WO2020082577A1 - Seal anti-counterfeiting verification method, device, and computer readable storage medium - Google Patents

Abstract

Provided are a seal anti-counterfeiting verification method, a device, and a computer readable storage medium. Said method comprises: establishing a reference database for feature points of a reference seal image; extracting, from a document, a seal image to be verified; extracting feature points of said seal image to be verified; comparing the feature points of said seal image to be verified, with the feature points of the reference seal image in the reference database, to obtain matching feature points; and calculating the number of the matching feature points, and verifying, according to the number of the matching feature points, the authenticity of said seal image to be verified. This method accurately and quickly detects the authenticity of a seal.

Description

Seal anti-counterfeiting inspection method, device and computer readable storage medium

This application requires the priority of the Chinese patent application submitted to the China Patent Office on October 26, 2018 with the application number 201811256366.X and the invention titled "Seal Anti-counterfeiting Inspection Method, Device, and Computer-readable Storage Medium", all of its contents Incorporated by reference in this application.

Technical field

The present application relates to the field of image recognition technology, and in particular, to a seal anti-counterfeiting inspection method and device based on key feature point detection and comparison, and a computer-readable storage medium.

Background technique

In China, the seal is an important certificate that proves the identity of the party and government organs, enterprises, social organizations and other unit organizations, represents their rights and interests, and has legal effect. It is an important means for the state to exercise power, manage society, and citizens and legal persons to exercise civil rights. It plays an extremely important role in social, political and economic life. Due to the current backward technical means in seal management, seal anti-counterfeiting, and identification of counterfeit seals, driven by huge economic interests, criminals have furiously forged administrative agencies, law enforcement departments, financial institutions, enterprises, institutions and legal persons at all levels. The seal has caused serious economic losses to the state, collectives and individuals. During the two sessions each year, deputies to the National People's Congress called for the improvement of the national unified seal management mechanism as soon as possible, speeding up the production of seal anti-counterfeiting technology, and the study of counterfeit seal identification technology, and three-pronged measures to effectively combat the criminal activities of counterfeit seals.

Summary of the invention

The present application provides a seal anti-counterfeiting inspection method, device and computer-readable storage medium, and its main purpose is to accurately and quickly detect the authenticity of the seal.

To achieve the above purpose, a seal anti-counterfeiting inspection method provided by this application includes:

Establish a reference database of feature points that refer to the seal image;

Extract the seal image to be checked from the document;

Extract the feature points of the seal image to be checked;

Matching the feature points of the seal image to be inspected with the reference seal image feature points in the reference database to obtain matching feature points; and

The number of matching feature points in the feature image of the seal image to be tested and the reference seal image is counted, and the authenticity of the seal image to be tested is verified according to the number of matching feature points.

In addition, in order to achieve the above object, the present application also provides a seal anti-counterfeiting inspection device, which includes a memory and a processor, and the memory stores a seal anti-counterfeiting inspection program that can run on the processor, and the seal anti-counterfeiting When the verification program is executed by the processor, the following steps are implemented:

Establish a reference database of feature points that refer to the seal image;

Extract the seal image to be checked from the document;

Extract the feature points of the seal image to be checked;

Matching the feature points of the seal image to be checked with the reference seal image feature points in the reference database to obtain matching feature points; and

The number of matching feature points in the feature image of the seal image to be tested and the reference seal image is counted, and the authenticity of the seal image to be tested is verified according to the number of matching feature points.

In addition, in order to achieve the above object, the present application also provides a computer-readable storage medium on which a seal anti-counterfeiting verification program is stored, which can be executed by one or more processors, In order to realize the steps of the seal anti-counterfeiting inspection method as described above.

The method, device and computer-readable storage medium for seal anti-counterfeiting inspection proposed in this application establish a reference database of feature points of reference seal images; extract seal images to be checked from documents; extract feature points of seal images to be checked; and seals to be checked The feature points of the image are matched with the feature points of the reference seal image in the reference database to obtain the matched feature points; and the number of feature points matching the feature points of the seal image to be checked and the feature points of the reference seal image is calculated according to The number of matching feature points verifies the authenticity of the seal image to be verified. This application can accurately and quickly detect the authenticity of the seal.

BRIEF DESCRIPTION

FIG. 1 is a schematic flowchart of a seal anti-counterfeiting inspection method provided by an embodiment of the present application;

2 is a schematic diagram of an internal structure of a seal anti-counterfeiting inspection device provided by an embodiment of the present application;

FIG. 3 is a schematic diagram of modules of a seal anti-counterfeiting inspection program in a seal anti-counterfeiting inspection device provided by an embodiment of the present application.

The implementation, functional characteristics and advantages of the present application will be further described in conjunction with the embodiments and with reference to the drawings.

detailed description

It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

This application provides a seal anti-counterfeiting inspection method. Referring to FIG. 1, it is a schematic flowchart of a seal anti-counterfeiting inspection method provided by an embodiment of the present application. The method may be executed by an apparatus, and the apparatus may be implemented by software and / or hardware.

In this embodiment, the seal anti-counterfeiting inspection method includes:

S10. Establish a reference database for the feature points of the reference stamp image.

The preferred embodiment of the present application acquires the seal image of the real seal as an image of the reference seal through an image acquisition device such as a camera, a scanner, a CCD camera, and gives each reference seal image a fixed ID, which corresponds to a specific real seal .

A feature point extraction method is used to extract feature points in the reference seal image, and data such as position information and descriptor information of the feature points contained in the reference seal image are stored in the reference database.

The feature point extraction method used in this proposal is SIFT (Scale Invariant Feature Transform) algorithm.

The SIFT operator is an operator with better performance in the image matching algorithm, and the feature image registration based on the SIFT algorithm can be roughly divided into feature detection, description, and matching. Feature detection is performed in scale space. First, the image scale space is generated, and then the local extremum values in the scale space are detected, and then the local extremum points are accurately located by excluding low-contrast points and edge response points; When calculating the principal direction of each extremum point, the main direction of the extremum point and the histogram gradient direction of the area with the extreme point as the center are calculated to generate the feature point descriptor; the feature point descriptor can be used to find the match Features to establish connections between images.

SIFT feature detection mainly includes the following four basic steps: 1. Scale space extreme value detection: search for image positions on all scales. Gaussian differential functions are used to identify potential candidate points that are invariant to scale and rotation. 2. Feature point positioning: at each candidate position, a precise model is used to determine the position and scale. The selection of feature points depends on their stability. 3. Direction determination: Based on the local gradient direction of the image, one or more directions are assigned to each feature point location. All subsequent operations on image data are transformed with respect to the direction, scale, and position of the feature points, thereby providing invariance to these transformations. 4. Feature point description: In the neighborhood around each feature point, measure the local gradient of the image at the selected scale. These gradients are transformed into a representation that allows larger local shape deformations and light changes. In the description of feature points, the neighborhood of 16 × 16 is taken as the sampling window with the feature point as the center, and the relative direction of the sample point and the feature point is weighted by Gaussian into a direction histogram containing 8 bins, and finally 4 × 4 × 8 128-dimensional feature point descriptor.

S20. Extract the seal image to be checked from the document.

In the preferred embodiment of the present application, the steps of extracting the seal image to be inspected include:

Substep 1. Separate the feature data of the preset color from the document.

It should be understood that most of the seals are red, so this embodiment uses red as an example.

In order to highlight the difference in value between red and other colors, this scheme proposes three ways to describe colors according to the characteristics of colors expressed in different color spaces. They are:

(1) R-G, R-B difference in RGB space.

The three components R, G, and B in the RGB space represent the mixed amounts of red, green, and blue, respectively. Intuitively speaking, the red R value, that is, the red component must be relatively large, in fact, the RGB value of the standard red in RGB space is (255,0,0), but a large R value does not mean that the color is close to red , The size of G value and B value will have a significant effect on the color, for example (255,255,0) represents yellow, and (255,0,255) represents purple. It can be seen from this that when the R value is large and the G and B values are small, the corresponding color is closer to red.

Correspondingly, this solution calculates the difference RG of the R component and the G component of a pixel, and the difference RB of the R component and the B component, when both the difference RG and the difference RB are greater than the corresponding preset threshold To determine that the pixel is red.

(2) H value of HSV space.

H, S, and V in HSV space represent hue, saturation, and intensity, respectively. The description of colors in the HSV space is closer to humans ’understanding of colors: H is the hue, and the phase of the color is expressed in the form of angle values, from 0 ° to 360 °, which indicates the transition from red to green to blue and back to red, S Represents the fill ratio of the color under the corresponding hue, taking 0-100%, V represents the brightness, or light intensity, taking 0-100%. For the seal, the change in red is mainly affected by the light collected by the image (corresponding to the brightness V) and the wear of the paper itself (corresponding to the saturation S), so the value of H can better describe the "color" difference in the general sense .

Therefore, this solution calculates the H value of a pixel, and when the H value of the pixel is less than the corresponding preset threshold, it is determined that the pixel is red.

(3) YCbCr space deflection angle θ.

YCbCr space is not an absolute color space, but an encoding of RGB space. The encoding rules are as follows:

In fact, each component of the YCbCr space is still meaningful, Y represents the grayscale of the pixel, and Cb and Cr represent the blue and red density offset components of the pixel, respectively. In this solution, an angle θ between a vector mapped on the CbCr plane and the Cr axis can be used to describe the color of the pixel. The effect is similar to H in HSV, but the distribution of H is not the same. among them:

θ = atan2 (Cb, Cr);

When the calculated θ is less than the corresponding preset threshold, it is determined that the pixel is red.

Sub-step 2. According to the characteristic data of the preset color, the seal image is separated by a double-threshold color separation method.

In order to solve the problem that the noise is close to the color of the stamp itself, this scheme uses the k-means clustering method. Assuming that the θ value in the YCbCr space is used as the color description method, the image is divided into clusters according to the θ value of each pixel. To obtain the upper and lower bounds of the θ value of the cluster to which the seal color belongs [a, b]. The color of the seal edge tends to be closer to the background color, which means that the color of this part may be very close to the color of part of the noise and it is considered to be removed by noise, so this scheme adopts the double threshold method, based on k-means aggregation On the basis of the threshold [a, b] obtained by the class, the value range is first narrowed, and a strict threshold is used to obtain the pixels that absolutely belong to the seal, and the set of pixel points of the seal is formed, and then the value range is expanded to adopt a more relaxed Threshold range to determine whether the adjacent points of each point in the seal pixel set belong to the seal, so as to supplement the seal pixel set and ensure the integrity of the seal image.

Sub-step 3. Perform contour detection of the seal image, and cut out the seal image according to the detected contour.

In a preferred embodiment of the present application, the outline detection of the seal includes any one of the following methods or a combination of several methods:

(1) Edge detection. This solution uses the Canny edge detection algorithm based on Sobel operator. This method can detect the edge of the seal well, and the gradient value of the edge point can be obtained through Sobel operator.

(2) Random Hough transform detects ellipse. Random Hough transform can effectively detect ellipse, find the position of the ellipse and draw a more accurate ellipse outline, but its mechanism of randomly taking points causes it to be easily disturbed by noise , Easy to appear after reaching the maximum number of cycles still can not find any ellipse results. In addition, its ellipse parameter calculation method also makes it impossible to calculate a very accurate ellipse equation.

(3) Direct least square fitting of ellipse, direct least square fitting of ellipse is an optimized method of ellipse detection. It uses a set of sampling point data to solve unknown parameters in elliptic equations by algebraic method. The key problem to solve with the least square method is the selection of sample points, so other methods need to be combined to obtain better samples.

(4) Combine the random Hough transform with the direct least squares fitting of the ellipse to detect the ellipse. Since the steps of using these two methods are more complicated, they will not be elaborated here. In short, using this method to detect the ellipse of the stamp image, the result often contains two ellipses with different sizes of the same centroid, corresponding to the inner and outer edges of the stamp border, and the larger one is obtained. Precise edge. The color of pixels outside the edge is set to white, and the image is cut according to the position of the edge to obtain an elliptical image within the outline.

S30. Extract the feature points of the seal image to be checked.

In this step, SIFT (Scale Invariant Feature Transform) algorithm can also be used as a method for extracting feature points.

S40. Match the feature points of the seal image to be tested with the reference seal image feature point reference database to obtain matching feature points.

Feature points have two types of descriptors such as feature vector and feature matrix.

Using feature vectors of feature points as feature point feature descriptors, this scheme uses the Euclidean distance evaluation method for feature point matching:

Obtain the feature vector V _j ^{recg of the} feature point i of the reference seal image = [x ₁ , x ₂ … x _n ], and the feature vector V _j ^{recg of the} feature point j of the seal image to be tested = [x ₁ ′, x ₂ ′… x _n '], Where n represents the dimension of the feature vector, and the Euclidean distance of the two feature vectors is:

The two corresponding feature vectors with the smallest Euclidean distance are the matching feature vectors, and the corresponding reference stamp image feature points are the matching feature points.

The feature matrix of feature points is used as the feature descriptor of feature points. Since such feature matrices are generally positive definite matrices, this solution uses the following methods to perform feature point matching:

待检验印章图像特征点j的特征矩阵

其中n表示特征矩阵的行、列数，两个特征矩阵的ρ距离为： Obtain the feature matrix of the reference point image feature point i

Feature matrix of feature point j of seal image to be tested

Where n represents the number of rows and columns of the feature matrix, and the ρ distance between the two feature matrices is:

λ _k (V _i ^ref , V _j ^recg ) λ _k V _i ^ref x _k -V _j ^recg x _k = 0 _{k = 1 ... n} , x _k ≠ 0,

Where V _i ^ref and V _j ^recg are two covariance matrices, and ρ (V _i ^ref , V _j ^recg ) represents their evaluation distance. λ _k (V _i ^ref , V _j ^recg ) represents the generalized eigenvalues of V _i ^ref and V _j ^recg and is calculated by the following formula: λ _k V _i ^ref x _k -V _j ^recg x _k = 0 _{k = 1 .. .n} where x _k ≠ 0 is the generalized eigenvector of V _i ^ref and V _j ^recg , and n is the dimension of the generalized eigenvector.

The two corresponding feature vectors with the smallest distance ρ (V _i ^ref , V _j ^recg ) are matching feature vectors, and the corresponding feature points of the reference seal image and the feature image of the seal image to be tested are matching feature points.

After the matching of the feature points of the seal image to be inspected and that of the reference seal image is completed, the feature points of the reference seal image [P ₁ P ₂ … P _n ] and the feature points of the seal image to be tested [P ₁ ′ P ₂ ′… P _n ′], where n represents the number of matching feature points. The above formula indicates that the reference image feature point P ₁ corresponds to the whole point P ₁ ′ of the image to be inspected, P ₂ corresponds to P ₂ ′, and so on.

S50. Count the number of feature points matching the feature points of the seal image to be tested and the reference seal image feature points, and verify the authenticity of the seal image to be tested according to the number of matching feature points.

Where, if the number of matching feature points of the seal image to be checked and the reference seal image feature points in the total number of feature points of the reference seal image exceeds the preset value, such as 85%, the pending Verify that the stamp image is true.

The present application also provides a seal anti-counterfeiting inspection device. Referring to FIG. 2, it is a schematic diagram of an internal structure of a seal anti-counterfeiting inspection device provided by an embodiment of the present application.

In this embodiment, the seal anti-counterfeiting inspection device 1 may be a PC (Personal Computer), a terminal device such as a smart phone, tablet computer, or portable computer, or a server or server cluster. The seal anti-counterfeiting verification device 1 includes at least a memory 11, a processor 12, a communication bus 13, and a network interface 14.

Wherein, the memory 11 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), magnetic memory, magnetic disk, optical disk, and the like. In some embodiments, the memory 11 may be an internal storage unit of the seal anti-counterfeiting verification device 1, such as a hard disk of the seal anti-counterfeiting verification device 1. In other embodiments, the memory 11 may also be an external storage device of the seal anti-counterfeiting inspection device 1, for example, a plug-in hard disk equipped on the seal anti-counterfeiting inspection device 1, a smart memory card (Smart Media Card, SMC), and a secure digital (Secure Digital, SD) card, flash card (Flash Card), etc. Further, the memory 11 may also include both the internal storage unit of the seal anti-counterfeiting verification device 1 and the external storage device. The memory 11 can be used not only to store application software and various types of data installed in the seal anti-counterfeiting inspection device 1, such as codes of the seal anti-counterfeiting inspection program 01, but also to temporarily store data that has been output or is about to be output.

In some embodiments, the processor 12 may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip for running the program code or processing stored in the memory 11 Data, for example, the execution of the seal anti-counterfeiting inspection program 01.

The communication bus 13 is used to realize connection and communication between these components.

The network interface 14 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface), and is generally used to establish a communication connection between the device 1 and other electronic devices.

Optionally, the device 1 may further include a user interface. The user interface may include a display and an input unit such as a keyboard. The optional user interface may also include a standard wired interface and a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode, organic light emitting diode) touch device, or the like. Among them, the display may also be appropriately referred to as a display screen or a display unit for displaying information processed in the seal anti-counterfeiting inspection device 1 and for displaying a visual user interface.

FIG. 2 only shows the seal anti-counterfeiting inspection device 1 having the components 11-14 and the seal anti-counterfeiting inspection program 01. Those skilled in the art can understand that the structure shown in FIG. , May include fewer or more components than shown, or combine certain components, or different component arrangements.

In the embodiment of the device 1 shown in FIG. 2, the seal anti-counterfeit verification program 01 is stored in the memory 11; the processor 12 implements the following steps when executing the seal anti-counterfeit verification program 01 stored in the memory 11:

Step 1: Establish a reference database for the feature points of the reference stamp image.

The preferred embodiment of the present application acquires the seal image of the real seal as an image of the reference seal through an image acquisition device such as a camera, a scanner, a CCD camera, and gives each reference seal image a fixed ID, which corresponds to a specific real seal .

A feature point extraction method is used to extract feature points in the reference seal image, and data such as position information and descriptor information of the feature points contained in the reference seal image are stored in the reference database.

The feature point extraction method used in this proposal is SIFT (Scale Invariant Feature Transform) algorithm.

The SIFT operator is an operator with better performance in the image matching algorithm, and feature image registration based on the SIFT algorithm can be roughly divided into feature detection, description, and matching. Feature detection is performed in scale space. First, the image scale space is generated, and then the local extremum values in the scale space are detected, and then the local extremum points are accurately located by excluding low-contrast points and edge response points; When calculating the principal direction of each extremum point, the main direction of the extremum point and the histogram gradient direction of the area with the extreme point as the center are calculated to generate the feature point descriptor; the feature point descriptor can be used to find the match Features to establish connections between images.

SIFT feature detection mainly includes the following four basic steps: 1. Scale space extreme value detection: search for image positions on all scales. Gaussian differential functions are used to identify potential candidate points that are invariant to scale and rotation. 2. Feature point positioning: at each candidate position, a precise model is used to determine the position and scale. The selection of feature points depends on their stability. 3. Direction determination: Based on the local gradient direction of the image, one or more directions are assigned to each feature point location. All subsequent operations on image data are transformed with respect to the direction, scale, and position of the feature points, thereby providing invariance to these transformations. 4. Feature point description: In the neighborhood around each feature point, measure the local gradient of the image at the selected scale. These gradients are transformed into a representation that allows larger local shape deformations and light changes. In the description of feature points, the neighborhood of 16 × 16 is taken as the sampling window with the feature point as the center, and the relative direction of the sample point and the feature point is weighted by Gaussian into a direction histogram containing 8 bins, and finally 4 × 4 × 8 128-dimensional feature point descriptor.

Step 2: Extract the seal image to be checked from the document.

In the preferred embodiment of the present application, the process of extracting the seal image to be inspected includes:

1. Separate feature data of preset colors from the document.

It should be understood that most of the seals are red, so this embodiment uses red as an example.

In order to highlight the difference in value between red and other colors, this scheme proposes three ways to describe colors according to the characteristics of colors expressed in different color spaces. They are:

(1), RGB space-R-G, R-B.

The three components R, G, and B in the RGB space represent the mixed amounts of red, green, and blue, respectively. Intuitively speaking, the red R value, that is, the red component must be relatively large, in fact, the RGB value of the standard red in RGB space is (255,0,0), but a large R value does not mean that the color is close to red , The size of G value and B value will have a significant effect on the color, for example (255,255,0) represents yellow, and (255,0,255) represents purple. It can be seen from this that when the R value is large and the G and B values are small, the corresponding color is closer to red.

Correspondingly, this solution calculates the difference RG of the R component and the G component of a pixel, and the difference RB of the R component and the B component, when both the difference RG and the difference RB are greater than the corresponding preset threshold To determine that the pixel is red.

(2), HSV space-H.

H, S, and V in HSV space represent hue, saturation, and intensity, respectively. The description of colors in the HSV space is closer to humans ’understanding of colors: H is the hue, and the phase of the color is expressed in the form of angle values, from 0 ° to 360 °, which indicates the transition from red to green to blue and back to red, S Represents the fill ratio of the color under the corresponding hue, taking 0-100%, V represents the brightness, or light intensity, taking 0-100%. For the seal, the change in red is mainly affected by the light collected by the image (corresponding to the brightness V) and the wear of the paper itself (corresponding to the saturation S), so the value of H can better describe the "color" difference in the general sense .

Therefore, this solution calculates the H value of a pixel, and when the H value of the pixel is less than the corresponding preset threshold, it is determined that the pixel is red. (3) YCbCr space-deflection angle θ.

YCbCr space is not an absolute color space, but an encoding of RGB space. The encoding rules are as follows:

In fact, each component of the YCbCr space is still meaningful, Y represents the grayscale of the pixel, and Cb and Cr represent the blue and red density offset components of the pixel, respectively. In this solution, an angle θ between a vector mapped on the CbCr plane and the Cr axis can be used to describe the color of the pixel. The effect is similar to H in HSV, but the distribution of H is not the same. among them:

θ = atan2 (Cb, Cr);

When the calculated θ is less than the corresponding preset threshold, it is determined that the pixel is red. .

2. According to the characteristic data of the preset color, the seal image is separated by a double threshold color separation method.

In order to solve the problem that the noise is close to the color of the stamp itself, this scheme uses the k-means clustering method. Assuming that the θ value in the YCbCr space is used as the color description method, the image is divided into clusters according to the θ value of each pixel. To obtain the upper and lower bounds of the θ value of the cluster to which the seal color belongs [a, b]. The color of the seal edge tends to be closer to the background color, which means that the color of this part may be very close to the color of part of the noise and it is considered to be removed by noise, so this scheme adopts the double threshold method, based on k-means aggregation On the basis of the threshold [a, b] obtained by the class, the value range is first narrowed, and a strict threshold is used to obtain the pixels that absolutely belong to the seal, and the set of pixel points of the seal is formed, and then the value range is expanded to adopt a more relaxed Threshold range to determine whether the adjacent points of each point in the seal pixel set belong to the seal, so as to supplement the seal pixel set and ensure the integrity of the seal image.

3. Perform contour detection of the seal image, and cut out the seal image according to the detected contour.

In a preferred embodiment of the present application, the outline detection of the seal includes any one of the following methods or a combination of several methods:

(1). Edge detection. This solution uses the Canny edge detection algorithm based on Sobel operator. This method can detect the edge of the seal well, and the gradient value of the edge point can be obtained through Sobel operator.

(2) Random Hough transform detects ellipse. Random Hough transform can effectively detect ellipse, find the position of the ellipse and draw a more accurate ellipse outline, but its mechanism of randomly taking points makes it very susceptible to noise. Interference, easy to appear after reaching the maximum number of cycles and still can not find any elliptical results. In addition, its ellipse parameter calculation method also makes it impossible to calculate a very accurate ellipse equation.

(3). Direct least squares fitting of ellipse. Direct least squares fitting of ellipse is an optimized method of ellipse detection. It uses a set of sampling point data to solve unknown parameters in the elliptic equation by algebraic method. The key problem to solve with the least square method is the selection of sample points, so other methods need to be combined to obtain better samples.

(4) Combining random Hough transform and direct least squares fitting of ellipse to detect ellipse. Since the steps of using these two methods are more complicated, they will not be elaborated here. In short, using this method to detect the ellipse of the stamp image, the result often contains two ellipses with different sizes of the same centroid, corresponding to the inner and outer edges of the stamp border, and the larger one is obtained. Precise edge. The color of pixels outside the edge is set to white, and the image is cut according to the position of the edge to obtain an elliptical image within the outline.

Step 3: Extract the feature points of the seal image to be checked.

In this step, SIFT (Scale Invariant Feature Transform) algorithm can also be used as a method for extracting feature points.

Step 4: Match the feature points of the seal image to be tested with the reference seal image feature point reference database to obtain matching feature points.

Feature points have two types of descriptors such as feature vector and feature matrix.

Using feature vectors of feature points as feature point feature descriptors, this scheme uses the Euclidean distance evaluation method for feature point matching:

Obtain the feature vector V _j ^{recg of the} feature point i of the reference seal image = [x ₁ , x ₂ … x _n ], and the feature vector V _j ^{recg of the} feature point j of the seal image to be tested = [x ₁ ′, x ₂ ′… x _n '], Where n represents the dimension of the feature vector, and the Euclidean distance of the two feature vectors is:

Among them, the two corresponding feature vectors with the smallest Euclidean distance are matching feature vectors, and the corresponding reference stamp image feature points are matching feature points.

The feature matrix of feature points is used as the feature descriptor of feature points. Since such feature matrices are generally positive definite matrices, this solution uses the following methods to perform feature point matching:

待检验印章图像特征点j的特征矩阵

其中n表示特征矩阵的行、列数，两个特征矩阵的ρ距离为： Obtain the feature matrix of the reference point image feature point i

Feature matrix of feature point j of seal image to be tested

Where n represents the number of rows and columns of the feature matrix, and the ρ distance between the two feature matrices is:

λ _k (V _i ^ref , V _j ^recg ) λ _k V _i ^ref x _k -V _j ^recg x _k = 0 _{k = 1 ... n} , x _k ≠ 0,

Where V _i ^ref and V _j ^recg are two covariance matrices, and ρ (V _i ^ref , V _j ^recg ) represents their evaluation distance. λ _k (V _i ^ref , V _j ^recg ) represents the generalized eigenvalues of V _i ^ref and V _j ^recg and is calculated by the following formula: λ _k V _i ^ref x _k -V _j ^recg x _k = 0 _{k = 1 .. .n} , where x _k ≠ 0 is the generalized eigenvector of V _i ^ref and V _j ^recg , and n is the dimension of the generalized eigenvector.

Among them, the two corresponding feature vectors with the smallest distance between ρ (V _i ^ref and V _j ^recg ) are the matching feature vectors, and the corresponding feature points of the reference seal image and the feature image points of the seal image to be tested are matching feature points.

After the matching of the feature points of the seal image to be checked and the feature points of the reference seal image is completed, the feature points of the reference seal image [P ₁ P ₂ … 2 _n ] and the feature points of the seal image to be tested [P ₁ ′ P ₂ ′… P _n ′], where n represents the number of matching feature points. The above formula indicates that the reference image feature point P ₁ corresponds to the whole point P ₁ ′ of the image to be inspected, P ₂ corresponds to P ₂ ′, and so on.

Step 5: Count the number of matching feature points in the feature image of the seal image to be checked and the reference seal image, and verify the authenticity of the seal image to be checked according to the number of matching feature points.

Optionally, in other embodiments, the seal anti-counterfeiting verification program may also be divided into one or more modules, and the one or more modules are stored in the memory 11 and are processed by one or more processors (this embodiment is The processor 12) is executed to complete this application. The module referred to in this application refers to a series of computer program instruction segments capable of performing specific functions, and is used to describe the execution process of the seal anti-counterfeiting verification program in the seal anti-counterfeiting verification device.

For example, referring to FIG. 3, it is a schematic diagram of a program module of a seal anti-counterfeiting inspection program in an embodiment of a seal anti-counterfeiting inspection device of the present application. The module 20, the feature point extraction module 30, the feature point matching module 40 and the seal verification module 50, exemplarily:

The reference database creation 10 is used to: establish a reference database of feature points of reference seal images;

The seal image extraction module 20 is used to: extract the seal image to be checked from the document;

The feature point extraction module 30 is used to: extract the feature points of the seal image to be checked;

The feature point matching module 40 is used to: match the feature points of the seal image to be inspected with the reference seal image feature point reference database to obtain matching feature points; and

The seal verification module 50 is used to count the number of matching feature points in the feature points of the seal image to be verified and the reference seal image feature points, and verify the trueness of the seal image to be verified according to the number of matching feature points Pseudo.

The above reference database establishment 10, the seal image extraction module 20, the feature point extraction module 30, the feature point matching module 40, and the seal verification module 50, etc., when the program modules are executed, the functions or operation steps are substantially the same as those in the above embodiment, here No longer.

In addition, the embodiments of the present application also provide a computer-readable storage medium having a seal anti-counterfeit verification program stored on the computer-readable storage medium, which may be executed by one or more processors to implement the following operating:

Establish a reference database of feature points that refer to the seal image;

Extract the seal image to be checked from the document;

Extract the feature points of the seal image to be checked;

Matching the feature points of the seal image to be tested with the reference seal image feature point reference database to obtain matching feature points; and

The number of matching feature points in the feature image of the seal image to be tested and the reference seal image is counted, and the authenticity of the seal image to be tested is verified according to the number of matching feature points.

The specific implementation of the computer-readable storage medium of the present application is basically the same as the embodiments of the above-mentioned seal anti-counterfeiting verification device and method, and will not be repeated here.

It should be noted that the sequence numbers of the above embodiments of the present application are for description only, and do not represent the advantages and disadvantages of the embodiments. And the terms "include", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method that includes a series of elements includes not only those elements, but also includes no explicit The other elements listed may also include elements inherent to this process, device, article, or method. Without more restrictions, the element defined by the sentence "include one ..." does not exclude that there are other identical elements in the process, device, article or method that includes the element.

Through the description of the above embodiments, those skilled in the art can clearly understand that the methods in the above embodiments can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware, but in many cases the former is better Implementation. Based on this understanding, the technical solution of the present application can be embodied in the form of a software product in essence or part that contributes to the existing technology, and the computer software product is stored in a storage medium (such as ROM / RAM) as described above , Magnetic disks, optical disks), including several instructions to enable a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to perform the method described in each embodiment of the present application.

The above are only the preferred embodiments of the present application, and do not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made by using the description and drawings of this application, or directly or indirectly used in other related technical fields , The same reason is included in the scope of patent protection of this application.

Claims (20)

A seal anti-counterfeiting inspection method, characterized in that the method includes: Establish a reference database of feature points that refer to the seal image; Extract the seal image to be checked from the document; Extract the feature points of the seal image to be checked; Matching the feature points of the seal image to be inspected with the reference seal image feature points in the reference database to obtain matching feature points; and The number of matching feature points in the feature image of the seal image to be tested and the reference seal image is counted, and the authenticity of the seal image to be tested is verified according to the number of matching feature points. The seal anti-counterfeiting inspection method according to claim 1, wherein the establishment of a reference database of feature points of reference seal images includes: Obtain the seal image of the real seal as the reference seal image through the image acquisition device, and give each reference seal image a fixed ID, which corresponds to the specific real seal; A feature point extraction method is used to extract feature points in the reference seal image, and the position information and descriptor information of the feature points contained in the reference seal image are stored in the reference database. The seal anti-counterfeiting inspection method according to claim 1, wherein the extraction of the seal image to be inspected from the document includes: Separating feature data of preset colors from the document; According to the characteristic data of the preset color, the seal image is separated by a double threshold color separation method; and The contour of the seal image is detected, and the seal image is cut according to the detected contour. The seal anti-counterfeiting inspection method according to claim 2, wherein the extraction of the seal image to be inspected from the document includes: Separating feature data of preset colors from the document; According to the characteristic data of the preset color, the seal image is separated by a double threshold color separation method; and The contour of the seal image is detected, and the seal image is cut according to the detected contour. The seal anti-counterfeiting inspection method according to claim 3, wherein the preset color includes red, and the method for separating feature data of the preset color from the document includes one or more of the following The combination: Calculate the difference between R-G and R-B in the RGB space of the pixel, and if the difference between R-G and R-B is greater than the corresponding preset threshold, determine that the pixel is red; Calculate the H value of the pixel's HSV space, and if the H value of the pixel is less than the corresponding preset threshold, determine that the pixel is red; and Calculate the deflection angle θ of the YCbCr space of the pixel point = atan2 (Cb, Cr), and if the calculated θ is less than the corresponding preset threshold, determine that the pixel point is red. The seal anti-counterfeiting inspection method according to claim 2, wherein the preset color includes red, and the method for separating feature data of the preset color from the document includes one or more of the following The combination: Calculate the difference between R-G and R-B in the RGB space of the pixel, and if the difference between R-G and R-B is greater than the corresponding preset threshold, determine that the pixel is red; Calculate the H value of the pixel's HSV space, and if the H value of the pixel is less than the corresponding preset threshold, determine that the pixel is red; and Calculate the deflection angle θ of the YCbCr space of the pixel point = atan2 (Cb, Cr), and if the calculated θ is less than the corresponding preset threshold, determine that the pixel point is red. The seal anti-counterfeiting inspection method according to any one of claims 1 to 6, characterized in that the feature points of the seal image to be checked are matched with the reference seal image feature points in the reference database to obtain the matching feature points Steps, use one of the following methods: Obtain the feature vector V _j ^recg of the feature point i of the reference seal image = [x ₁ , x ₂ … x _n ] and the feature vector V _j ^{recg of the} feature point j of the seal image to be tested = [x ₁ ′, x ₂ ′… x _n ′], where n represents the dimension of the feature vector, and calculate the Euclidean distance between the two feature vectors:

Among them, the two corresponding feature vectors with the smallest Euclidean distance are matching feature vectors, and the corresponding reference stamp image feature points and the feature image points to be tested are matching feature points; or Obtain the feature matrix of the reference point image feature point i

Feature matrix of feature point j of seal image to be tested

Where n represents the number of rows and columns of the feature matrix, and the ρ distance between the two feature matrices is:

λ _k (V _i ^ref , V _j ^recg ) λ _k V _i ^ref x _k -V _j ^recg x _k = 0 _{k = 1 ... n} , x _k ≠ 0, Among them, the two corresponding feature vectors with the smallest distance between ρ (V _i ^ref and V _j ^recg ) are the matching feature vectors, and the corresponding feature points of the reference seal image and the feature image points of the seal image to be tested are matching feature points. A seal anti-counterfeiting inspection device, characterized in that the device includes a memory and a processor, and a seal anti-counterfeiting inspection program operable on the processor is stored on the memory, and the seal anti-counterfeiting inspection program is processed The following steps are implemented when the device is executed: Establish a reference database of feature points that refer to the seal image; Extract the seal image to be checked from the document; Extract the feature points of the seal image to be checked; Matching the feature points of the seal image to be inspected with the reference seal image feature points in the reference database to obtain matching feature points; and The number of matching feature points in the feature image of the seal image to be tested and the reference seal image is counted, and the authenticity of the seal image to be tested is verified according to the number of matching feature points. The seal anti-counterfeiting inspection device according to claim 8, wherein the feature point reference database for establishing a reference seal image includes: Obtain the seal image of the real seal as the reference seal image through the image acquisition device, and give each reference seal image a fixed ID, which corresponds to the specific real seal; A feature point extraction method is used to extract feature points in the reference seal image, and the position information and descriptor information of the feature points contained in the reference seal image are stored in the reference database. The seal anti-counterfeiting inspection device according to claim 8, wherein the extraction of the seal image to be inspected from the document includes: Separating feature data of preset colors from the document; According to the characteristic data of the preset color, the seal image is separated by a double threshold color separation method; and The contour of the seal image is detected, and the seal image is cut according to the detected contour. The seal anti-counterfeiting inspection device according to claim 9, wherein the extraction of the seal image to be inspected from the document includes: Separating feature data of preset colors from the document; According to the characteristic data of the preset color, the seal image is separated by a double threshold color separation method; and The contour of the seal image is detected, and the seal image is cut according to the detected contour. The seal anti-counterfeiting inspection device according to claim 9, wherein the preset color includes red, and the method for separating the feature data of the preset color from the document includes one or more of the following The combination: Calculate the difference between R-G and R-B in the RGB space of the pixel, and if the difference between R-G and R-B is greater than the corresponding preset threshold, determine that the pixel is red; Calculate the H value of the pixel's HSV space, and if the H value of the pixel is less than the corresponding preset threshold, determine that the pixel is red; and Calculate the deflection angle θ of the YCbCr space of the pixel point = atan2 (Cb, Cr), and if the calculated θ is less than the corresponding preset threshold, determine that the pixel point is red. The seal anti-counterfeiting inspection device according to claim 9, wherein the preset color includes red, and the method for separating the feature data of the preset color from the document includes one or more of the following The combination: Calculate the difference between R-G and R-B in the RGB space of the pixel, and if the difference between R-G and R-B is greater than the corresponding preset threshold, determine that the pixel is red; Calculate the H value of the pixel's HSV space, and if the H value of the pixel is less than the corresponding preset threshold, determine that the pixel is red; and Calculate the deflection angle θ of the YCbCr space of the pixel point = atan2 (Cb, Cr), and if the calculated θ is less than the corresponding preset threshold, determine that the pixel point is red. The seal anti-counterfeiting inspection device according to any one of claims 8 to 14, wherein the feature points of the seal image to be checked are matched with the reference seal image feature points in the reference database to obtain the matching feature points Steps, use one of the following methods: Obtain the feature vector V _j ^recg of the feature point i of the reference seal image = [x ₁ , x ₂ … x _n ] and the feature vector V _j ^{recg of the} feature point j of the seal image to be tested = [x ₁ ′, x ₂ ′… x _n ′], where n represents the dimension of the feature vector, and calculate the Euclidean distance between the two feature vectors:

Among them, the two corresponding feature vectors with the smallest Euclidean distance are matching feature vectors, and the corresponding reference stamp image feature points and the feature image points to be tested are matching feature points; or Obtain the feature matrix of the reference point image feature point i

Feature matrix of feature point j of seal image to be tested

Where n represents the number of rows and columns of the feature matrix, and the ρ distance between the two feature matrices is:

λ _k (V _i ^ref , V _j ^recg ) λ _k V _i ^ref x _k -V _j ^recg x _k = 0 _{k = 1 ... n} , x _k ≠ 0, Among them, the two corresponding feature vectors with the smallest distance between ρ (V _i ^ref and V _j ^recg ) are the matching feature vectors, and the corresponding feature points of the reference seal image and the feature image points of the seal image to be tested are matching feature points. A computer-readable storage medium, characterized in that a seal anti-counterfeiting verification program is stored on the computer-readable storage medium, and the seal anti-counterfeiting verification program can be executed by one or more processors to implement the following steps: Establish a reference database of feature points that refer to the seal image; Extract the seal image to be checked from the document; Extract the feature points of the seal image to be checked; Matching the feature points of the seal image to be inspected with the reference seal image feature points in the reference database to obtain matching feature points; and The number of matching feature points in the feature image of the seal image to be tested and the reference seal image is counted, and the authenticity of the seal image to be tested is verified according to the number of matching feature points. The computer-readable storage medium according to claim 15, wherein the feature point reference database for establishing a reference seal image includes: Obtain the seal image of the real seal as the reference seal image through the image acquisition device, and give each reference seal image a fixed ID, which corresponds to the specific real seal; A feature point extraction method is used to extract feature points in the reference seal image, and the position information and descriptor information of the feature points contained in the reference seal image are stored in the reference database. The computer-readable storage medium of claim 15, wherein the extracting of the seal image to be verified from the document includes: Separating feature data of preset colors from the document; According to the characteristic data of the preset color, the seal image is separated by a double threshold color separation method; and The contour of the seal image is detected, and the seal image is cut according to the detected contour. The computer-readable storage medium of claim 16, wherein the extraction of the stamp image to be verified from the document includes: Separating feature data of preset colors from the document; According to the characteristic data of the preset color, the seal image is separated by a double threshold color separation method; and The contour of the seal image is detected, and the seal image is cut according to the detected contour. The computer-readable storage medium according to claim 16 or 17, wherein the preset color includes red, and the method for separating feature data of the preset color from the document includes one of the following Or a combination of several: Calculate the difference between R-G and R-B in the RGB space of the pixel, and if the difference between R-G and R-B is greater than the corresponding preset threshold, determine that the pixel is red; Calculate the H value of the pixel's HSV space, and if the H value of the pixel is less than the corresponding preset threshold, determine that the pixel is red; and Calculate the deflection angle θ of the YCbCr space of the pixel point = atan2 (Cb, Cr), and if the calculated θ is less than the corresponding preset threshold, determine that the pixel point is red. The computer-readable storage medium according to claim 19, wherein the step of matching the feature points of the seal image to be checked with the reference seal image feature points in the reference database to obtain matching feature points is as follows One of the methods: Obtain the feature vector V _j ^recg of the feature point i of the reference seal image = [x ₁ , x ₂ … x _n ] and the feature vector V _j ^{recg of the} feature point j of the seal image to be tested = [x ₁ ′, x ₂ ′… x _n ′], where n represents the dimension of the feature vector, and calculate the Euclidean distance between the two feature vectors:

Among them, the two corresponding feature vectors with the smallest Euclidean distance are matching feature vectors, and the corresponding reference stamp image feature points and the feature image points to be tested are matching feature points; or Obtain the feature matrix of the reference point image feature point i

Feature matrix of feature point j of seal image to be tested

Where n represents the number of rows and columns of the feature matrix, and the ρ distance between the two feature matrices is:

λ _k (V _i ^ref , V _j ^recg ) λ _k V _i ^ref x _k -V _j ^recg x _k = 0 _{k = 1 ... n} , x _k ≠ 0, Among them, the two corresponding feature vectors with the smallest distance of ρ (V _i ^ref , V _j ^recg ) are matching feature vectors, and the corresponding reference stamp image feature points and the to-be-tested seal image feature points are matching feature points.

Images