A social security card data recognition method based on image processing

Abstract

The application provides a social security card data recognition method based on image processing, relates to the technical field of social security card data processing, and performs pretreatment on an original social security card image, aligns the original social security card image with a standard template, and obtains a structured image; the structured image is segmented according to minimum units to obtain a plurality of sub-regional blocks, and adjacent sub-regional blocks are constructed; adjacent sub-regional blocks belonging to a character sub-regional block are merged to obtain a character region block; adjacent sub-regional blocks belonging to an image sub-regional block are merged to obtain an image region block; character information recognition is performed on the character region block, image recognition is performed on the image region block, the image recognition result is used to verify the character information recognition result, and if the verification is successful, it is proved that the social security card data recognition is successful.

Description

Technical Field

[0001] This invention proposes a social security card data recognition method based on image processing, which relates to the field of social security card data processing technology. Background Technology

[0002] With the widespread use of social security cards in government affairs, finance and other scenarios, manual verification is inefficient and requires automated technology to achieve fast and accurate data extraction and verification. With the upgrade of the third-generation social security card, higher requirements are placed on the identification technology, and self-service terminals need to complete identity verification in seconds.

[0003] However, due to the complex collection scenarios of social security cards, the original images often have the following problems: strong light or backlighting causes overexposure or shadow occlusion of character areas; deviations in shooting angle or errors in scanning equipment cause geometric distortion of the image, affecting the accuracy of character segmentation; aging scanning equipment or shooting shake introduces noise such as spots and scratches, which obscure the edges of characters; the social security card layout contains multiple types of information, which are difficult to effectively parse using traditional methods, with elements such as text, tables, barcodes, and photos interspersed, such as names and ID numbers possibly located in different areas; differences exist between social security cards in different regions, causing the universal template to fail; fonts, font sizes, and colors vary, with handwritten and printed fonts coexisting; relying solely on character recognition is easily affected by image quality, and traditional segmentation methods based on thresholds or edge detection are prone to splitting continuous text into multiple sub-regions or mistakenly treating background noise as characters; the logical relationship between fields such as "gender" and "ethnicity" cannot be distinguished, leading to information misalignment. Most importantly, traditional solutions lack a verification mechanism for the recognition results. Character and image information are disconnected. For example, when the photo and name do not match, they cannot be detected automatically and require manual verification. This results in poor verification and anti-counterfeiting capabilities. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes a social security card data recognition method based on image processing, comprising:

[0005] The original social security card image is preprocessed and aligned with a standard template to obtain a structured image;

[0006] The structured image is segmented according to the smallest unit to obtain multiple sub-region blocks, and adjacent sub-region blocks are constructed.

[0007] Construct adjacent sub-region blocks, merge multiple adjacent sub-region blocks belonging to the character sub-region block to obtain the character region block; merge multiple adjacent sub-region blocks belonging to the image sub-region block to obtain the image region block;

[0008] Character information recognition is performed on the character region block, and image recognition is performed on the image region block. The image recognition result is used to verify the character information recognition result. If the verification is successful, it proves that the social security card data has been successfully recognized.

[0009] In a preferred embodiment, character information recognition is performed on the character region block using an equal-interval sampling method to select key points on the character region block; the stroke width from each key point to the character edge is calculated, and a set of stroke widths {Z1, Z2, ..., Zn} for m key points is constructed. k …,Z m The algorithm calculates the average and median of the stroke width set, uses the average and median as the stroke features of the character, and outputs the character information recognition result.

[0010] In a preferred embodiment, the average value W of the stroke width set avg for:

[0011] ;

[0012] Let the stroke width set {Z1, Z2, ..., Z} k …,Z m Sort the data in ascending order to get {Z(1), Z(2), ..., Z(m')}, and calculate the median W. med :

[0013] If m' is odd, then the median ;

[0014] If m' is even, then the median .

[0015] In a preferred embodiment, image recognition is performed on image regions to extract the face image region recognition results. A face image output module and a judgment module are constructed and trained alternately. The face image output module outputs a real face image based on the face image region recognition results. The judgment module judges the authenticity of the real face image output by the face image output module while judging the gender information. The gender information judgment result is used to verify the gender digits of the social security number in the character information recognition result.

[0016] In a preferred embodiment, a difference function L is constructed in the face image output module. G :

[0017]

[0018]

[0019]

[0020] Among them, L fake It is a difference function that guides the judgment module to determine the authenticity of the image, which is a function of the face image output module. It is the difference function used by the judgment module to classify gender based on the generated noisy face image G(z). The desired gender label is... The judgment module correctly classifies the generated noisy face image G(z) into the desired gender label. The probability is given by z, which is noise data sampled from the noise distribution pb(z), and E is the mathematical expectation symbol. This represents the probability that the generated noisy face image G(z) is judged as real by the judgment module.

[0021] In a preferred embodiment, the difference function L set in the determination module D Represented as:

[0022] L D =L real +L gender ;

[0023] in:

[0024] ;

[0025] ;

[0026] Among them, L real L is the difference function used by the judgment module to determine the authenticity of the real face image Q generated by the face image output module. gender is the difference function used by the judgment module to classify gender based on the real face image Q, where h is the gender label corresponding to the real face image. This is the output of the judgment module regarding the probability that the real face image Q is labeled with gender h; pa(Q) is the true data distribution of the real face image Q, E is the mathematical expectation symbol, and D... real (Q) represents the probability that the generated real face image Q is real.

[0027] In a preferred embodiment, image recognition is performed on the image region block to extract the recognition result of the bank logo sub-region block and output the bank name; the recognition result of the bank card number sub-region block of the character region block is obtained, and the issuing bank name is obtained by querying the mapping database through IIN; if the bank name matches the issuing bank name, the verification is successful.

[0028] In a preferred embodiment, adjacent sub-region blocks are constructed using adjacency weights, where the adjacency weight W between sub-region blocks i and j is... i,j The calculation formula is:

[0029]

[0030] Wherein, the center distance D(i,j) represents the distance between the geometric centers of sub-region blocks i and j. It's a hyperparameter.

[0031] In a preferred embodiment, the neighboring sub-region block verification method is as follows:

[0032] When the distance between the right boundary of sub-region block i and the left boundary of sub-region block j is ≤ M1 pixels, the sub-region blocks are judged to be horizontally adjacent.

[0033] If the distance between the lower boundary of sub-region block i and the upper boundary of sub-region block j is ≤ M2 pixels, the sub-region blocks are considered to be vertically adjacent.

[0034] In a preferred embodiment, a detection algorithm is used to extract the four vertices of the quadrilateral in the social security card image, fix the positions of three vertices, and dynamically adjust the coordinates of the remaining vertices to transform the social security card image into a structured image aligned with a standard template.

[0035] Compared with the prior art, the present invention has the following beneficial technical effects:

[0036] 1. By preprocessing, interference such as uneven lighting, tilting distortion, and printing defects in the original image are eliminated, making the character edges clear and the image area features stable, thus transforming the social security card image into a structured image and reducing processing errors caused by image diversity.

[0037] 2. The image is segmented into the smallest unit sub-region blocks, and regions with the same attributes (character regions or image regions) are merged based on adjacency relationships, which conforms to the physical structure of the social security card design (such as characters arranged in rows / columns, and image markers as independent blocks). Incomplete regions are repaired by adjacency merging, thereby improving the integrity of the region.

[0038] 3. For character regions (text and numbers), character recognition technology is used to quickly extract key information such as name, card number, and expiration date by leveraging its strong parsing ability for serialized symbols; for image regions (logos and patterns), image detection methods are used to identify graphic information such as card issuer logos and facial images; image recognition is used to verify character information, constructing a dual verification mechanism to form a dual-source evidence chain of "character-image". Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is a flowchart of the social security card data recognition method based on image processing according to the present invention;

[0041] Figure 2 This is a flowchart illustrating the character information recognition process for character region blocks according to the present invention. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0043] like Figure 1 The diagram shown is a flowchart of the social security card data recognition method based on image processing according to the present invention. The method includes:

[0044] To improve the efficiency and accuracy of subsequent data recognition, the original social security card image first needs to be preprocessed to obtain the preprocessed social security card image.

[0045] Specifically, the original social security card image is preprocessed, including converting the color image to a grayscale image to reduce the amount of data; using Gaussian filtering to remove noise from the image to make the image smoother; and enhancing the contrast of the image through histogram equalization to highlight text and key information.

[0046] Preferably, a grid sampler is used to implement the thin-plate spline interpolation algorithm to sample the original social security card image and process local distortions in the original image. Specifically, pixel values ​​are extracted from the original social security card image according to the position of the sampling grid, and these pixel values ​​are interpolated using the thin-plate spline interpolation algorithm to generate the interpolated image. During the interpolation process, the thin-plate spline interpolation algorithm considers local image deformations, thus making the interpolated image more natural and accurate.

[0047] Next, the preprocessed social security card image is aligned with a standard template to obtain a structured image.

[0048] Locate the four vertices of the quadrilateral in the preprocessed social security card image: fix the positions of three vertices, dynamically adjust the coordinates of the remaining vertices, transform the quadrilateral perspective to a preset standard template, eliminate local area distortion, and obtain a structured image aligned with the standard template.

[0049] Specifically, the contour of the social security card image is approximated as a quadrilateral by combining the contour approximation algorithm. The four vertices of the quadrilateral in the social security card image are extracted by the detection algorithm and arranged in ascending order according to the sum of the horizontal and vertical coordinates to determine the top left, top right, bottom left and bottom right vertices, and obtain the coordinate set of the four vertices P={p1, p2, p3, p4}, where p1, p2, p3, p4 correspond to the top left, top right, bottom left and bottom right vertices of the quadrilateral, respectively.

[0050] Define the coordinate set of the four vertices of the standard template as Q = {q1, q2, q3, q4}, where:

[0051] q1=(x0,y0), where (x0,y0) are the coordinates of the top left vertex q1, and the top left vertex is fixed as the preset reference point;

[0052] q2=(x0+w,y0), where (x0+w,y0) are the coordinates of the top right vertex q2, and w is the width of the standard template;

[0053] q3 = (x0, y0 + h), where (x0, y0 + h) are the coordinates of the lower left vertex q3, and h is the height of the standard template;

[0054] q4 = (x0 + w, y0 + h), where (x0 + w, y0 + h) are the coordinates of the lower right vertex q4.

[0055] The four vertices of the quadrilateral in the detected social security card image are matched with the four vertices of the standard template. First, the first three vertices of the quadrilateral in the social security card image are fixed. By adjusting only the coordinates of the fourth vertex p4, the bottom right vertex of the quadrilateral in the adjusted social security card image is obtained. .

[0056] By applying perspective transformation constraints, establish the mapping relationship between the vertices of the quadrilateral and the vertices of the standard template:

[0057]

[0058] in, To satisfy the constraints of the transformation matrix H of the perspective transformation, i.e. H is the transformation matrix, H is obtained through The mapping relationship was obtained by fitting.

[0059] The above method can effectively eliminate local deformation of social security card images and achieve precise alignment of structured images.

[0060] Next, the structured image is segmented according to the smallest unit to obtain multiple sub-region blocks, and adjacent sub-region blocks are constructed.

[0061] The structured image is segmented according to the smallest unit, resulting in multiple sub-region blocks. The top-left corner coordinate (x, y) of each sub-region block i is then determined. i1 ,y i1 Width and height information (w) i ,h i Normalize to the [0,1] interval.

[0062] By adjacency weight W i,j Construct adjacent sub-region blocks, with an adjacency weight W. i,j The closer the value is to 1, the closer the two sub-region blocks are.

[0063] Adjacency weight W i,j The calculation formula is as follows:

[0064]

[0065] Among them, W i,j The adjacency weight represents the adjacency weight between sub-region blocks i and j, which is used to measure the degree of spatial association between the two sub-region blocks. The larger the adjacency weight, the closer the spatial association between the two.

[0066] Center distance D(i,j): Represents the distance between the geometric centers of sub-region blocks i and j, reflecting the spatial relationship between the two sub-region blocks. The closer the distance, the better for W. i,j The greater the positive impact, the better.

[0067] It is a hyperparameter, preferably. =10 pixels.

[0068] Based on the standard structure of the social security card, the adjacency relationships of core sub-region blocks are predefined (e.g., the "Photo" sub-region block must be adjacent to the "Name" sub-region block), and W is forcibly set. i,j =1.

[0069] The method for verifying neighboring sub-region blocks is as follows:

[0070] If the distance between the right boundary of sub-region block i and the left boundary of sub-region block j is ≤ M1 pixels, the sub-region blocks are considered to be horizontally adjacent.

[0071] If the distance between the lower boundary of sub-region block i and the upper boundary of sub-region block j is ≤ M2 pixels, the sub-region blocks are considered to be vertically adjacent.

[0072] By adjacency weight W i,j The algorithm verifies adjacent sub-region blocks and represents the spatial relationship of adjacent sub-region blocks in the form of parameters, which facilitates subsequent processing using algorithms. This allows for the merging of multiple adjacent sub-region blocks that both belong to the character sub-region block to obtain the character region block, and the merging of multiple adjacent sub-region blocks that both belong to the image sub-region block to obtain the image region block.

[0073] In a preferred embodiment, for non-standard formats (such as social security cards from different provinces), reinforcement learning is used to automatically adjust adjacent thresholds to adapt to layout changes.

[0074] Next, character information recognition is performed on each character region block, and the recognition results are output. The recognition process is as follows: Figure 2 As shown, it includes the following steps:

[0075] (1) Select key points on each character region block.

[0076] Using an equal-interval sampling method, a key point is selected at regular pixel intervals d on the character region block. That is, the pixel distance between the selected key points is d. Let the set of key point coordinates be K = {(X...} k ,Y k )}.

[0077] (2) Calculate the stroke width from the key point to the character edge.

[0078] For the k-th key point P(X) k ,Y k Starting from this key point, search horizontally to the right to find the first background pixel (the pixel value of the background pixel is 0), and record the stroke width Z from this key point to the character edge. k .

[0079] The stroke width Z from each keypoint to the character edge k This reflects the local width of the strokes near the key point. By setting a reasonable interval d, these key points can represent the overall situation of the character area to a certain extent.

[0080] Construct a set of stroke widths {Z1, Z2, ..., Zn} for m key points. k …,Z m} Calculate the average value W of the set of stroke widths. avg and median W med This can statistically characterize the overall width of a stroke. Even with interval selection, as long as the interval d is reasonable (neither too large nor too small), it can avoid excessive computation while retaining enough information to reflect stroke features, thus achieving effective extraction of character stroke features.

[0081] (3) Calculate the stroke characteristics of the character.

[0082] Use the mean and median as stroke characteristics for characters.

[0083] The average value W of the stroke width set avg The calculation formula is:

[0084]

[0085] Let the stroke width set {Z1, Z2, ..., Z} k …,Z m Sort the data in ascending order to get {Z(1), Z(2), ..., Z(m')}. Calculate the median W using the following method. med :

[0086] If m' is odd, then the median ;

[0087] If m' is even, then the median ;

[0088] Through the above steps, each character can be converted into stroke features containing the average and median.

[0089] Due to the different stroke characteristics of numbers or characters (W) avg W med There are differences between numbers and characters. By using these differences to train classification models (such as support vector machines, neural networks, etc.), the recognition of numbers and characters can be achieved.

[0090] For example, the stroke structures of "1" and "7" are different, and their stroke features will be clearly distinguishable, so the classification model can classify them accurately.

[0091] (4) Output the recognition results of each character region block.

[0092] Output the recognition text results for fields such as name and card number. Fill the structured database with the recognition text according to the spatial position of adjacent sub-region blocks, such as in the form of key-value pairs: {"adjacency weight": "1"; "name": "Zhang San"; "card number": "XXXX"}.

[0093] (5) Verify the data logic of the identification results and trigger the anomaly marker.

[0094] Data logic includes, for example, whether the social security card number, bank card number, and social security number have the required number of digits.

[0095] The data populated into the structured database is validated using predefined rules. Each field is checked individually to ensure it conforms to the corresponding rules. If any field is found to be inconsistent with the rules, an exception is flagged.

[0096] For data marked as anomalous, appropriate processing measures are taken. Anomaly information can be recorded, including the anomalous field, the reason for the anomaly, and the specific identified text content. Simultaneously, the anomalous data is fed back to the manual review process for further verification and correction.

[0097] After verification and anomaly handling, the final structured social security card data is output. This data is presented in a clear and standardized format and can be directly used for subsequent data analysis, storage, or other business processes.

[0098] In a preferred embodiment, image recognition is performed on the image region block to extract the recognition result of the face image region block, and gender information is determined to verify the gender digits of the social security number in the recognition result of the character region block.

[0099] Social Security Number (i.e., ID number) Gender Character Rules: The 17th digit of the Social Security Number indicates gender; odd numbers represent males, and even numbers represent females. Extract the identified 17th digit of the Social Security Number and determine its odd / even value and gender.

[0100] Specifically, image recognition is performed on image regions, the recognition results of face image regions are extracted, and a face image output module and a judgment module are constructed.

[0101] The face image output module and the judgment module are trained alternately. The face image output module outputs a real face image based on the recognition result of the face image region block. The judgment module judges the authenticity of the real face image output by the face image output module, and at the same time judges the gender information. The judgment result of the gender information is used to verify the gender digit of the social security number in the character information recognition result.

[0102] In order for the judgment module to simultaneously determine the authenticity and gender probability of the generated face image, the judgment module is configured to output two results: one is the judgment of the authenticity of the face image, and the other is the judgment of gender.

[0103] Construct the difference function L in the face image output module G :

[0104]

[0105]

[0106]

[0107] Among them, L fake It is a difference function that guides the judgment module to determine the authenticity of the image, which is a function of the face image output module. It is the difference function used by the judgment module to classify gender based on the generated noisy face image G(z). The desired gender label is... The judgment module correctly classifies the generated noisy face image G(z) into the desired gender label. The probability is given by z, which is noise data sampled from the noise distribution pb(z), and E is the mathematical expectation symbol. This represents the probability that the generated noisy face image G(z) is judged as real by the judgment module.

[0108] After the above training, the noisy face image G(z) generated by the face image output module is finally trained to generate a real face image Q.

[0109] Construct the difference function L in the judgment module D :

[0110] L D =L real +L gender ;

[0111] in:

[0112] ;

[0113] ;

[0114] Among them, L real L is the difference function used by the judgment module to determine the authenticity of the generated real face image Q. gender is the difference function used by the judgment module to classify gender based on the real face image Q, where h is the gender label corresponding to the real face image. This is the output of the judgment module regarding the probability that the real face image Q is labeled with gender h; pa(Q) is the true data distribution of the real face image Q, E is the mathematical expectation symbol, and D... real (Q) represents the probability that the generated real face image Q is real.

[0115] By maximizing the probability of real face images being correctly identified as real and minimizing the probability of generated real face images being incorrectly identified as real, the judgment module is trained to better distinguish between real and generated real face images; at the same time, by maximizing the probability of correct gender classification, the judgment module is trained to perform well on gender classification tasks.

[0116] Ultimately, by alternating the training of the face image output module and the judgment module, the face image output module is able to generate realistic face images with specific gender characteristics, while the judgment module is able to accurately determine the authenticity and gender of the image.

[0117] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0118] In one embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0119] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and executes the computer instructions, causing the computer device to perform the steps in the above method embodiments.

[0120] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. The databases involved in the embodiments provided in this application can include at least one of relational and non-relational databases. Non-relational databases can include, but are not limited to, distributed databases based on blockchain. The processors involved in the embodiments provided in this application can be general-purpose processors, central processing units, graphics processors, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0121] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0122] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for recognizing social security card data based on image processing, characterized in that, include: The original social security card image is preprocessed and aligned with a standard template to obtain a structured image; The structured image is segmented according to the smallest unit to obtain multiple sub-region blocks, and adjacent sub-region blocks are constructed. Construct adjacent sub-region blocks, merge multiple adjacent sub-region blocks belonging to the character sub-region block to obtain the character region block; merge multiple adjacent sub-region blocks belonging to the image sub-region block to obtain the image region block; Character information recognition is performed on the character region block, and image recognition is performed on the image region block. The image recognition result is used to verify the character information recognition result. If the verification is successful, it proves that the social security card data has been successfully recognized.

2. The social security card data recognition method based on image processing according to claim 1, characterized in that, For the character region block, character information recognition is performed by using an equal-interval sampling method to select key points on the character region block; Calculate the stroke width from each keypoint to the character edge, and construct a set of stroke widths {Z1, Z2, ..., Zn} for m keypoints. k …,Z m The algorithm calculates the average and median of the stroke width set, uses the average and median as the stroke features of the character, and outputs the character information recognition result.

3. The social security card data recognition method based on image processing according to claim 2, characterized in that, The average value W of the stroke width set avg for: ； Let the stroke width set {Z1, Z2, ..., Z} k …,Z m Sort the data in ascending order to get {Z(1), Z(2), ..., Z(m')}, and calculate the median W. med : If m' is odd, then the median ; If m' is even, then the median .

4. The social security card data recognition method based on image processing according to claim 1, characterized in that, Image recognition is performed on image regions to extract the face image region recognition results. A face image output module and a judgment module are constructed and trained alternately. The face image output module outputs a real face image based on the face image region recognition results. The judgment module judges the authenticity of the real face image output by the face image output module and performs gender information judgment. The gender information judgment result is used to verify the gender digits of the social security number in the character information recognition result.

5. The social security card data recognition method based on image processing according to claim 4, characterized in that, Construct the difference function L in the face image output module G : ； ； ； Among them, L fake It is a difference function that guides the judgment module to determine the authenticity of the image, which is a function of the face image output module. It is the difference function used by the judgment module to classify gender based on the generated noisy face image G(z). The desired gender label is... The judgment module correctly classifies the generated noisy face image G(z) into the desired gender label. The probability is given by z, which is noise data sampled from the noise distribution pb(z), and E is the mathematical expectation symbol. This represents the probability that the generated noisy face image G(z) is judged as real by the judgment module.

6. The social security card data recognition method based on image processing according to claim 5, characterized in that, The difference function L set in the judgment module D Represented as: L D =L real +L gender ; in: ； ； Among them, L real L is the difference function used by the judgment module to determine the authenticity of the real face image Q generated by the face image output module. gender is the difference function used by the judgment module to classify gender based on the real face image Q, where h is the gender label corresponding to the real face image. This is the output of the judgment module regarding the probability that the real face image Q is labeled with gender h; pa(Q) is the true data distribution of the real face image Q, E is the mathematical expectation symbol, and D... real (Q) represents the probability that the generated real face image Q is real.

7. The social security card data recognition method based on image processing according to claim 1, characterized in that, For each image region block, image recognition is performed to extract the recognition result of the bank logo sub-region block and output the bank name; the recognition result of the bank card number sub-region block of the character region block is obtained, and the issuing bank name is obtained by querying the mapping database through IIN. If the bank name matches the issuing bank name, the verification is successful.

8. The social security card data recognition method based on image processing according to claim 1, characterized in that, Adjacent sub-region blocks are constructed using adjacency weights, where W is the adjacency weight between sub-region blocks i and j. i,j The calculation formula is: ； Wherein, the center distance D(i,j) represents the distance between the geometric centers of sub-region blocks i and j. It's a hyperparameter.

9. The social security card data recognition method based on image processing according to claim 8, characterized in that, The method for verifying neighboring sub-region blocks is as follows: When the distance between the right boundary of sub-region block i and the left boundary of sub-region block j is ≤ M1 pixels, the sub-region blocks are judged to be horizontally adjacent. If the distance between the lower boundary of sub-region block i and the upper boundary of sub-region block j is ≤ M2 pixels, the sub-region blocks are considered to be vertically adjacent.

10. The social security card data recognition method based on image processing according to claim 1, characterized in that, A detection algorithm is used to extract the four vertices of the quadrilateral in the social security card image. The positions of three vertices are fixed, and the coordinates of the remaining vertices are dynamically adjusted to transform the social security card image into a structured image aligned with a standard template.

Images