CN107657241B - Signature pen-oriented signature authenticity identification system - Google Patents
Signature pen-oriented signature authenticity identification system Download PDFInfo
- Publication number
- CN107657241B CN107657241B CN201710933556.XA CN201710933556A CN107657241B CN 107657241 B CN107657241 B CN 107657241B CN 201710933556 A CN201710933556 A CN 201710933556A CN 107657241 B CN107657241 B CN 107657241B
- Authority
- CN
- China
- Prior art keywords
- information
- pen
- signature
- sample
- characteristic
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/30—Writer recognition; Reading and verifying signatures
- G06V40/37—Writer recognition; Reading and verifying signatures based only on signature signals such as velocity or pressure, e.g. dynamic signature recognition
- G06V40/376—Acquisition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/30—Writer recognition; Reading and verifying signatures
- G06V40/37—Writer recognition; Reading and verifying signatures based only on signature signals such as velocity or pressure, e.g. dynamic signature recognition
- G06V40/382—Preprocessing; Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/30—Writer recognition; Reading and verifying signatures
- G06V40/37—Writer recognition; Reading and verifying signatures based only on signature signals such as velocity or pressure, e.g. dynamic signature recognition
- G06V40/394—Matching; Classification
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Human Computer Interaction (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Collating Specific Patterns (AREA)
Abstract
The invention discloses a signature pen-oriented signature authenticity identification system. The system comprises an information input module, a sample database module, a characteristic model construction module, a characteristic extraction module, a similarity measurement module and an identification and output module. The identification process is that the characteristic model building module carries out space modeling on the characteristic information collected by the information input module, and characteristic parameters of the change relation between the writing horizontal pressure and the pen holding posture along with time and the writing length are extracted from the space geometric model to build a matrix input system. And calculating the weight coefficient of each type of feature by using the average information entropy of each type of feature, and calculating the comprehensive feature similarity of the information to be detected and the sample information by combining a matrix similarity measurement method. And in the identification and output module, comparing the comprehensive similarity with similarity thresholds corresponding to different signers, identifying the authenticity of the signature and outputting a final identification result. The signature authentication and identity authentication method provided by the invention has objective and stable result.
Description
Technical Field
The invention relates to a user signature authenticity identification system for a sign pen, and belongs to the technical field of digital image processing and signature authenticity identification.
Background
The 21 st century is an information era, key industries and enterprises such as finance, medical treatment, government and public utilities are comprehensively covered by big data, and the appearance of various digital products brings efficient and convenient life to people, greatly improves the life quality of people, but simultaneously inevitably brings about a plurality of problems in the aspect of information safety. The identity authentication of personnel involved in daily transactions and affairs is one of the problems that needs to be solved urgently. With the intensive research on identification techniques based on human body biological characteristics, the human body biological characteristics currently investigated by researchers have started to shift from the inherent characteristics of personal DNA, fingerprints, retina, etc. to the acquired characteristics such as signature, voice, gait, etc. Among the acquired features, the hand-written signature is influenced by the nervous system and muscle reflection of human and is related to the habit, character and accepted writing training of the writer, so that the hand-written signature has strong stability and diversity and can effectively realize the identification of the identity of the signer.
At present, researchers divide handwritten signature identification into an online mode and an offline mode according to different signature obtaining modes, the offline signature identification mostly aims at identification of static characteristics such as strokes and tracks of fonts, and the online signature identification utilizes online advantages of the offline signature identification to investigate dynamic characteristics of the fonts, such as pen point speed and acceleration. However, in both off-line signature and on-line signature, current researchers are focusing on the research on the signature font itself, and neglecting other features that may be generated with the signature and reflect the authenticity of the signature. The sign pen is used as a signature tool and is indispensable equipment in any signature process, and because the habits and the accepted training of each person are different, the characteristics of the pen holding posture, the force and the like are different during signature, and the sign pen is similar to the handwriting characteristics and has stronger stability and diversity. Compared with the dynamic characteristics of handwriting, such as speed, acceleration and the like, the signature pen is more convenient and faster to extract the characteristics. Therefore, if the characteristics extracted from the sign pen are utilized, the accuracy of signature authentication can be improved, and the research direction of signature authentication technology can be expanded. However, no researcher focuses on the characteristics obtained from the sign pen, and no research on signature authentication technology oriented to the sign pen exists at present, whether the research is theoretical research or practical application.
Disclosure of Invention
The invention provides a signature pen-oriented signature authenticity identification system aiming at the technical defects in the aspects of handwritten signature authenticity identification and identity authentication popular in the current society, so as to further improve the accuracy of signature authenticity identification.
In order to solve the technical problem, the invention provides a signature pen-oriented signature authenticity identification system which is characterized by comprising an information acquisition unit and the following program modules:
1) an information entry program module: inputting signature information and personal identity information of a signer to be tested, wherein the signature information comprises writing horizontal pressure dynamic characteristics, pen holding pressure dynamic characteristics, writing time and writing length of the signer during writing;
2) a sample database program module: the method comprises the steps that dynamic characteristic information and personal identity information, collected by a sign pen, of a sample signer during pen holding writing are stored in advance, wherein the dynamic characteristic information comprises writing horizontal pressure dynamic characteristics and pen holding pressure dynamic characteristics;
3) a characteristic model building program module: acquiring and constructing a writing horizontal pressure dynamic characteristic model and a pen holding pressure dynamic characteristic model when a signer holds a pen:
writing horizontal pressure dynamics: the screw thread cylinder is located sign pen refill lower extreme and refill laminating department, and screw thread cylindrical internal surface is provided with a plurality of pressure sensing device one along the circumferencial direction, because everybody's signature writes the difference of custom, and during the signature person's signature, the pressure change condition of the not equidirectional refill of the cylindrical internal surface of pen post screw thread is also different to have the anisotropy, consequently can gather the pressure change condition of the not equidirectional internal surface of pen post screw thread cylinder as writing horizontal pressure dynamic characteristic.
Specifically, the writing level pressure dynamic characteristic construction process comprises the following steps: the threaded cylinder is located at the joint of the lower end of the sign pen column and the pen core, a plurality of pressure sensing devices I are arranged on the inner surface of the threaded cylinder along the circumferential direction, the pressure sensing devices I collect pressure information applied by the pen core in a plurality of directions on the inner surface of the threaded cylinder during writing, writing horizontal pressure information with the pressure not being zero in the direction is calculated from the beginning of a pen-falling signature to the end of the signature, and writing horizontal pressure information, writing length and time change curves are recorded respectively and expressed in a three-dimensional space coordinate system for subsequent feature extraction and similarity measurement.
The dynamic characteristics of pen holding pressure: because everyone holds the pen custom different, when holding a pen signature, the change condition of finger to the position pressure size is held to the sign pen also different to holding a pen pressure dynamic characteristic and having stability and anisotropy, consequently, can gather the change condition of the pressure of the position and user's thumb and forefinger contact surface department are held to the pen post and regard as holding a pen pressure dynamic characteristic.
Specifically, the process for constructing the pen holding pressure dynamic characteristics comprises the following steps: collecting finger pressure information of pressure collection surfaces corresponding to the thumb and the forefinger, recording the change curves of the finger pressure information, writing length and time collected by the thumb pressure collection surface and the forefinger pressure collection surface from the beginning of pen-down signature to the end of signature, representing in a three-dimensional space coordinate system, and using for subsequent feature extraction and similarity measurement.
In the case of a pressure dynamic characteristic p at the writing level1Time t1Length of writing1In the formed space coordinate system xyz, sampling is carried out for multiple times at equal time intervals from the beginning of pen drop of a signer to the end of signature in the period of time, corresponding points on curves in three planes of xoz, xoy and yoz at a certain moment are connected at different time points to form a plurality of feature triangles reflecting the relation between writing horizontal pressure dynamic features and time and handwriting length, three vertexes of the feature triangles are connected with the origin of coordinates to form a plurality of feature tetrahedrons, and the feature tetrahedrons are used as feature models to carry out writing horizontal pressure dynamic feature extraction.
Similarly, in the dynamic characteristic p of the pen holding pressure2Time t2Length of writing2In the formed space coordinate system xyz, from the start of the pen drop of the signer to the end of the signature,sampling for multiple times at equal time intervals in the period of time, connecting corresponding points on curves in three planes of xoz, xoy and yoz at a certain moment at different time points to form a plurality of feature triangles reflecting the relation between the dynamic features of the pen holding pressure and the lengths of time and handwriting, connecting three vertexes of the feature triangles with the origin of coordinates to form a plurality of feature tetrahedrons, and extracting the dynamic features of the pen holding pressure by taking the feature tetrahedrons as feature models.
4) A feature extraction program module: and recording three side lengths, areas and volumes of the characteristic tetrahedrons of the corresponding characteristic triangles for the writing horizontal pressure dynamic characteristics and the pen holding pressure dynamic characteristics when the signer holds the pen, constructing a characteristic vector corresponding to the characteristics by using the 5 types of information, combining the characteristic vectors corresponding to a plurality of characteristic triangles, and constructing a matrix for similarity measurement and signature authenticity identification of the signer.
5) A similarity measurement module: for different signatories, respectively utilizing the writing level pressure dynamic characteristic and the pen holding pressure dynamic characteristic corresponding to p1X 5 matrix, p1And 5, for the number of the selected feature triangles, taking 5 pieces of information of the volume of a feature tetrahedron formed by connecting three side lengths, the area and three vertexes of the feature triangles with the origin, and performing similarity measurement on the information to be measured and the sample information to obtain the comprehensive similarity between the information to be measured and the sample information.
6) An authentication and output program module: setting a similarity threshold according to the minimum value of similarity measurement among a plurality of sample information of the same sample signer, comparing the magnitude relation between the comprehensive similarity and the similarity threshold, identifying the authenticity of the signer to be detected, if the comprehensive similarity is more than or equal to the comprehensive similarity threshold, considering that the signer to be detected is the same as the sample signer, and if the comprehensive similarity is less than the comprehensive similarity threshold, considering that the signer to be detected is different from the sample signer.
The invention achieves the following beneficial effects: the invention provides a signature pen-oriented signature authenticity identification system, which considers the problem of feature extraction from the viewpoint of a signature pen, extracts two dynamic features of writing horizontal pressure and pen holding pressure through the signature pen, gets rid of the dependence of most current signature authenticity identification methods on a handwriting structure, widens the research direction of signature authenticity identification, and has high accuracy and strong robustness.
Drawings
FIG. 1 is a schematic diagram of an external structure of a sign pen;
FIG. 2 is a schematic diagram of a system module architecture;
FIG. 3 is a schematic diagram of a system process;
FIG. 4 is a schematic cross-sectional view of a threaded cylinder;
FIG. 5 is a schematic cross-sectional view of a pen-holding portion;
FIG. 6 is a schematic diagram of a refinement algorithm template;
FIG. 7a is a schematic diagram of a writing level pressure signature model;
FIG. 7b is a diagram of a model of the pressure characteristics of the pen grip.
Detailed Description
The signature authenticity identification system for the sign pen is applied to the sign pen with the information acquisition function, the appearance structure of the sign pen is shown in figure 1, the sign pen comprises a pen holder, and a pen holding part at the lower part of the pen holder is provided with two pressure information acquisition surfaces, wherein the surface A is a thumb pressure acquisition surface, the surface B is a forefinger pressure acquisition surface, and the surface 1 is a prompting lamp, so that the holding position is guaranteed to be effective.
The program module structure is shown in fig. 2, and includes an information entry program module, a sample database program module, a feature model construction program module, a feature extraction program module, a similarity measurement program module, and an identification and output program module.
Fig. 3 is a processing process diagram of each module of the system, describing the information processing process of each module, and the processing process of each module is as follows:
(1) an information entry program module: at the beginning of a signature authenticity identification system, the identity information of a signer is input, firstly, the signature information and the non-signature information of the signer are input in real time through a signature pen and a signature board with an information acquisition function, a signer writes own signatures on the signature board for many times by using a specific signature pen, handwriting images at all moments are acquired from the beginning of the signature to the end of the signature, the non-handwriting information of the signer is acquired through the signature pen when the signer signs, and the acquired handwriting image information and the acquired non-handwriting information at different moments are transmitted to a sample database program module for storage and are used for construction of a subsequent characteristic model and extraction of corresponding characteristics. The non-handwriting information comprises signature information and personal identity information, and the signature information comprises writing level pressure dynamic characteristics, pen holding pressure dynamic characteristics, writing time and writing length of the signature handwriting written by the signer.
(2) A sample database program module: pre-storing handwriting signature images of sample signers at different moments, dynamic characteristic information during pen holding and personal identity information, classifying a sample library into K types according to the number K of people who enter the sample signers, wherein each type comprises s entered by the signer1Group information, recording each type of sample set as QkK is 1, 2, …, K, m is selected from the ith subclassiGroup information, mi=1,2,…,s1Constructing a special sample set M, M ═ qij|qij∈Qi,i=1,2,…,K,j=1,2…,miThe total number of sample sets isII denotes "taking the product", i.e. m0=m1×m2×…×mKThe numerical values below and above the symbol are the initial value and the final value of the product respectively, and the special sample set M is used for calculating the information entropy and the weight coefficient corresponding to various dynamic features and integrating the feature similarity measurement.
(3) For signature images acquired by an information input program module and corresponding to different moments, defining writing length as the number of character pixel points in a signature skeleton image, and calculating the change relation of the writing length along with time, wherein the method specifically comprises the following steps:
31) performing graying and binarization operation on the handwriting image at each moment to obtain a target binarization image;
32) then, carrying out standardization processing on the target binary image to obtain a standardized image with the pixel size of 800 × 600;
33) and (3) carrying out rapid refinement processing on the standardized image, adopting a classical rapid parallel refinement algorithm, stripping black pixels on the periphery of the stroke layer by layer, and reserving pixel points belonging to the skeleton. After the image is binarized, the pixel values are only 0 and 1, 0 is black, 1 is white, black is a signature handwriting part, and white is a signature background, wherein the black pixel refers to a pixel with the pixel value of 0.
In order to ensure the continuity of the skeleton, each iteration is divided into a secondary sub-process, a binarized Chinese character image is marked as shown in fig. 6 for a certain inspection point P by setting a foreground point value of the binarized Chinese character image to be 1 and a background point value of the binarized Chinese character image to be 0, each foreground point in the Chinese character image is inspected in a circulating mode in the first sub-process, pixel points which simultaneously meet the following conditions (1), (2), (3) and (4) are marked as deletable points, and when one-time circulating inspection is finished, all points which are marked as the deletable points are deleted; in the second sub-process, each foreground point in the Chinese character image is inspected in a circulating way, pixel points which simultaneously satisfy the following conditions (1), (2), (5) and (6) are marked as deletable points, when one-time circulation inspection is finished, all the points marked as the deletable points are deleted, the sub-process is repeatedly executed for two times until no deletable pixel point exists,
(1)2≤B(P)≤6;
(2)A(P)=1;
(3)P1×P3×P5=0;
(4)P3×P5×P7=0;
(5)P1×P3×P7=0;
(6)P1×P5×P7=0。
wherein A (P) represents the number of P1, P2, … and P8, wherein 01 pairs in the P1 sequence appear, and B (P) is the number of 1 in an eight-neighborhood window of the point P; 0 represents a white background, 1 represents a black handwriting, and 01 represents an edge point of a signature handwriting;
condition (1) is to retain isolated point (b) (p) 0), end point (b (p) 1), approximate inner point (b (p) 7), and inner point (b (p) 8);
the condition (2) is to retain the skeletal points;
the conditions (3) and (4) are to ensure that the right boundary, the lower boundary, and the upper left corner are deleted;
conditions (5) and (6) are to ensure that the left boundary, the upper boundary, and the lower right corner are deleted;
34) and for skeleton images at different moments obtained by thinning, calculating the number of character pixel points, and fitting by using a least square method to obtain a change relation curve of the length of the written handwriting along with time.
(4) A characteristic model building program module: and acquiring writing horizontal pressure dynamic characteristics and pen holding pressure dynamic characteristics of a signer when holding a pen.
Writing horizontal pressure dynamics: as shown in figure 4, the screw thread cylinder is located sign pen refill laminating department of pen lower extreme, and screw thread cylinder's internal surface is provided with a plurality of pressure sensing device one along the circumferencial direction, because everyone's signature writes the difference of custom, and during the signature person's signature, the pressure change condition that the different directions of the cylindrical internal surface of pen post screw thread received the refill also is different to have the difference in nature, consequently can gather the pressure change condition of the different directions of the cylindrical internal surface of pen post screw thread cylinder as writing horizontal pressure dynamic characteristic.
Specifically, the threaded cylinder is located at the joint of the lower end of the sign pen column and the pen core, the inner surface of the threaded cylinder is provided with a plurality of first pressure sensing devices in the circumferential direction, the first pressure sensing devices collect pressure information applied by the pen core in a plurality of directions on the inner surface of the threaded cylinder during writing, writing horizontal pressure information with pressure not equal to zero in the direction is calculated from the beginning of a pen-falling signature to the end of the signature, and the writing horizontal pressure information, writing length and time change curves are recorded respectively and expressed in a three-dimensional space coordinate system for subsequent feature extraction and similarity measurement.
The dynamic characteristics of pen holding pressure: because everyone holds the pen custom different, when holding a pen signature, the change condition of finger to the position pressure size is held to the sign pen also different to holding a pen pressure dynamic characteristic and having stability and anisotropy, consequently, can gather the change condition of the pressure of the position and user's thumb and forefinger contact surface department are held to the pen post and regard as holding a pen pressure dynamic characteristic.
Specifically, finger pressure information of pressure acquisition surfaces corresponding to the thumb and the forefinger is acquired, change curves of the finger pressure information, writing length and time acquired by the thumb pressure acquisition surface and the forefinger pressure acquisition surface are recorded from the beginning of pen-down signature to the end of signature, and are expressed in a three-dimensional space coordinate system for subsequent feature extraction and similarity measurement.
Specifically, a schematic transverse cross-sectional view of a pen holding part of the sign pen is shown in fig. 5, wherein a surface a in fig. 5 is a thumb pressure information acquisition surface, a surface B is a forefinger pressure acquisition surface, and a position 1 is a prompting lamp, so that the holding position is guaranteed to be effective. Collecting finger pressure information of pressure collection surfaces corresponding to the thumb and the forefinger, recording the change curves of the finger pressure information, writing length and time collected by the thumb pressure collection surface and the forefinger pressure collection surface from the beginning of pen-down signature to the end of signature, representing in a three-dimensional space coordinate system, and using for subsequent feature extraction and similarity measurement.
In the case of a pressure dynamic characteristic p at the writing level1Time t1Length of writing1In the formed space coordinate system xyz, sampling is carried out for multiple times at equal time intervals from the beginning of pen drop of a signer to the end of signature in the period of time, corresponding points on curves in three planes of xoz, xoy and yoz at a certain moment are connected at different time points to form a plurality of feature triangles reflecting the relation between writing horizontal pressure dynamic features and time and handwriting length, three vertexes of the feature triangles are connected with the origin of coordinates to form a plurality of feature tetrahedrons, and the feature tetrahedrons are used as feature models to carry out writing horizontal pressure dynamic feature extraction.
Similarly, in the dynamic characteristic p of the pen holding pressure2Time t2Length of writing2In the formed space coordinate system xyz, sampling is carried out for multiple times at equal time intervals from the beginning of pen drop of a signer to the end of signature in the period of time, corresponding points on curves in three planes of xoz, xoy and yoz at a certain moment are connected at different time points to form a plurality of feature triangles reflecting the relation between the dynamic feature of pen holding pressure and the lengths of time and handwriting, and three vertexes of the feature triangles are connected with the origin of coordinates to form a plurality of feature trianglesAnd the characteristic tetrahedron is used as a characteristic model to extract the dynamic characteristics of the pen holding pressure.
As shown in the characteristic model diagram of fig. 7, according to the relationship between the writing horizontal pressure dynamic characteristic and time as well as the writing length when the signer holds the pen, the relationship between the pen holding pressure dynamic characteristic and time as well as the writing length is expressed in a three-dimensional space coordinate system, the x axis represents the writing length, the y axis represents time, the z axis represents the dynamic characteristic, and from the origin, the relationship curves of the dynamic characteristic, the time and the writing length are respectively recorded in each digital axis plane. Sampling is carried out for 6 times at equal time intervals from the beginning of pen drop of a signer to the end of signature, corresponding points on curves in three planes of xoz, xoy and yoz at a certain moment are connected at different time points to form 6 feature triangles capable of reflecting the relation between the dynamic features, the time and the length of a written handwriting, three vertexes of the feature triangles are connected with an origin of coordinates to form 6 feature tetrahedrons, and the feature tetrahedrons are used as feature models to carry out feature extraction for subsequent processing.
(5) A feature extraction program module: and recording three side lengths, areas and volumes of the characteristic triangles corresponding to the writing horizontal pressure dynamic characteristics and the pen holding pressure dynamic characteristics when the signer holds the pen, constructing characteristic vectors corresponding to the characteristics by using the 5 types of information, combining the characteristic vectors corresponding to a plurality of characteristic triangles, and constructing a matrix for similarity measurement of a subsequent module and authenticity identification of the signature of the signer to be detected.
Specifically, the writing horizontal pressure dynamic characteristic and the pen holding pressure dynamic characteristic respectively mark the area of a characteristic triangle as S according to the sequence of time sampling pointsh,1≤h≤p1. The vertex lying in the xOz plane is denoted as ahAnd the vertex in the xOy plane is denoted as bhAnd the vertex lying in the yOz plane is denoted as chThe volume of the corresponding characteristic tetrahedron is denoted VhThe elements in the constructed corresponding feature vector are expressed as (a)hbh,bhch,chah,Sh,Vh) Will not be arranged inConstructing a matrix capable of reflecting dynamic characteristics by using the same characteristic vector as a row vector group of the matrix, and marking the matrix as Bh,
(6) Similarity measurement program module: for different signatories, similarity measurement is carried out on the information to be measured and the sample information by utilizing the 6 x 5 matrixes corresponding to the writing horizontal pressure dynamic characteristic and the pen holding pressure dynamic characteristic respectively, and weighting coefficients are given to the two similarity measurement results respectively.
In order to obtain the weight coefficients of the writing level pressure dynamic characteristic and the pen holding pressure dynamic characteristic in the similarity measurement, samples are extracted from a sample database to calculate the dynamic characteristic weight coefficients, the average information entropy corresponding to each type of characteristic is calculated, the corresponding weight coefficient is further calculated, and the similarity of the two types of dynamic characteristics is weighted and summed to obtain the comprehensive similarity.
In order to obtain the weight coefficient of the similarity of each type of feature, samples are extracted from a sample database to calculate the weight coefficient, the similarity measurement is carried out on the samples in the special sample set by respectively utilizing the writing level pressure dynamic feature and the pen holding pressure dynamic feature, the number of the samples with the similarity reaching a specific threshold value is recorded as C, and C is the total number m of the special sample set0In the method, the similarity between every two sample information reaches the number of the set threshold value.
a) Calculating the average information entropy corresponding to each type of feature:
dividing the sample library into K classes according to the number K of the signers entering the samples, recording each class of dynamic characteristic sample set as Q, wherein each class of dynamic characteristic sample set comprises z groups of information entered by the signerskK is 1, 2, …, K, m is selected from the ith subclassiGroup information, mi=1,2,…,z;
Constructing a special sample set M, M ═ qij|qij∈Qi,i=1,2,…,K,j=1,2…,miThe total number of sample sets is
Sample set Q of each type of dynamic featurekHas a number of samples ofWriting horizontal pressure dynamics is noted as F1Dynamic pen holding pressure characteristic is recorded as F2;
wherein the content of the first and second substances,dynamic characteristics FlCorresponding mean entropy of information
The smaller the average information entropy, the higher the validity of the feature for the signer identity authentication.
the weight coefficient satisfiesThe smaller the average information entropy of the features is, the higher the weight coefficient is, and the weight coefficient is calculated and used for measuring the similarity of the dynamic features;the weight coefficients corresponding to the writing level pressure dynamics,the weight coefficient is corresponding to the dynamic characteristics of the pen holding pressure;
c) calculating the similarity between the information to be detected and the sample information:
recording a matrix corresponding to the dynamic characteristics of the writing horizontal pressure in the information to be detected as Du(u is more than or equal to 1 and less than or equal to 4), and the matrix corresponding to the dynamic characteristics of the pen holding pressure is recorded as Ev(v is more than or equal to 1 and less than or equal to 2), and the horizontal pressure characteristic matrix of the sample information is recorded as Du' (u is more than or equal to 1 and less than or equal to 4) and the matrix corresponding to the dynamic characteristics of the pen holding pressure is marked as Ev' (v is more than or equal to 1 and less than or equal to 2), and marking the element of the mth row and the nth column in the information matrix to be detected as amn(1≤m≤p1N is more than or equal to 1 and less than or equal to 5), and the element of the mth row and the nth column in the corresponding sample information matrix is recorded as bmn(1≤m≤p1N is more than or equal to 1 and less than or equal to 5), calculating the mean value of the sum of squares of the difference values of the corresponding elements among the matrixes by using the matrixes, normalizing the mean value, defining the result after normalization as the similarity of the information to be detected and the sample information corresponding to the characteristic, defining the similarity of the information to be detected and the sample information as P,
writing waterThe flat pressure dynamic feature similarity is the similarity mean value of the information of the 4 acquisition surfaces, and the pen-holding pressure dynamic feature similarity is the similarity mean value of the information of the two acquisition surfaces corresponding to the thumb and the forefinger. p is a radical of1Can take 6.
d) Calculating the comprehensive similarity between the information to be detected and the sample information:
obtaining the comprehensive similarity SIM of the information to be measured and the sample information by utilizing the weight coefficients of the two types of dynamic characteristics obtained by the special sample set M,
e) calculating the similarity between a plurality of sample information:
for each sample signer, according to the similarity measurement method of the information to be measured and the sample information, similarity measurement is carried out on a plurality of corresponding sample information in the sample database pairwise, if each signer records s1Group information, a common s2Result of similarity
The minimum value of the saved result is recorded as T, representing the similarity of the two sample information with respect to the writing level pressure characteristic,representing the similarity of the two sample information in terms of the pen-holding pressure characteristics.
And selecting a threshold for identifying similarity with the output program module.
(7) An authentication and output program module: and taking the comprehensive similarity threshold as bT and b as coefficients, and comparing the comprehensive similarity SIM of the information to be detected and the sample information with the comprehensive similarity threshold bT to identify the authenticity of the signature of the signer to be detected, wherein as shown in the identification result of the table 1, if the SIM is more than or equal to the bT, the signer to be detected is considered to be the same as the sample signer, if the SIM < the bT, the signer to be detected is considered to be different from the sample signer, a judgment result is output, and corresponding identification information is visually fed back to an operator through a software operation interface so as to identify the identity of the signer to be detected. Here b may take 0.8.
During actual identification, namely when the signature information of a signer to be detected is input, the signature information is only acquired once, and is acquired for multiple times during sample acquisition, similarity measurement is respectively carried out on the information to be detected and multiple groups of sample information, and a result with the largest similarity is selected to be compared with a threshold value, so that authenticity identification is realized.
TABLE 1 statistical table of discrimination results
The present invention has been disclosed in terms of the preferred embodiment, but it is not intended to be limited to the embodiment, and all technical solutions obtained by using equivalent substitution or equivalent transformation fall within the scope of the present invention.
Claims (7)
1. The signature pen-oriented signature authenticity identification system is characterized by comprising an information acquisition unit and the following program modules:
1) an information entry program module: inputting signature information and personal identity information of a signer to be tested, wherein the signature information comprises writing horizontal pressure dynamic characteristics, pen holding pressure dynamic characteristics, writing time and writing length of the signer when the signer holds a pen for writing;
2) a sample database program module: the method comprises the steps that dynamic characteristic information and personal identity information, collected by a sign pen, of a sample signer during pen holding writing are stored in advance, wherein the dynamic characteristic information comprises writing horizontal pressure dynamic characteristics and pen holding pressure dynamic characteristics;
3) a characteristic model building program module: acquiring and constructing a writing horizontal pressure dynamic characteristic model and a pen holding pressure dynamic characteristic model when a signer holds a pen:
4) a feature extraction program module: in the case of a pressure dynamic characteristic p at the writing level1Or dynamic character p of pen holding pressure2Time t1Length of writing1In the formed space coordinate system xyz, sampling for multiple times at equal time intervals in the period from the start of pen drop of a signer to the end of signature, connecting corresponding points on curves in three planes of xoz, xoy and yoz at a certain moment at different time points to form a plurality of feature triangles reflecting the relation between the dynamic feature of writing horizontal pressure and the length of time and handwriting, and connecting three vertexes of the feature triangles with the origin of coordinates to form a plurality of feature tetrahedrons;
recording three side lengths, areas and volumes of characteristic tetrahedrons of corresponding characteristic triangles for the writing horizontal pressure dynamic characteristics and the pen holding pressure dynamic characteristics when a signer holds a pen, constructing a characteristic vector corresponding to the characteristics by using the 5 types of information, combining the characteristic vectors corresponding to a plurality of characteristic triangles, and constructing a matrix for similarity measurement and signature authenticity identification of the signer to be detected;
5) similarity measurement program module: for different signatories, respectively utilizing the writing level pressure dynamic characteristic and the pen holding pressure dynamic characteristic corresponding to p1X 5 matrix, p1For the number of the selected feature triangles, 5 is 5 pieces of information of the volume of a feature tetrahedron formed by connecting three side lengths, the area and three vertexes of the feature triangles with the original point, and similarity measurement is carried out on the information to be measured and the sample information to obtain the comprehensive similarity between the information to be measured and the sample information;
6) an authentication and output program module: setting a comprehensive similarity threshold value according to the minimum value of similarity measurement among a plurality of sample information of the same sample signer, comparing the magnitude relation between the comprehensive similarity and the comprehensive similarity threshold value, identifying the authenticity of the signer to be detected, if the comprehensive similarity is greater than or equal to the comprehensive similarity threshold value, considering that the signer to be detected is the same as the sample signer, and if the comprehensive similarity is less than the comprehensive similarity threshold value, considering that the signer to be detected is different from the sample signer.
2. The system for authenticating the authenticity of the signature pen as claimed in claim 1, wherein: in the feature model building program module, the writing horizontal pressure dynamic feature building process comprises the following steps:
the threaded cylinder is located at the joint of the lower end of the sign pen column and the pen core, a plurality of pressure sensing devices I are arranged on the inner surface of the threaded cylinder along the circumferential direction, the pressure sensing devices I collect pressure information applied by the pen core in a plurality of directions on the inner surface of the threaded cylinder during writing, writing horizontal pressure information with the pressure not being zero in the direction is calculated from the beginning of a pen-falling signature to the end of the signature, and writing horizontal pressure information, writing length and time change curves are recorded respectively and expressed in a three-dimensional space coordinate system for subsequent feature extraction and similarity measurement.
3. The system for authenticating the authenticity of the signature pen as claimed in claim 1, wherein: in the characteristic model building program module, the process of building the pen holding pressure dynamic characteristic is as follows: collecting finger pressure information of pressure collection surfaces corresponding to the thumb and the forefinger, recording the change curves of the finger pressure information, writing length and time collected by the thumb pressure collection surface and the forefinger pressure collection surface from the beginning of pen-down signature to the end of signature, representing in a three-dimensional space coordinate system, and using for subsequent feature extraction and similarity measurement.
4. The system for authenticating the authenticity of the signature as claimed in claim 3, wherein: in the module of the feature extraction program,
for the writing horizontal pressure dynamic characteristic and the pen holding pressure dynamic characteristic, the area of the characteristic triangle is recorded as S according to the sequence of time sampling pointsh,1≤h≤p1Position of replacementThe vertex in the xOz plane is denoted as ahAnd the vertex in the xOy plane is denoted as bhAnd the vertex lying in the yOz plane is denoted as chThe volume of the corresponding characteristic tetrahedron is denoted VhThe elements in the constructed corresponding feature vector are expressed as (a)hbh,bhch,chah,Sh,Vh) The different characteristic vectors are used as row vector groups of the matrix to construct a matrix capable of reflecting the dynamic characteristics, and the matrix is marked as Bh,
5. The system for authenticating the authenticity of the signature pen as claimed in claim 1, wherein: in the similarity measurement program module, in order to obtain the weight coefficients of the writing level pressure dynamic characteristic and the pen holding pressure dynamic characteristic in the similarity measurement respectively, samples are extracted from a sample database to calculate the dynamic characteristic weight coefficients, the average information entropy corresponding to each type of characteristics is calculated, the corresponding weight coefficients are further calculated, and the similarity of the two types of dynamic characteristics is weighted and summed to obtain the comprehensive similarity.
6. The system for authenticating the authenticity of the signature as claimed in claim 5, wherein: the method specifically comprises the following steps:
a) calculating the average information entropy corresponding to each type of feature:
dividing the sample library into K classes according to the number K of the signers recording the samples, wherein each class comprises s recorded by the signer1Group information, recording each type of dynamic characteristic sample set as QkK is 1, 2, …, K, m is selected from the ith subclassiGroup information, mi=1,2,…,s1;
Constructing a special sample set M, M ═ qij|qij∈Qi,i=1,2,…,K,j=1,2…,miThe total number of sample sets is
Sample set Q of each type of dynamic featurekHas a number of samples ofWriting horizontal pressure dynamics is noted as F1Dynamic pen holding pressure characteristic is recorded as F2;
wherein the content of the first and second substances,dynamic characteristics FlCorresponding mean entropy of informationC is the total number m of the special sample sets0In the method, the number of similarity between every two pieces of sample information reaches a set threshold value;
the weight coefficient satisfies The weight coefficients corresponding to the writing level pressure dynamics,the weight coefficient is corresponding to the dynamic characteristics of the pen holding pressure;
c) calculating the similarity between the information to be detected and the sample information:
recording a matrix corresponding to the dynamic characteristics of the writing horizontal pressure in the information to be detected as DuU is more than or equal to 1 and less than or equal to 4, and the matrix corresponding to the dynamic characteristics of the pen holding pressure is recorded as EvV is more than or equal to 1 and less than or equal to 2, and the sample information horizontal pressure characteristic matrix is recorded as D'uU is more than or equal to 1 and less than or equal to 4, and a matrix corresponding to the dynamic pen-holding pressure characteristic is recorded as E'vV is more than or equal to 1 and less than or equal to 2, and the element of the mth row and the nth column in the information matrix to be detected is marked as amn,1≤m≤p1N is more than or equal to 1 and less than or equal to 5, and the element of the mth row and the nth column in the corresponding sample information matrix is recorded as bmn,1≤m≤p1N is more than or equal to 1 and less than or equal to 5, calculating the mean value of the sum of squares of the difference values of the corresponding elements between the matrixes, normalizing the mean value, defining the result after normalization as the similarity of the information to be measured and the sample information corresponding to the characteristic, defining the similarity of the information to be measured and the sample information as P,
d) calculating the comprehensive similarity between the information to be detected and the sample information:
obtaining the comprehensive similarity SIM of the information to be measured and the sample information by utilizing the weight coefficients of the two types of dynamic characteristics obtained by the special sample set M,
representing the similarity of the information to be measured and the sample information with respect to the writing level pressure characteristic,representing the similarity of the information to be detected and the sample information in terms of the pen holding pressure characteristic;
e) calculating the similarity between a plurality of sample information:
for each sample signer, according to the similarity measurement method of the information to be measured and the sample information, similarity measurement is carried out on a plurality of corresponding sample information in the sample database pairwise, if each signer records s1Group information, a common s2Result of similarity
7. The system for authenticating the authenticity of the signature as claimed in claim 6, wherein: in the identification and output program module, the comprehensive similarity threshold value bT and the comprehensive similarity threshold value bT are taken as coefficients, the authenticity of the signature of the signer to be detected is identified by comparing the comprehensive similarity SIM of the information to be detected and the sample information with the comprehensive similarity threshold value bT, if the SIM is more than or equal to the bT, the signer to be detected is considered to be the same as the signer of the sample, if the SIM is less than the bT, the signer to be detected is considered to be different from the signer of the sample, and a judgment result is output.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710933556.XA CN107657241B (en) | 2017-10-09 | 2017-10-09 | Signature pen-oriented signature authenticity identification system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710933556.XA CN107657241B (en) | 2017-10-09 | 2017-10-09 | Signature pen-oriented signature authenticity identification system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107657241A CN107657241A (en) | 2018-02-02 |
CN107657241B true CN107657241B (en) | 2021-03-26 |
Family
ID=61116464
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710933556.XA Active CN107657241B (en) | 2017-10-09 | 2017-10-09 | Signature pen-oriented signature authenticity identification system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107657241B (en) |
Families Citing this family (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108921126B (en) * | 2018-07-20 | 2019-07-23 | 北京开普云信息科技有限公司 | A kind of automatic identification signature stamp or the method and device of handwritten signature |
CN110245571A (en) * | 2019-05-20 | 2019-09-17 | 深圳壹账通智能科技有限公司 | Contract signature checking method, device, computer equipment and storage medium |
CN110222660B (en) * | 2019-06-12 | 2022-09-23 | 河海大学常州校区 | Signature authentication method and system based on dynamic and static feature fusion |
CN110942042B (en) * | 2019-12-02 | 2022-11-08 | 深圳棒棒帮科技有限公司 | Three-dimensional handwritten signature authentication method, system, storage medium and equipment |
CN111555889A (en) * | 2020-04-27 | 2020-08-18 | 深圳壹账通智能科技有限公司 | Electronic signature verification method and device, computer equipment and storage medium |
CN113065414A (en) * | 2021-03-15 | 2021-07-02 | 国网河北省电力有限公司 | Handwriting identification method based on writing kinematics theory |
CN112929173A (en) * | 2021-03-17 | 2021-06-08 | 讯翱(上海)科技有限公司 | Digital certificate authentication method based on signature recognition |
CN113158819B (en) * | 2021-03-29 | 2022-04-22 | 郑静 | Signature verification method based on handheld terminal |
CN114463858B (en) * | 2022-01-12 | 2024-05-24 | 广州市双照电子科技有限公司 | Signature behavior recognition method and system based on deep learning |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1227373A (en) * | 1998-02-23 | 1999-09-01 | 三菱电机株式会社 | Handwriting verification device |
KR20050048977A (en) * | 2003-11-20 | 2005-05-25 | 주식회사 투비링크 | Method for proving the identity using the biometrics signature verification system |
JP2007317022A (en) * | 2006-05-26 | 2007-12-06 | Canon Inc | Handwritten character processor and method for processing handwritten character |
CN101246543A (en) * | 2008-03-18 | 2008-08-20 | 苏州纳米技术与纳米仿生研究所 | Examiner identity appraising system based on bionic and biological characteristic recognition |
CN103218624A (en) * | 2013-04-25 | 2013-07-24 | 华东理工大学 | Recognition method and recognition device based on biological characteristics |
-
2017
- 2017-10-09 CN CN201710933556.XA patent/CN107657241B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1227373A (en) * | 1998-02-23 | 1999-09-01 | 三菱电机株式会社 | Handwriting verification device |
KR20050048977A (en) * | 2003-11-20 | 2005-05-25 | 주식회사 투비링크 | Method for proving the identity using the biometrics signature verification system |
JP2007317022A (en) * | 2006-05-26 | 2007-12-06 | Canon Inc | Handwritten character processor and method for processing handwritten character |
CN101246543A (en) * | 2008-03-18 | 2008-08-20 | 苏州纳米技术与纳米仿生研究所 | Examiner identity appraising system based on bionic and biological characteristic recognition |
CN103218624A (en) * | 2013-04-25 | 2013-07-24 | 华东理工大学 | Recognition method and recognition device based on biological characteristics |
Non-Patent Citations (3)
Title |
---|
A Novel Digitizing Pen for the Analysis of Pen Pressure and Inclination in Handwriting Biometrics;Hook C , Jürgen Kempf, Scharfenberg G;《 Biometric Authentication, Eccv International Workshop》;20040515;第243-294页 * |
基于图像特征融合识别的中文签名鉴伪方法;马云鹏,李庆武,刘艳,曹美;《应用科技》;20151215;第10-14页 * |
签名笔迹笔压特征显现方法的比较研究;刘宁,裴雷,吴亦;《中国司法鉴定》;20141215;第25-29页 * |
Also Published As
Publication number | Publication date |
---|---|
CN107657241A (en) | 2018-02-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107657241B (en) | Signature pen-oriented signature authenticity identification system | |
CN106778586B (en) | Off-line handwritten signature identification method and system | |
CN105956582B (en) | A kind of face identification system based on three-dimensional data | |
Feng et al. | Robust and efficient algorithms for separating latent overlapped fingerprints | |
CN110543822A (en) | finger vein identification method based on convolutional neural network and supervised discrete hash algorithm | |
CN104200239A (en) | Image feature fusion identification based signature authentic identification system and method | |
CN104809451B (en) | A kind of person's handwriting identification system based on stroke curvature measuring | |
CN107392136B (en) | Signature authenticity identification system based on feature self-adaptive oscillation attenuation | |
CN110874587B (en) | Face characteristic parameter extraction system | |
Samuel et al. | Novel feature extraction technique for off-line signature verification system | |
CN109145704B (en) | Face portrait recognition method based on face attributes | |
CN110222660B (en) | Signature authentication method and system based on dynamic and static feature fusion | |
Engin et al. | Offline signature verification on real-world documents | |
CN110119693B (en) | English handwriting identification method based on improved VGG-16 model | |
CN107103289B (en) | Method and system for handwriting identification by using handwriting outline characteristics | |
CN110866442B (en) | Real-time face recognition-based testimony-of-person integrated checking system and method | |
Paudel et al. | Handwritten signature verification for mobile phones | |
Houtinezhad et al. | Off-line signature verification system using features linear mapping in the candidate points | |
Aravinda et al. | An approach for signature recognition using contours based technique | |
Singh et al. | Survey on offline signature recognition and verification schemes | |
Abdulbaqi et al. | Biometrics detection and recognition based-on geometrical features extraction | |
Zaghloul et al. | Recognition of Hindi (Arabic) handwritten numerals | |
Yan et al. | A novel bimodal identification approach based on hand-print | |
Douma et al. | Applying the Histogram of Oriented Gradients to Recognize Arabic Letters | |
Palanikumar et al. | Advanced palmprint recognition using unsharp masking and histogram equalization |
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 |