围内的相邻像素点记为细节单位,并将细节单位的颜色数值记为相邻像素点的颜色数值的平均值;获取相邻的细节单位的颜色数值变化趋势,将细节单位、细节单位的颜色数值和颜色数值变化趋势作为手写签名的字迹特征,并提取字迹特征的方式,利用细节单位、细节单位的颜色数值和颜色数值变化趋势作为后续识别手写签名的书写者的基础。其中由于不同书写者在书写时的用力习惯不同,会导致用力轻重不同时的细节单位的形状、颜色和颜色数值变化趋势存在细微的区别,据此可识别出正确的书写者。
在一个实施方式中,获取用户输入的手写签名与签章图案的步骤S1,包括:
S101、获取用户输入的手写签名、签章图案和签章图案的生成时间;
获取预存的虚拟立体印章在指定投影方向的指定投影图案的步骤S5,包括:
S501、根据预设的签名与虚拟立体印章的对应关系,调取手写签名对应的指定虚拟立体印章;
S502、以指定虚拟立体印章的正面中心为原点,根据预设的时间与空间坐标点的对应关系,获取与签章图案的生成时间对应的指定坐标点;
S503、将指定坐标点指向原点的方向记为指定投影方向,并从指定投影方向对指定虚拟立体印章进行投影,从而得到指定投影图案。
如上,实现了从指定投影方向对指定虚拟立体印章进行投影,从而得到指定投影图案。为了提高印章的安全性,本申请采用了虚拟立体印章以防止印章被伪造。本申请采用根据预设的时间与空间坐标点的对应关系,获取签章图案的生成时间的指定坐标点,将指定坐标点指向原点的方向记为指定投影方向,并从指定投影方向对指定虚拟立体印章进行投影,从而得到平面投影图像的方式,保证了签章的安全性(不同时间的平面投影图像,因此逆向倒推平面投影图像或者指定虚拟立体印章是不可能的)。其中指定虚拟立体印章的正面可为指定虚拟立体印章预设的任意一个面,优选为指定虚拟立体印章的一个具有特定图案的面,其中特定图案例如与实体印章的签章相同或者与实体印章的签章对应的阳文(以实体印章的签章为阴文)。
在一个实施方式中,以指定虚拟立体印章的正面中心为原点,根据预设的时间与空间坐标点的对应关系,获取与签章图案的生成时间对应的指定坐标点的步骤S502,包括:
S5021、以指定虚拟立体印章的正面中心为原点,原点与正面中的预设点的连线作为x轴,正面中与x轴垂直的且过原点的直线作为y轴,正面的过原点的垂线为z轴,从而建立平面直角坐标系;
S5022、获取当前时间,并根据公式:
x=k1×M+a1;y=k2×D+a2;z=k3×T+a3,获取指定坐标点(x,y,z),其中签章图案的生成时间为签章年份的第M月中的第D天中的第T个小时,其中k1、k2、k3、b1、b2和b3均为预设的参数。
如上,实现了根据预设的时间与空间坐标点的对应关系,获取与签章图案的生成时间对应的指定坐标点。本申请将签章图案的生成时间分解为当前年份的第M月中的第D天中的第T个小时,并且根据第M月、第D天、第T个小时,利用公式x=k1×M+a1;y=k2×D+a2;z=k3×T+a3,获取指定坐标点(x,y,z),从而进一步保证了签章的安全性。并且由于x轴、y轴、z轴分别与月、日、小时相关,也即平面投影图像与月、日、小时相关,因此指定投影图案能够进行一定程度的信息反馈,在保证信息安全的前提下,有利于提高信息的利用率。
在一个实施方式中,利用预设的图像相似判断方法判断指定投影图案与签章图案是否相同的步骤S5,包括:
S501、分别对指定投影图案与签章图案进行灰度化处理,得到第一灰度图片和第二灰度图片;
S502、计算灰度图片的第m列或者第m行的所有像素点的灰度值的平均值Am,以及计算灰度图片中所有像素点的灰度值的平均值B;
S503、根据公式:
Figure PCTCN2020088000-appb-000001
计算灰度图片的第m列或者第m行 的总体方差
Figure PCTCN2020088000-appb-000002
其中N为灰度图片中的列或者行的总数量;
S504、根据公式:
Figure PCTCN2020088000-appb-000003
获得两张灰度图片的第m列或者第m行的总体方差之差
Figure PCTCN2020088000-appb-000004
其中,
Figure PCTCN2020088000-appb-000005
为第一张灰度图片的第m列或者第m行的总体方差,
Figure PCTCN2020088000-appb-000006
为第二张灰度图片的第m列或者第m行的总体方差;
S505、判断
Figure PCTCN2020088000-appb-000007
是否小于预设的方差误差阈值;
S506、若
Figure PCTCN2020088000-appb-000008
小于预设的方差误差阈值,则判定指定投影图案与签章图案相同。
如上,实现了利用预设的图像相似判断方法判断指定投影图案与签章图案是否相同。其中,灰度化指将彩色表示一种灰度颜色,例如在在RGB模型中,如果R=G=B时,则彩色表示一种灰度颜色,其中R=G=B的值叫灰度值,因此,灰度图像每个像素只需一个字节存放灰度值(又称强度值、亮度值),从而减少存储量。根据公式:
Figure PCTCN2020088000-appb-000009
计算灰度图片的第m列或者第m行的总体方差
Figure PCTCN2020088000-appb-000010
其中N为灰度图片中的列或者行的总数量。在本申请中,采用总体方差来衡量灰度图片的第m列或者第m行的像素点的灰度值的平均值Am与灰度图片中所有像素点的灰度值的平均值B之间的差异。根据公式:
Figure PCTCN2020088000-appb-000011
获得两张灰度图片的第m列或者第m行的总体方差之差
Figure PCTCN2020088000-appb-000012
总体方差之差
Figure PCTCN2020088000-appb-000013
反应了两张灰度图片的第m列或者第m行的灰度值的差异。当
Figure PCTCN2020088000-appb-000014
较小时,例如为0时,表明
Figure PCTCN2020088000-appb-000015
等于或者近似等于
Figure PCTCN2020088000-appb-000016
可视为第一张灰度图片第m列或者第m行的灰度值与第二张灰度图片第m列或者第m行的灰度值相同或者近似相同(近似判断,以节省算力,并且由于不同的两张图片的总体方差一般不相等,因此该判断的准确性很高),反之认为第一张灰度图片第m列或者第m行的灰度值与第二张灰度图片第m列或者第m行的灰度值不相同。判断
Figure PCTCN2020088000-appb-000017
是否小于预设的方差误差阈值。其中
Figure PCTCN2020088000-appb-000018
的返回值即为
Figure PCTCN2020088000-appb-000019
中的最大值。若
Figure PCTCN2020088000-appb-000020
小于预设的方差误差阈值,则判定指定投影图案与签章图案相同。
本申请的基于印章和签名的身份验证方法,获取用户输入的手写签名与签章图案;识别手写签名从而获取签名文本;若签名文本与预设的姓名相同,则提取手写签名的字迹特征,将字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到字迹识别模型输出的识别结果;若识别结果与签名文本相同,则获取预存的虚拟立体印章在指定投影方向的指定投影图案;若指定投影图案与签章图案相同,则判定用户的身份验证通过,从而提高了身份验证的准确性。
参照图2,本申请实施例提供一种基于印章和签名的身份验证装置,包括:
获取单元10,用于获取用户输入的手写签名与签章图案,其中签章图案是预存的虚拟立体印章的投影图案;
姓名判断单元20,用于采用预设的文字识别技术识别手写签名从而获取签名文本,并判断签名文本是否与预设的姓名相同;
字迹识别单元30,用于若签名文本与预设的姓名相同,则提取手写签名的字迹特征,将字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到字迹识别模型输出的识别结果,其中字迹识别模型基于预先采集的手写文字,以及与预先采集的手写文字对应的书写者组成的样本数据训练而成;
签名文本相同判断单元40,用于判断识别结果是否与签名文本相同;
签章图案判断单元50,用于若识别结果与签名文本相同,则获取预存的虚拟立体印章在指定投影方向的指定投影图案,并利用预设的图像相似判断方法判断指定投影图案与签章图案是否相同;
身份验证通过判定单元60,用于若指定投影图案与签章图案相同,则判定用户的身份验证通过。
其中上述单元分别用于执行的操作与前述实施方式的基于印章和签名的身份验证方法的步骤一一对应,在此不再赘述。
在一个实施方式中,手写签名位于指定图片中,姓名判断单元20,包括:
参考数值获取子单元,用于采集指定图片中的像素点的RGB颜色模型中的R颜色通道的数值、G颜色通道的数值和B颜色通道的数值,并根据公式:F1=MIN{ROUND[(a1R+a2G+a3B)/L,0],A},获取参考数值F1,其中MIN为最小值函数,ROUND为四舍五入函数,a1、a2、a3和L均为预设参数,a1、a2、a3均为大于0且小于L的正数,L为大于0的整数,A为预设的取值在范围(0,255)之内阈值参数,R、G、B分别为指定图片中的指定像素点的RGB颜色模型中的R颜色通道的数值、G颜色通道的数值和B颜色通道的数值;
手写签名像素点获取子单元,用于获取参考数值F1的值不等于A的像素点,记为手写签名像素点,并将手写签名像素点构成的图形记为手写签名图形;
分类子单元,用于提取手写签名图形的文字特征,并输入预设的支持向量机中进行分类,从而获得识别而得的手写文字文本和印刷体文字文本。
其中上述子单元分别用于执行的操作与前述实施方式的基于印章和签名的身份验证方法的步骤一一对应,在此不再赘述。
在一个实施方式中,装置,包括:
样本数据调用单元,用于调用预先采集的样本数据,并分成训练集和测试集;其中,样本数据包括预先采集的手写文字,以及与预先采集的手写文字对应的书写者;
训练单元,用于将训练集的样本数据输入到预设的神经网络模型中进行训练,从而得到初始字迹识别模型,其中,训练的过程中采用随机梯度下降法;
验证单元,用于利用测试集的样本数据验证初始字迹识别模型;
标记单元,用于若初始字迹识别模型验证通过,则将初始字迹识别模型记为字迹识别模型。
其中上述单元分别用于执行的操作与前述实施方式的基于印章和签名的身份验证方法的步骤一一对应,在此不再赘述。
在一个实施方式中,手写签名位于指定图片中,字迹识别单元30,包括:
手写签名像素点标记子单元,用于获取手写签名对应的像素点,并记为手写签名像素点;
细节单位标记子单元,用于获取手写签名像素点的颜色数值,并将颜色数值处于同一预设范围内的相邻像素点记为细节单位,并将细节单位的颜色数值记为相邻像素点的颜色数值的平均值;
字迹特征提取子单元,用于获取相邻的细节单位的颜色数值变化趋势,将细节单位、细节单位的颜色数值和颜色数值变化趋势作为手写签名的字迹特征,并提取字迹特征。
其中上述子单元分别用于执行的操作与前述实施方式的基于印章和签名的身份验证方法的步骤一一对应,在此不再赘述。
在一个实施方式中,获取单元10,包括:
获取子单元,用于获取用户输入的手写签名、签章图案和签章图案的生成时间;
签章图案判断单元50,包括:
指定虚拟立体印章调取子单元,用于根据预设的签名与虚拟立体印章的对应关系,调取手写签名对应的指定虚拟立体印章;
指定坐标点获取子单元,用于以指定虚拟立体印章的正面中心为原点,根据预设的时间与空间坐标点的对应关系,获取与签章图案的生成时间对应的指定坐标点;
指定投影图案获取子单元,用于将指定坐标点指向原点的方向记为指定投影方向,并从指定投影方向对指定虚拟立体印章进行投影,从而得到指定投影图案。
其中上述子单元分别用于执行的操作与前述实施方式的基于印章和签名的身份验证方法的步骤一一对应,在此不再赘述。
在一个实施方式中,指定坐标点获取子单元,包括:
建立平面直角坐标系模块,用于以指定虚拟立体印章的正面中心为原点,原点与正面中的预设点的连线作为x轴,正面中与x轴垂直的且过原点的直线作为y轴,正面的过原点的垂线为z轴,从而建立平面直角坐标系;
指定坐标点计算模块,用于获取当前时间,并根据公式:
x=k1×M+a1;y=k2×D+a2;z=k3×T+a3,获取指定坐标点(x,y,z),其中签章图案的生成时间为签章年份的第M月中的第D天中的第T个小时,其中k1、k2、k3、b1、b2和b3均为预设的参数。
其中上述模块分别用于执行的操作与前述实施方式的基于印章和签名的身份验证方法的步骤一一对应,在此不再赘述。
在一个实施方式中,签章图案判断单元50,包括:
灰度化处理子单元,用于分别对指定投影图案与签章图案进行灰度化处理,得到第一灰度图片和第二灰度图片;
平均值计算子单元,用于计算灰度图片的第m列或者第m行的所有像素点的灰度值的平均值Am,以及计算灰度图片中所有像素点的灰度值的平均值B;
总体方差计算子单元,用于根据公式:
Figure PCTCN2020088000-appb-000021
计算灰度图片的第m列或者第m行的总体方差
Figure PCTCN2020088000-appb-000022
其中N为灰度图片中的列或者行的总数量;
总体方差之差计算子单元,用于根据公式:
Figure PCTCN2020088000-appb-000023
获得两张灰度图片的第m列或者第m行的总体方差之差
Figure PCTCN2020088000-appb-000024
其中,
Figure PCTCN2020088000-appb-000025
为第一张灰度图片的第m列或者第m行的总体方差,
Figure PCTCN2020088000-appb-000026
为第二张灰度图片的第m列或者第m行的总体方差;
方差误差阈值判断子单元,用于判断
Figure PCTCN2020088000-appb-000027
是否小于预设的方差误差阈值;
相同判定子单元,用于若
Figure PCTCN2020088000-appb-000028
小于预设的方差误差阈值,则判定指定投影图案与签章图案相同。
其中上述子单元分别用于执行的操作与前述实施方式的基于印章和签名的身份验证方法的步骤一一对应,在此不再赘述。
本申请的基于印章和签名的身份验证装置,获取用户输入的手写签名与签章图案;识别手写签名从而获取签名文本;若签名文本与预设的姓名相同,则提取手写签名的字迹特征,将字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到字迹识别模型输出的识别结果;若识别结果与签名文本相同,则获取预存的虚拟立体印章在指定投影方向的指定投影图案;若指定投影图案与签章图案相同,则判定用户的身份验证通过,从而提高了身份验证的准确性。
参照图3,本申请实施例中还提供一种计算机设备,该计算机设备可以是服务器,其内部结构可以如图所示。该计算机设备包括通过系统总线连接的处理器、存储器、网络接口和数据库。其中,该计算机设计的处理器用于提供计算和控制能力。该计算机设备的存储器包括非易失性存储介质、内存储器。该非易失性存储介质存储有操作系统、计算机程序和数据库。该内存器为非易失性存储介质中的操作系统和计算机程序的运行提供环境。该计算机设备的数据库用于存储基于印章和签名的身份验证方法所用数据。该计算机设备的网络接口用于与外部的终端通过网络连接通信。该计算机程序被处理器执行时以实现一种基于印章和签名的身份验证方法。
上述处理器执行上述基于印章和签名的身份验证方法,其中方法包括的步 骤分别与执行前述实施方式的基于印章和签名的身份验证方法的步骤一一对应,在此不再赘述。
本申请的计算机设备,获取用户输入的手写签名与签章图案;识别手写签名从而获取签名文本;若签名文本与预设的姓名相同,则提取手写签名的字迹特征,将字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到字迹识别模型输出的识别结果;若识别结果与签名文本相同,则获取预存的虚拟立体印章在指定投影方向的指定投影图案;若指定投影图案与签章图案相同,则判定用户的身份验证通过,从而提高了身份验证的准确性。
本申请一实施例还提供一种计算机可读存储介质,所述计算机可读存储介质为易失性存储介质或非易失性存储介质,其上存储有计算机程序,计算机程序被处理器执行时实现基于印章和签名的身份验证方法,其中方法包括的步骤分别与执行前述实施方式的基于印章和签名的身份验证方法的步骤一一对应,在此不再赘述。
本申请的计算机可读存储介质,获取用户输入的手写签名与签章图案;识别手写签名从而获取签名文本;若签名文本与预设的姓名相同,则提取手写签名的字迹特征,将字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到字迹识别模型输出的识别结果;若识别结果与签名文本相同,则获取预存的虚拟立体印章在指定投影方向的指定投影图案;若指定投影图案与签章图案相同,则判定用户的身份验证通过,从而提高了身份验证的准确性。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程,是可以通过计算机程序来指令相关的硬件来完成,所述的计算机程序可存储于一非易失性计算机可读取存储介质中,该计算机程序在执行时,可包括如上述各方法的实施例的流程。其中,本申请所提供的和实施例中所使用的对存储器、存储、数据库或其它介质的任何引用,均可包括非易失性和/或易失性存储器。非易失性存储器可以包括只读存储器(ROM)、可编程ROM(PROM)、电可编程ROM(EPROM)、电可擦除可编程ROM(EEPROM)或闪存。易失性存储器可包括随机存取存储器(RAM)或者外部高速缓冲存储器。作为说明而非局限,RAM以多种形式可得,诸如静态RAM(SRAM)、动态RAM(DRAM)、同步DRAM(SDRAM)、双速据率SDRAM(SSRSDRAM)、增强型SDRAM(ESDRAM)、同步链路(Synchlink)DRAM(SLDRAM)、存储器总线(Rambus)直接RAM(RDRAM)、直接存储器总线动态RAM(DRDRAM)、以及存储器总线动态RAM(RDRAM)等。

Claims (20)

  1. 一种基于印章和签名的身份验证方法,其中,包括:
    获取用户输入的手写签名与签章图案,其中所述签章图案是预存的虚拟立体印章的投影图案;
    采用预设的文字识别技术识别所述手写签名从而获取签名文本,并判断所述签名文本是否与预设的姓名相同;
    若所述签名文本与预设的姓名相同,则提取所述手写签名的字迹特征,将所述字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到所述字迹识别模型输出的识别结果,其中所述字迹识别模型基于预先采集的手写文字,以及与所述预先采集的手写文字对应的书写者组成的样本数据训练而成;
    判断所述识别结果是否与所述签名文本相同;
    若所述识别结果与所述签名文本相同,则获取预存的虚拟立体印章在指定投影方向的指定投影图案,并利用预设的图像相似判断方法判断所述指定投影图案与所述签章图案是否相同;
    若所述指定投影图案与所述签章图案相同,则判定所述用户的身份验证通过。
  2. 根据权利要求1所述的基于印章和签名的身份验证方法,其中,所述手写签名位于指定图片中,所述采用预设的文字识别技术识别所述手写签名从而获取签名文本的步骤,包括:
    采集所述指定图片中的像素点的RGB颜色模型中的R颜色通道的数值、G颜色通道的数值和B颜色通道的数值,并根据公式:F1=MIN{ROUND[(a1R+a2G+a3B)/L,0],A},获取参考数值F1,其中MIN为最小值函数,ROUND为四舍五入函数,a1、a2、a3和L均为预设参数,a1、a2、a3均为大于0且小于L的正数,L为大于0的整数,A为预设的取值在范围(0,255)之内阈值参数,R、G、B分别为所述指定图片中的指定像素点的RGB颜色模型中的R颜色通道的数值、G颜色通道的数值和B颜色通道的数值;
    获取所述参考数值F1的值不等于A的像素点,记为手写签名像素点,并将所述手写签名像素点构成的图形记为手写签名图形;
    提取所述手写签名图形的文字特征,并输入预设的支持向量机中进行分类,从而获得识别而得的手写文字文本和印刷体文字文本。
  3. 根据权利要求1所述的基于印章和签名的身份验证方法,其中,所述若所述签名文本与预设的姓名相同,则提取所述手写签名的字迹特征,将所述字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到所述字迹识别模型输出的识别结果,其中所述字迹识别模型基于预先采集的手写文字,以及与所述预先采集的手写文字对应的书写者组成的样本数据训练而成的步骤之前,包括:
    调用预先采集的样本数据,并分成训练集和测试集;其中,所述样本数据包括预先采集的手写文字,以及与所述预先采集的手写文字对应的书写者;
    将训练集的样本数据输入到预设的神经网络模型中进行训练,从而得到初始字迹识别模型,其中,训练的过程中采用随机梯度下降法;
    利用测试集的样本数据验证所述初始字迹识别模型;
    若所述初始字迹识别模型验证通过,则将所述初始字迹识别模型记为所述字迹识别模型。
  4. 根据权利要求1所述的基于印章和签名的身份验证方法,其中,所述手写签名位于指定图片中,所述提取所述手写签名的字迹特征的步骤,包括:
    获取所述手写签名对应的像素点,并记为手写签名像素点;
    获取所述手写签名像素点的颜色数值,并将颜色数值处于同一预设范围内的相邻像素点记为细节单位,并将所述细节单位的颜色数值记为所述相邻像素点的颜色数值的平均值;
    获取相邻的所述细节单位的颜色数值变化趋势,将所述细节单位、所述细节单位的颜色数值和所述颜色数值变化趋势作为所述手写签名的字迹特征,并提取所述字迹特征。
  5. 根据权利要求1所述的基于印章和签名的身份验证方法,其中,所述获取用户输入的手写签名与签章图案的步骤,包括:
    获取用户输入的手写签名、签章图案和签章图案的生成时间;
    所述获取预存的虚拟立体印章在指定投影方向的指定投影图案的步骤,包括:
    根据预设的签名与虚拟立体印章的对应关系,调取所述手写签名对应的指定虚拟立体印章;
    以所述指定虚拟立体印章的正面中心为原点,根据预设的时间与空间坐标点的对应关系,获取与所述签章图案的生成时间对应的指定坐标点;
    将所述指定坐标点指向所述原点的方向记为指定投影方向,并从所述指定投影方向对所述指定虚拟立体印章进行投影,从而得到指定投影图案。
  6. 根据权利要求5所述的基于印章和签名的身份验证方法,其中,所述以所述指定虚拟立体印章的正面中心为原点,根据预设的时间与空间坐标点的对应关系,获取与所述签章图案的生成时间对应的指定坐标点的步骤,包括:
    以所述指定虚拟立体印章的正面中心为原点,所述原点与所述正面中的预设点的连线作为x轴,所述正面中与所述x轴垂直的且过原点的直线作为y轴,所述正面的过所述原点的垂线为z轴,从而建立平面直角坐标系;
    获取当前时间,并根据公式:
    x=k1×M+b1;y=k2×D+b2;z=k3×T+b3,获取指定坐标点(x,y,z),其中所述签章图案的生成时间为签章年份的第M月中的第D天中的第T个小时,其中k1、k2、k3、b1、b2和b3均为预设的参数。
  7. 根据权利要求1所述的基于印章和签名的身份验证方法,其中,所述利用预设的图像相似判断方法判断所述指定投影图案与所述签章图案是否相同的步骤,包括:
    分别对所述指定投影图案与所述签章图案进行灰度化处理,得到第一灰度图片和第二灰度图片;
    计算灰度图片的第m列或者第m行的所有像素点的灰度值的平均值Am,以及计算灰度图片中所有像素点的灰度值的平均值B;
    根据公式:
    Figure PCTCN2020088000-appb-100001
    计算灰度图片的第m列或者第m行的总体方差
    Figure PCTCN2020088000-appb-100002
    其中N为灰度图片中的列或者行的总数量;
    根据公式:
    Figure PCTCN2020088000-appb-100003
    获得两张灰度图片的第m列或者第m行的总体方差之差
    Figure PCTCN2020088000-appb-100004
    其中,
    Figure PCTCN2020088000-appb-100005
    为第一张灰度图片的第m列或者第m行的总体方差,
    Figure PCTCN2020088000-appb-100006
    为第二张灰度图片的第m列或者第m行的总体方差;
    判断
    Figure PCTCN2020088000-appb-100007
    是否小于预设的方差误差阈值;
    Figure PCTCN2020088000-appb-100008
    小于预设的方差误差阈值,则判定所述指定投影图案与所述签章图案相同。
  8. 一种基于印章和签名的身份验证装置,其中,包括:
    获取单元,用于获取用户输入的手写签名与签章图案,其中所述签章图案是预存的虚拟立体印章的投影图案;
    姓名判断单元,用于采用预设的文字识别技术识别所述手写签名从而获取签名文本,并判断所述签名文本是否与预设的姓名相同;
    字迹识别单元,用于若所述签名文本与预设的姓名相同,则提取所述手写签名的字迹特征,将所述字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到所述字迹识别模型输出的识别结果,其中所述字迹识别模型基于预先采集的手写文字,以及与所述预先采集的手写文字对应的书写者组成的样本数据训练而成;
    签名文本相同判断单元,用于判断所述识别结果是否与所述签名文本相同;
    签章图案判断单元,用于若所述识别结果与所述签名文本相同,则获取预存的虚拟立体印章在指定投影方向的指定投影图案,并利用预设的图像相似判断方法判断所述指定投影图案与所述签章图案是否相同;
    身份验证通过判定单元,用于若所述指定投影图案与所述签章图案相同,则判定所述用户的身份验证通过。
  9. 一种计算机设备,其中,包括:
    一个或多个处理器;
    存储器;
    一个或多个计算机程序,其中所述一个或多个计算机程序被存储在所述存储器中并被配置为由所述一个或多个处理器执行,所述一个或多个计算机程序配置用于执行一种基于印章和签名的身份验证方法:
    其中,所述基于印章和签名的身份验证方法包括:
    获取用户输入的手写签名与签章图案,其中所述签章图案是预存的虚拟立 体印章的投影图案;
    采用预设的文字识别技术识别所述手写签名从而获取签名文本,并判断所述签名文本是否与预设的姓名相同;
    若所述签名文本与预设的姓名相同,则提取所述手写签名的字迹特征,将所述字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到所述字迹识别模型输出的识别结果,其中所述字迹识别模型基于预先采集的手写文字,以及与所述预先采集的手写文字对应的书写者组成的样本数据训练而成;
    判断所述识别结果是否与所述签名文本相同;
    若所述识别结果与所述签名文本相同,则获取预存的虚拟立体印章在指定投影方向的指定投影图案,并利用预设的图像相似判断方法判断所述指定投影图案与所述签章图案是否相同;
    若所述指定投影图案与所述签章图案相同,则判定所述用户的身份验证通过。
  10. 根据权利要求9所述的计算机设备,其中,所述手写签名位于指定图片中,所述采用预设的文字识别技术识别所述手写签名从而获取签名文本的步骤,包括:
    采集所述指定图片中的像素点的RGB颜色模型中的R颜色通道的数值、G颜色通道的数值和B颜色通道的数值,并根据公式:F1=MIN{ROUND[(a1R+a2G+a3B)/L,0],A},获取参考数值F1,其中MIN为最小值函数,ROUND为四舍五入函数,a1、a2、a3和L均为预设参数,a1、a2、a3均为大于0且小于L的正数,L为大于0的整数,A为预设的取值在范围(0,255)之内阈值参数,R、G、B分别为所述指定图片中的指定像素点的RGB颜色模型中的R颜色通道的数值、G颜色通道的数值和B颜色通道的数值;
    获取所述参考数值F1的值不等于A的像素点,记为手写签名像素点,并将所述手写签名像素点构成的图形记为手写签名图形;
    提取所述手写签名图形的文字特征,并输入预设的支持向量机中进行分类,从而获得识别而得的手写文字文本和印刷体文字文本。
  11. 根据权利要求9所述的计算机设备,其中,所述若所述签名文本与预设的姓名相同,则提取所述手写签名的字迹特征,将所述字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到所述字迹识别模型输出的识别结果,其中所述字迹识别模型基于预先采集的手写文字,以及与所述预先采集的手写文字对应的书写者组成的样本数据训练而成的步骤之前,包括:
    调用预先采集的样本数据,并分成训练集和测试集;其中,所述样本数据包括预先采集的手写文字,以及与所述预先采集的手写文字对应的书写者;
    将训练集的样本数据输入到预设的神经网络模型中进行训练,从而得到初始字迹识别模型,其中,训练的过程中采用随机梯度下降法;
    利用测试集的样本数据验证所述初始字迹识别模型;
    若所述初始字迹识别模型验证通过,则将所述初始字迹识别模型记为所述字迹识别模型。
  12. 根据权利要求9所述的计算机设备,其中,所述手写签名位于指定图片中,所述提取所述手写签名的字迹特征的步骤,包括:
    获取所述手写签名对应的像素点,并记为手写签名像素点;
    获取所述手写签名像素点的颜色数值,并将颜色数值处于同一预设范围内的相邻像素点记为细节单位,并将所述细节单位的颜色数值记为所述相邻像素点的颜色数值的平均值;
    获取相邻的所述细节单位的颜色数值变化趋势,将所述细节单位、所述细节单位的颜色数值和所述颜色数值变化趋势作为所述手写签名的字迹特征,并提取所述字迹特征。
  13. 根据权利要求9所述的计算机设备,其中,所述获取用户输入的手写签名与签章图案的步骤,包括:
    获取用户输入的手写签名、签章图案和签章图案的生成时间;
    所述获取预存的虚拟立体印章在指定投影方向的指定投影图案的步骤,包括:
    根据预设的签名与虚拟立体印章的对应关系,调取所述手写签名对应的指定虚拟立体印章;
    以所述指定虚拟立体印章的正面中心为原点,根据预设的时间与空间坐标点的对应关系,获取与所述签章图案的生成时间对应的指定坐标点;
    将所述指定坐标点指向所述原点的方向记为指定投影方向,并从所述指定投影方向对所述指定虚拟立体印章进行投影,从而得到指定投影图案。
  14. 根据权利要求13所述的计算机设备,其中,所述以所述指定虚拟立体印章的正面中心为原点,根据预设的时间与空间坐标点的对应关系,获取与所述签章图案的生成时间对应的指定坐标点的步骤,包括:
    以所述指定虚拟立体印章的正面中心为原点,所述原点与所述正面中的预设点的连线作为x轴,所述正面中与所述x轴垂直的且过原点的直线作为y轴,所述正面的过所述原点的垂线为z轴,从而建立平面直角坐标系;
    获取当前时间,并根据公式:
    x=k1×M+b1;y=k2×D+b2;z=k3×T+b3,获取指定坐标点(x,y,z),其中所述签章图案的生成时间为签章年份的第M月中的第D天中的第T个小时,其中k1、k2、k3、b1、b2和b3均为预设的参数。
  15. 根据权利要求9所述的计算机设备,其中,所述利用预设的图像相似判断方法判断所述指定投影图案与所述签章图案是否相同的步骤,包括:
    分别对所述指定投影图案与所述签章图案进行灰度化处理,得到第一灰度图片和第二灰度图片;
    计算灰度图片的第m列或者第m行的所有像素点的灰度值的平均值Am,以及计算灰度图片中所有像素点的灰度值的平均值B;
    根据公式:
    Figure PCTCN2020088000-appb-100009
    计算灰度图片的第m列或者第m行的总体方差
    Figure PCTCN2020088000-appb-100010
    其中N为灰度图片中的列或者行的总数量;
    根据公式:
    Figure PCTCN2020088000-appb-100011
    获得两张灰度图片的第m列或者第m行的总体方差之差
    Figure PCTCN2020088000-appb-100012
    其中,
    Figure PCTCN2020088000-appb-100013
    为第一张灰度图片的第m列或者第m行的总体方差,
    Figure PCTCN2020088000-appb-100014
    为第二张灰度图片的第m列或者第m行的总体方差;
    判断
    Figure PCTCN2020088000-appb-100015
    是否小于预设的方差误差阈值;
    Figure PCTCN2020088000-appb-100016
    小于预设的方差误差阈值,则判定所述指定投影图案与所述签章图案相同。
  16. 一种计算机可读存储介质,其中,所述计算机可读存储介质上存储有计算机程序,该计算机程序被处理器执行时实现基于印章和签名的身份验证方法,其中,所述基于印章和签名的身份验证方法包括以下步骤:
    获取用户输入的手写签名与签章图案,其中所述签章图案是预存的虚拟立体印章的投影图案;
    采用预设的文字识别技术识别所述手写签名从而获取签名文本,并判断所述签名文本是否与预设的姓名相同;
    若所述签名文本与预设的姓名相同,则提取所述手写签名的字迹特征,将所述字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到所述字迹识别模型输出的识别结果,其中所述字迹识别模型基于预先采集的手写文字,以及与所述预先采集的手写文字对应的书写者组成的样本数据训练而成;
    判断所述识别结果是否与所述签名文本相同;
    若所述识别结果与所述签名文本相同,则获取预存的虚拟立体印章在指定投影方向的指定投影图案,并利用预设的图像相似判断方法判断所述指定投影图案与所述签章图案是否相同;
    若所述指定投影图案与所述签章图案相同,则判定所述用户的身份验证通过。
  17. 根据权利要求16所述的计算机可读存储介质,其中,所述手写签名位于指定图片中,所述采用预设的文字识别技术识别所述手写签名从而获取签名文本的步骤,包括:
    采集所述指定图片中的像素点的RGB颜色模型中的R颜色通道的数值、G颜色通道的数值和B颜色通道的数值,并根据公式:F1=MIN{ROUND[(a1R+a2G+a3B)/L,0],A},获取参考数值F1,其中MIN为最小值函数,ROUND为四舍五入函数,a1、a2、a3和L均为预设参数,a1、a2、a3均为大于0且小于L的正数,L为大于0的整数,A为预设的取值在范围(0,255)之内阈值参数,R、G、B分别为所述指定图片中的指定像素点的RGB颜色模型中的R颜色通道的数值、G颜色通道的数值和B颜色通道的数值;
    获取所述参考数值F1的值不等于A的像素点,记为手写签名像素点,并将所述手写签名像素点构成的图形记为手写签名图形;
    提取所述手写签名图形的文字特征,并输入预设的支持向量机中进行分类,从而获得识别而得的手写文字文本和印刷体文字文本。
  18. 根据权利要求16所述的计算机可读存储介质,其中,所述若所述签名文本与预设的姓名相同,则提取所述手写签名的字迹特征,将所述字迹特征输入基于神经网络模型训练完成的字迹识别模型中进行计算,从而得到所述字迹识别模型输出的识别结果,其中所述字迹识别模型基于预先采集的手写文字,以及与所述预先采集的手写文字对应的书写者组成的样本数据训练而成的步骤之前,包括:
    调用预先采集的样本数据,并分成训练集和测试集;其中,所述样本数据包括预先采集的手写文字,以及与所述预先采集的手写文字对应的书写者;
    将训练集的样本数据输入到预设的神经网络模型中进行训练,从而得到初始字迹识别模型,其中,训练的过程中采用随机梯度下降法;
    利用测试集的样本数据验证所述初始字迹识别模型;
    若所述初始字迹识别模型验证通过,则将所述初始字迹识别模型记为所述字迹识别模型。
  19. 根据权利要求16所述的计算机可读存储介质,其中,所述手写签名位于指定图片中,所述提取所述手写签名的字迹特征的步骤,包括:
    获取所述手写签名对应的像素点,并记为手写签名像素点;
    获取所述手写签名像素点的颜色数值,并将颜色数值处于同一预设范围内的相邻像素点记为细节单位,并将所述细节单位的颜色数值记为所述相邻像素点的颜色数值的平均值;
    获取相邻的所述细节单位的颜色数值变化趋势,将所述细节单位、所述细节单位的颜色数值和所述颜色数值变化趋势作为所述手写签名的字迹特征,并提取所述字迹特征。
  20. 根据权利要求16所述的计算机可读存储介质,其中,所述获取用户输入的手写签名与签章图案的步骤,包括:
    获取用户输入的手写签名、签章图案和签章图案的生成时间;
    所述获取预存的虚拟立体印章在指定投影方向的指定投影图案的步骤,包括:
    根据预设的签名与虚拟立体印章的对应关系,调取所述手写签名对应的指定虚拟立体印章;
    以所述指定虚拟立体印章的正面中心为原点,根据预设的时间与空间坐标点的对应关系,获取与所述签章图案的生成时间对应的指定坐标点;
    将所述指定坐标点指向所述原点的方向记为指定投影方向,并从所述指定投影方向对所述指定虚拟立体印章进行投影,从而得到指定投影图案。
PCT/CN2020/088000 2019-08-14 2020-04-30 基于印章和签名的身份验证方法、装置和计算机设备 WO2021027336A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910750602.1 2019-08-14
CN201910750602.1A CN110619274A (zh) 2019-08-14 2019-08-14 基于印章和签名的身份验证方法、装置和计算机设备

Publications (1)

Publication Number Publication Date
WO2021027336A1 true WO2021027336A1 (zh) 2021-02-18

Family

ID=68921917

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/088000 WO2021027336A1 (zh) 2019-08-14 2020-04-30 基于印章和签名的身份验证方法、装置和计算机设备

Country Status (2)

Country Link
CN (1) CN110619274A (zh)
WO (1) WO2021027336A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113158887A (zh) * 2021-04-19 2021-07-23 福建捷宇电脑科技有限公司 一种提高电子签名识别准率的电子签名认证方法及设备
CN113591155A (zh) * 2021-06-17 2021-11-02 傲雄在线(重庆)科技有限公司 一种电子印章授权方法、系统、电子设备及存储介质
CN114943063A (zh) * 2022-03-04 2022-08-26 杭州京胜航星科技有限公司 一种基于互联网的电子印章生成与识别系统
CN115035527A (zh) * 2022-08-10 2022-09-09 北京佳萌锐普科技发展有限公司 一种识别电子签章位置的方法、装置及设备

Families Citing this family (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110619274A (zh) * 2019-08-14 2019-12-27 深圳壹账通智能科技有限公司 基于印章和签名的身份验证方法、装置和计算机设备
CN111324589A (zh) * 2020-02-14 2020-06-23 中国平安人寿保险股份有限公司 基于相似程度值的日志发送方法、装置和计算机设备
CN111523537A (zh) * 2020-04-13 2020-08-11 联讯益康医疗信息技术(武汉)有限公司 一种文字识别方法、存储介质及系统
CN111832547A (zh) * 2020-06-24 2020-10-27 平安普惠企业管理有限公司 文字识别模型的动态部署方法、装置和计算机设备
TWI777188B (zh) * 2020-07-07 2022-09-11 新光人壽保險股份有限公司 契約簽名鑑別方法及其裝置
CN111931672A (zh) * 2020-08-17 2020-11-13 珠海大横琴科技发展有限公司 字迹识别方法、装置、计算机设备及存储介质
CN112036323B (zh) * 2020-09-01 2024-02-27 中国银行股份有限公司 签名笔迹鉴别方法、客户端及服务器
CN112580108B (zh) * 2020-12-10 2024-04-19 深圳证券信息有限公司 签名和印章完整性验证方法及计算机设备
CN112861649B (zh) * 2021-01-19 2024-05-10 平安科技(深圳)有限公司 指纹签名生成方法、装置、电子设备及计算机存储介质
CN113361547B (zh) * 2021-06-30 2024-06-14 深圳证券信息有限公司 一种签章识别方法、装置、设备及可读存储介质
CN113920589A (zh) * 2021-10-28 2022-01-11 平安银行股份有限公司 基于人工智能的签名识别方法、装置、设备及介质
CN114241463A (zh) * 2021-11-12 2022-03-25 中国南方电网有限责任公司 签名验证方法、装置、计算机设备和存储介质
CN114021099B (zh) * 2022-01-06 2022-03-25 北科信链(武汉)数字科技有限公司 一种基于指纹与印章的验证方法及装置
CN114037998A (zh) * 2022-01-11 2022-02-11 北京顶象技术有限公司 一种使用模型分析笔迹提升账户安全的方法及装置
CN115657900B (zh) * 2022-07-15 2023-09-22 内蒙古师范大学 基于自动翻译及文本转换的多文种私章定制方法与装置

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101008985A (zh) * 2007-01-18 2007-08-01 章毅 印章鉴别系统及其控制方法
CN106778586A (zh) * 2016-12-08 2017-05-31 武汉理工大学 离线手写签名鉴别方法及系统
CN109033954A (zh) * 2018-06-15 2018-12-18 西安科技大学 一种基于机器视觉的空中手写识别系统及方法
JP2019075045A (ja) * 2017-10-19 2019-05-16 株式会社 みずほ銀行 印鑑照合支援システム、印鑑照合支援方法及び印鑑照合支援プログラム
CN109934262A (zh) * 2019-01-31 2019-06-25 平安科技(深圳)有限公司 图片差异性判断方法、装置、计算机设备和存储介质
CN110619274A (zh) * 2019-08-14 2019-12-27 深圳壹账通智能科技有限公司 基于印章和签名的身份验证方法、装置和计算机设备

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101008985A (zh) * 2007-01-18 2007-08-01 章毅 印章鉴别系统及其控制方法
CN106778586A (zh) * 2016-12-08 2017-05-31 武汉理工大学 离线手写签名鉴别方法及系统
JP2019075045A (ja) * 2017-10-19 2019-05-16 株式会社 みずほ銀行 印鑑照合支援システム、印鑑照合支援方法及び印鑑照合支援プログラム
CN109033954A (zh) * 2018-06-15 2018-12-18 西安科技大学 一种基于机器视觉的空中手写识别系统及方法
CN109934262A (zh) * 2019-01-31 2019-06-25 平安科技(深圳)有限公司 图片差异性判断方法、装置、计算机设备和存储介质
CN110619274A (zh) * 2019-08-14 2019-12-27 深圳壹账通智能科技有限公司 基于印章和签名的身份验证方法、装置和计算机设备

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113158887A (zh) * 2021-04-19 2021-07-23 福建捷宇电脑科技有限公司 一种提高电子签名识别准率的电子签名认证方法及设备
CN113591155A (zh) * 2021-06-17 2021-11-02 傲雄在线(重庆)科技有限公司 一种电子印章授权方法、系统、电子设备及存储介质
CN113591155B (zh) * 2021-06-17 2024-04-05 重庆亲笔签数字科技有限公司 一种电子印章授权方法、系统、电子设备及存储介质
CN114943063A (zh) * 2022-03-04 2022-08-26 杭州京胜航星科技有限公司 一种基于互联网的电子印章生成与识别系统
CN115035527A (zh) * 2022-08-10 2022-09-09 北京佳萌锐普科技发展有限公司 一种识别电子签章位置的方法、装置及设备
CN115035527B (zh) * 2022-08-10 2022-11-11 北京佳萌锐普科技发展有限公司 一种识别电子签章位置的方法、装置及设备

Also Published As

Publication number Publication date
CN110619274A (zh) 2019-12-27

Similar Documents

Publication Publication Date Title
WO2021027336A1 (zh) 基于印章和签名的身份验证方法、装置和计算机设备
CN111401257B (zh) 一种基于余弦损失在非约束条件下的人脸识别方法
Atoum et al. Face anti-spoofing using patch and depth-based CNNs
US10635946B2 (en) Eyeglass positioning method, apparatus and storage medium
CN106951832B (zh) 一种基于手写字符识别的验证方法及装置
WO2019232862A1 (zh) 嘴巴模型训练方法、嘴巴识别方法、装置、设备及介质
CN110287971B (zh) 数据验证方法、装置、计算机设备及存储介质
CN111191568B (zh) 翻拍图像识别方法、装置、设备及介质
WO2016149944A1 (zh) 用于识别人脸的方法、系统和计算机程序产品
CN110569850A (zh) 字符识别模板匹配方法、装置和文本识别设备
US9489561B2 (en) Method and system for estimating fingerprint pose
CN111275685B (zh) 身份证件的翻拍图像识别方法、装置、设备及介质
WO2021042505A1 (zh) 基于文字识别技术的笔记生成方法、装置和计算机设备
CN105335719A (zh) 活体检测方法及装置
WO2019232850A1 (zh) 手写汉字图像识别方法、装置、计算机设备及存储介质
CN110838119A (zh) 人脸图像质量评估方法、计算机装置及计算机可读存储介质
CN109376717A (zh) 人脸对比的身份识别方法、装置、电子设备及存储介质
US10423817B2 (en) Latent fingerprint ridge flow map improvement
CN113011144A (zh) 表单信息的获取方法、装置和服务器
CN113111880B (zh) 证件图像校正方法、装置、电子设备及存储介质
CN110942067A (zh) 文本识别方法、装置、计算机设备和存储介质
CN112232336A (zh) 一种证件识别方法、装置、设备及存储介质
CN113361666B (zh) 一种手写字符识别方法、系统及介质
Miroslav et al. Basic on-line handwritten signature features for personal biometric authentication
CN111881880A (zh) 一种基于新型网络的票据文本识别方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20853440

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20853440

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 04.08.2022)

122 Ep: pct application non-entry in european phase

Ref document number: 20853440

Country of ref document: EP

Kind code of ref document: A1