Handwritten signature normalization detection method and device and storage medium

Abstract

The invention discloses a handwritten signature normalization detection method and device and a storage medium, relates to the technical field of digital medical treatment and the technical field of financial science and technology, and mainly aims to improve the detection comprehensiveness and detection precision of signature normalization. Comprising the steps of obtaining a to-be-detected handwritten signature of a target user and writing process data when the to-be-detected handwritten signature is written; judging whether the handwritten signature to be detected is matched with the real name of the target user or not; if the handwritten signature to be detected is matched with the real name, performing stroke connection state detection on the handwritten signature to be detected based on the writing process data to obtain a stroke connection state detection result, and performing stroke integrity detection on the handwritten signature to be detected to obtain a stroke integrity detection result; and determining the normalization of the handwritten signature to be detected based on the stroke connection state detection result and the stroke integrity detection result.

Description

Technical Field

[0001] This invention relates to the fields of digital medical technology and financial technology, and in particular to a method, apparatus and storage medium for detecting the standardization of handwritten signatures. Background Technology

[0002] In handwritten signature processing scenarios, occasional instances of non-standard signatures occur. For example, in a rental scenario, an improperly written signature can render the signature legally invalid. Therefore, to ensure smooth workflow, signature verification is necessary.

[0003] Currently, signature conformity checks are typically performed by calculating the similarity between a handwritten signature and a user's registered name. However, this similarity check method only determines whether a signature conforms to conformity standards if most of the handwritten signature matches the user's registered name. Therefore, this method can only detect whether the user is signing their real name, but it cannot detect the conformity of each stroke in the signature, resulting in an incomplete signature conformity check and thus low accuracy. Summary of the Invention

[0004] This invention provides a method, apparatus, and storage medium for detecting the standardization of handwritten signatures, which mainly improves the comprehensiveness and accuracy of signature standardization detection.

[0005] According to a first aspect of the present invention, a method for detecting the conformity of handwritten signatures is provided, comprising: Obtain the handwritten signature to be detected from the target user and the writing process data when writing the handwritten signature to be detected; Determine whether the handwritten signature to be detected matches the real name of the target user; If the handwritten signature to be detected matches the real name, then the handwritten signature to be detected is subjected to stroke connection state detection based on the writing process data to obtain stroke connection state detection result, and the handwritten signature to be detected is subjected to stroke integrity detection to obtain stroke integrity detection result. Based on the stroke connection state detection results and the stroke integrity detection results, the standardization of the handwritten signature to be detected is determined.

[0006] Optionally, the writing process data includes writing trajectory information and the number of stroke intersections; Based on the writing process data, the stroke connection state of the handwritten signature to be detected is detected to obtain the stroke connection state detection result, including: Based on the writing trajectory information, breakpoint detection within the strokes is performed on the handwritten signature to be detected to obtain the breakpoint detection result within the strokes. Based on the number of stroke intersections, the handwritten signature to be detected is subjected to stroke disconnection detection to obtain the stroke end-connection detection result. Based on the intra-stroke breakpoint detection results and the inter-stroke connection detection results, the stroke connection status detection results of the handwritten signature to be detected are determined.

[0007] Optionally, Based on the writing trajectory information, breakpoint detection is performed within the strokes of the handwritten signature to be detected, resulting in breakpoint detection results, including: In the handwritten signature to be detected, each stroke to be detected is identified, and any stroke to be detected in each stroke to be detected is taken as a target stroke to be detected. Multiple trajectory points are determined at equal intervals on the writing trajectory corresponding to the target stroke to be detected. Based on the writing trajectory information, the position coordinates and writing timestamp of each trajectory point are determined, and the spatial distance and time difference between each pair of adjacent trajectory points are determined. If the spatial distance of any trajectory point is greater than a preset distance threshold and the corresponding time difference is greater than a preset time threshold, then it is determined that there is a breakpoint in the target stroke to be detected; otherwise, it is determined that there is no breakpoint in the target stroke to be detected. Based on the number of stroke intersections, the handwritten signature to be detected is subjected to stroke disconnection detection to obtain stroke end-connection detection results, including: Construct the stroke graph structure of the handwritten signature to be detected, determine the number of connected components in the stroke graph structure, and obtain the number of standard connected components in the standard stroke graph structure of the standard signature and the number of standard intersections between each stroke in the standard signature, wherein the standard signature is a compliance signature of the real name corresponding to the handwritten signature to be detected. If the number of connected components is greater than the number of standard connected components, and / or the number of stroke intersections is less than the number of standard intersections, then it is determined that there are breaks between strokes in the handwritten signature to be detected; otherwise, it is determined that there are no breaks between strokes in the handwritten signature to be detected.

[0008] Optionally, the handwritten signature to be detected is subjected to stroke integrity detection to obtain stroke integrity detection results, including: Constructing a preset signature library, wherein the method for constructing the preset signature library includes: obtaining complete signature images of various characters and their corresponding strokes; dividing the complete signature images into multiple grids and determining the gray mean, gray standard deviation, and stroke density of each grid; concatenating the gray mean, gray standard deviation, and stroke density into a standard feature vector, and constructing the preset signature library from the complete signature images and their corresponding standard feature vectors; Determine a matching complete signature image from the preset signature library that matches the handwritten signature to be detected; The method involves determining the feature vector to be detected for the handwritten signature to be detected, calculating the similarity distance between the handwritten signature to be detected and the matching complete signature image based on the feature vector to be detected and the standard feature vector corresponding to the matching complete signature image, and determining whether the similarity distance is greater than a preset similarity threshold. If it is, the handwritten signature to be detected is determined to have complete strokes; otherwise, the handwritten signature to be detected is determined to have incomplete strokes. The method for determining the preset similarity threshold includes: obtaining different styles of complete stroke signatures and different styles of incomplete stroke signatures for the same sample text; calculating the complete stroke similarity distance between each pair of complete stroke signatures and calculating the incomplete stroke similarity distance between each incomplete stroke signature and each complete stroke signature; and determining the preset similarity threshold based on the complete stroke similarity distance and the incomplete stroke similarity distance.

[0009] Optionally, the handwritten signature to be detected is subjected to stroke integrity detection to obtain stroke integrity detection results, including: The handwritten signature to be detected is input into a preset integrity detection model for detection to obtain the stroke integrity detection result of the handwritten signature to be detected. The preset integrity detection model is pre-built based on a sample dataset, which includes sample handwritten signatures with annotation information, and the annotation information is the actual stroke integrity of the sample handwritten signature.

[0010] Optionally, based on the stroke connection state detection result and the stroke integrity detection result, the standardization of the handwritten signature to be detected is determined, including: Determine the stroke connection state score and its corresponding connection state weight coefficient corresponding to the stroke connection state detection result, and determine the stroke integrity score and its corresponding integrity weight coefficient corresponding to the stroke integrity detection result. Based on the connection state weight coefficient and the integrity weight coefficient, the stroke connection state score and the stroke integrity score are weighted and summed to obtain a comprehensive score, and the standardization of the handwritten signature to be detected is determined based on the comprehensive score.

[0011] Optionally, determining whether the handwritten signature to be detected matches the real name of the target user includes: The handwritten signature to be detected is segmented into multiple segments to be detected, and the real name of the target user is segmented into multiple real segments. Each of the characters to be detected and the corresponding real characters at their respective positions are matched for similarity to obtain a similarity matching score for each character position. The position weight of each character position is determined. Based on the position weight, the similarity matching scores of each character are weighted and summed to obtain a comprehensive score. Based on the comprehensive score, it is determined whether the handwritten signature to be detected matches the real name of the target user.

[0012] According to a second aspect of the present invention, a device for detecting the conformity of handwritten signatures is provided, comprising: The acquisition unit is used to acquire the handwritten signature to be detected of the target user and the writing process data when writing the handwritten signature to be detected; A judgment unit is used to determine whether the handwritten signature to be detected matches the real name of the target user; The detection unit is configured to, if the handwritten signature to be detected matches the real name, perform stroke connection state detection on the handwritten signature to be detected based on the writing process data to obtain a stroke connection state detection result, and perform stroke integrity detection on the handwritten signature to be detected to obtain a stroke integrity detection result. The determining unit is used to determine the standardization of the handwritten signature to be detected based on the stroke connection state detection result and the stroke integrity detection result.

[0013] According to a third aspect of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the above-described method for detecting the conformity of handwritten signatures.

[0014] According to a fourth aspect of the present invention, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-described method for detecting the conformity of handwritten signatures.

[0015] According to the present invention, a method, apparatus, and storage medium for detecting the standardization of handwritten signatures provide a method for detecting the standardization of signatures. Compared with the current method of detecting signatures by calculating the similarity between handwritten signatures and user registered names, the present invention determines whether the handwritten signature to be detected is the user's real registered name. After determining that the handwritten signature is the user's real registered name, it further performs stroke connection state detection and stroke integrity detection on the handwritten signature. This allows for detection of handwritten signatures from multiple aspects, ensuring the comprehensiveness of handwritten signature detection and thus improving the accuracy of handwritten signature standardization detection. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1This invention provides a flowchart of a method for detecting the standardization of handwritten signatures according to an embodiment of the present invention. Figure 2 This invention provides a flowchart of another method for detecting the standardization of handwritten signatures according to an embodiment of the invention. Figure 3 This diagram illustrates the structure of a handwritten signature conformity detection device provided in an embodiment of the present invention. Figure 4 This invention provides a schematic diagram of the structure of another handwritten signature conformity detection device according to an embodiment of the invention. Figure 5 A schematic diagram of the physical structure of a computer device provided in an embodiment of the present invention is shown. Detailed Implementation

[0017] The present invention will be described in detail below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the present application can be combined with each other.

[0018] Currently, the method of verifying signature conformity by calculating the similarity between handwritten signatures and user registered names can only detect whether the signature belongs to the user's real name. This method is not comprehensive enough, resulting in low accuracy in signature conformity verification.

[0019] To address the aforementioned problems, embodiments of the present invention provide a method for detecting the standardization of handwritten signatures, such as... Figure 1 As shown, the method includes: 101. Obtain the handwritten signature to be tested from the target user and the writing process data when writing the handwritten signature to be tested.

[0020] The handwritten signature to be detected can be a handwritten signature of the target user purchasing insurance in an insurance contract, a handwritten signature of the patient's family member on a surgical risk disclosure form, or any other handwritten signature in any scenario. The writing process data includes information such as the writing trajectory information when signing the signature (such as the position coordinates and timestamps of the writing trajectory points), the number of stroke intersections, and the angle of stroke tilt.

[0021] In this embodiment of the invention, a video acquisition device is used to acquire video data of a target user signing a handwritten signature, and the handwritten signature image to be detected is extracted from the video data. The handwritten signature to be detected is identified from the handwritten signature image to be detected, and the writing process data of the handwritten signature to be detected is extracted from the video data.

[0022] 102. Determine whether the handwritten signature to be detected matches the target user's real name.

[0023] The target user's real name is the legally valid registered real name.

[0024] In this embodiment of the invention, the handwritten signature to be detected is matched with the target user's real name for similarity. If the similarity between the handwritten signature to be detected and the real name is greater than a preset similarity threshold (wherein the preset similarity threshold is set according to actual needs), then the handwritten signature to be detected is determined to match the real name; otherwise, it is determined that the handwritten signature to be detected does not match the real name. This embodiment of the invention pre-determines whether the user's signature is their real name. If the user's signature is not their real name, the connection state detection and integrity detection of strokes are stopped. Only when it is the real name will the connection state detection and integrity detection of strokes continue. This avoids the time and resources wasted on stroke detection when the user's signature is not their real name.

[0025] 103. If the handwritten signature to be detected matches the real name, then the stroke connection status of the handwritten signature to be detected is detected based on the writing process data to obtain the stroke connection status detection result, and the stroke integrity of the handwritten signature to be detected is detected to obtain the stroke integrity detection result.

[0026] In this embodiment of the invention, when the target user signs with their real name, it is necessary to detect the connectivity between strokes and the coherence within each stroke of the handwritten signature based on data such as the writing trajectory points when the user signs their name and the number of intersections between strokes in the handwritten signature. Simultaneously, it is also necessary to detect whether there are missing strokes in the handwritten signature. Specifically, the method for detecting missing strokes includes: inputting the handwritten signature to be detected into a preset integrity detection model for detection, and obtaining the stroke integrity detection result of the handwritten signature to be detected. The preset integrity detection model is pre-constructed based on a sample dataset, which includes sample handwritten signatures with annotation information, where the annotation information represents the actual stroke integrity of the sample handwritten signature.

[0027] Specifically, to improve the detection accuracy of the preset integrity detection model, it is first necessary to train and construct the preset integrity detection model. Based on this, the method includes: constructing a preset initial integrity detection model; obtaining a sample dataset, wherein the sample dataset includes multiple handwritten signature images with annotation information, the annotation information including annotation boxes and the missing strokes of the handwritten signature within the annotation boxes; dividing the sample dataset into a training set and a test set, using the training set to train the preset initial integrity detection model, and using the test set to test the trained preset initial integrity detection model, and finally using the trained preset initial integrity detection model that meets the test conditions as the preset integrity detection model.

[0028] Specifically, during model training, a pre-defined initial integrity detection model is first constructed, followed by downloading a sample dataset from the network. The dataset is ensured to contain all necessary files, including various handwritten signature images. The labeled files are converted to a format understandable by the pre-defined initial integrity detection model. Finally, the model is trained and tested. Specifically, the dataset can be divided first: using randomness or a specific strategy (such as stratified sampling), the sample dataset is divided into training and testing sets. The training set is then used to train the model, and the testing set is used to test the trained model, evaluating its performance on unseen data. Metrics such as mCP, precision, and recall on the testing set are calculated and recorded. If the model performance does not meet requirements, it can return to the training phase for further iterations or adjustments. This process yields a pre-defined integrity detection model that meets the requirements. Finally, the handwritten signature to be detected is directly input into the pre-defined integrity detection model, which can directly output information such as whether the handwritten signature is missing strokes, and which specific strokes are missing.

[0029] After determining that the handwritten signature is the user's real name registered, this embodiment of the invention further performs stroke connection status detection and stroke integrity detection on the handwritten signature. It can detect the handwritten signature from multiple aspects, ensuring the comprehensiveness of the handwritten signature detection and thus improving the detection accuracy of the handwritten signature's standardization.

[0030] 104. Based on the stroke connection state detection results and the stroke integrity detection results, determine the standardization of the handwritten signature to be detected.

[0031] In this embodiment of the invention, to determine the standardization of the handwritten signature to be detected, step 104 specifically includes: determining the stroke connection state score and its corresponding connection state weight coefficient corresponding to the stroke connection state detection result, and determining the stroke integrity score and its corresponding integrity weight coefficient corresponding to the stroke integrity detection result; based on the connection state weight coefficient and the integrity weight coefficient, performing a weighted summation on the stroke connection state score and the stroke integrity score to obtain a comprehensive score, and determining the standardization of the handwritten signature to be detected based on the comprehensive score.

[0032] Specifically, a reasonable stroke connection status score is set for the stroke connection status detection result. For example, if multiple strokes that should be connected are not connected, a lower stroke connection status score is assigned. Similarly, a reasonable stroke integrity score is set for the stroke integrity detection result. The more missing strokes, the lower the stroke integrity score. Weighting coefficients are assigned to the stroke connection status score and the stroke integrity score according to actual needs. Then, the stroke connection status score and the stroke integrity score are weighted and summed. Finally, the standardization of the handwritten signature is determined based on the weighted sum result. A higher weighted sum score indicates a more standard handwritten signature, and a lower weighted sum score indicates a less standard handwritten signature. This embodiment of the invention comprehensively analyzes whether the handwritten signature is a real name, the stroke connection status in the handwritten signature, and whether any strokes are missing to determine the standardization of the signature, ensuring the comprehensiveness of the analysis and thus improving the accuracy of signature standardization detection.

[0033] According to the present invention, a method for detecting the standardization of handwritten signatures, compared with the current method of detecting signature standardization by calculating the similarity between the handwritten signature and the user's registered name, the present invention, after determining that the handwritten signature is the user's registered real name, further performs stroke connection state detection and stroke integrity detection on the handwritten signature. It can detect the handwritten signature from multiple aspects, ensuring the comprehensiveness of the handwritten signature detection, thereby improving the accuracy of handwritten signature standardization detection.

[0034] Furthermore, to better illustrate the process of performing integrity checks on handwritten signatures, as a refinement and extension of the above embodiments, this invention provides another method for detecting the standardization of handwritten signatures, such as... Figure 2 As shown, the method includes: 201. Obtain the handwritten signature to be detected from the target user and the writing process data when writing the handwritten signature to be detected. The writing process data includes writing trajectory information and the number of stroke intersections.

[0035] Specifically, video information of the user during the signing process is collected through a camera device, and the handwritten signature to be detected and the writing process data are extracted from the video information.

[0036] 202. Determine whether the handwritten signature to be detected matches the real name of the target user.

[0037] For the embodiments of the present invention, in order to determine whether it is necessary to perform stroke connection state detection and stroke integrity detection on the handwritten signature to be detected in the subsequent process, it is first necessary to determine whether the user signs his real name. Based on this, step 202 specifically includes: segmenting the handwritten signature to be detected to obtain a plurality of segmented characters to be detected, and segmenting the real name of the target user to obtain a plurality of real segmented characters; respectively performing similarity matching between each of the segmented characters to be detected and the real segmented characters at the corresponding character positions to obtain a similarity matching score for each character position; respectively determining the position weights of each character position, and based on the position weights, performing weighted summation on each of the similarity matching scores to obtain a comprehensive score, and based on the comprehensive score, determining whether the handwritten signature to be detected matches the real name of the target user. [[ID=II]]

[0038] Specifically, if the handwritten signature to be detected is "Zhang Yao" and the legally effective real name registered by the target user is "Chi Zhu", then the segmented characters to be detected corresponding to the handwritten signature to be detected are "弓 长 王 䍃", and the real segmented characters corresponding to the real name are "弓 也 王 朱". Then, similarity calculation is performed on the characters at the same character positions. It is obtained that the two characters at the first character position are completely matched, and the score is set to 10 points (full score). The complete similarity between the two characters at the second character position is relatively low, and the score is set to 1 point (full score). The two characters at the third character position are completely matched, and the score is set to 10 points (full score). The two characters at the fourth character position are completely matched, and the score is set to 10 points (full score). Since the surname in the name is relatively important, a higher position weight should be set for it. For example, the position weight of the first character position is 0.35, the position weight of the second character position is 0.35, the position weight of the third character position is 0.2, and the position weight of the second character position is 0.1. Then the final comprehensive score is 7.85. If the score threshold set according to actual requirements is 9, the comprehensive score is less than this score threshold, and it is finally determined that the handwritten signature to be detected does not match the real name, that is, the user does not sign his registered real name. By splitting the signature and the real name into single characters for comparison, the embodiments of the present invention avoid misjudgment caused by local differences (such as connected strokes and occlusion) in the overall similarity calculation, thereby improving the matching accuracy.

[0039] 2) If the handwritten signature to be detected matches the real name, then based on the writing trajectory information, perform breakpoint detection within the strokes of the handwritten signature to be detected to obtain the breakpoint detection result within the strokes.

[0040] For the embodiments of the present invention, if the handwritten signature to be detected does not match the real name, there is no need to continue the stroke detection of the handwritten signature to be detected, avoiding resource waste and time cost consumption. If the handwritten signature to be detected matches the real name, it is necessary to continue the stroke detection of the handwritten signature to be detected. The stroke detection includes the breakpoint detection within a single stroke. The specific method for detecting breakpoints in a single stroke includes: identifying each stroke to be detected in the handwritten signature to be detected, taking any stroke to be detected in each stroke to be detected as a target stroke to be detected, and determining a plurality of trajectory points at equal intervals on the writing trajectory corresponding to the target stroke to be detected; based on the writing trajectory information, determining the position coordinates and writing timestamps of each trajectory point, determining the spatial distance and time difference between adjacent trajectory points for each pair of adjacent trajectory points. If the spatial distance of any trajectory point is greater than the preset distance threshold and the corresponding time difference between trajectory points is greater than the preset time threshold, it is determined that there is a breakpoint within the target stroke to be detected, otherwise it is determined that there is no breakpoint within the target stroke to be detected.

[0041] Among them, the writing trajectory refers to the trajectory where the visible ink trace of the writing tool is left on the paper. Specifically, for example, if the real name of the user is "Wang Yi", if any target stroke to be detected in the handwritten signature to be detected is the first stroke "- -" in Wang, a plurality of trajectory points are determined at equal intervals on the first horizontal line in "- -" and a plurality of trajectory points are determined at equal intervals on the second horizontal line (the determination of trajectory points needs to be based on the starting point and ending point of the stroke), and the position coordinates and writing timestamps of each trajectory point are determined and recorded as , where is the position coordinate of the i-th trajectory point, is the writing timestamp of the trajectory point. Finally, the trajectory points determined on the target stroke to be detected "- -" are " 、 、...... 、 ”......, From the position coordinates and writing timestamps of the above-mentioned respective trajectory points, it can be seen that the spatial distance between trajectory points and is 3, which is greater than the preset distance threshold (the preset distance threshold is the same as the spacing in the equal interval, for example, if the spacing in the equal interval when determining a plurality of trajectory points at equal intervals is 1, the preset distance threshold is 1). At the same time, the writing time difference between these two trajectory points is 0.32, which is greater than the preset time threshold (where the preset time threshold is set according to the actual needs such as the writing habits of normal users, such as set to 0.02), it is determined that is the end point (ending point) of the first stroke in the target stroke to be detected "- -", The starting point of the second stroke in the target stroke "- -" is used to determine if there is a breakpoint in the target stroke "- -", meaning that the pen tip may briefly leave the writing surface during writing, resulting in a discontinuous trajectory. If the distance between each trajectory point is less than a preset distance threshold, or the writing time difference between each trajectory point is less than a preset time threshold, then it is determined that there is no breakpoint between the strokes. Therefore, breakpoint detection can be performed on each stroke in a handwritten signature using the above method. This embodiment of the invention detects the existence of breakpoints within strokes by using writing trajectory information, combining spatial and temporal continuity, and has advantages such as high accuracy, strong robustness, and adaptability to dynamic writing characteristics.

[0042] 204. Based on the number of stroke intersections, perform stroke break detection on the handwritten signature to be detected to obtain the stroke end connection detection results.

[0043] In this embodiment of the invention, to ensure the comprehensiveness of signature conformity detection, it is also necessary to perform stroke disconnection detection on the handwritten signature. Based on this, step 204 specifically includes: constructing the stroke graph structure of the handwritten signature to be detected, determining the number of connected components in the stroke graph structure, and obtaining the number of standard connected components in the standard stroke graph structure of the standard signature and the number of standard intersections between each stroke in the standard signature, wherein the standard signature is a compliant signature of the real name corresponding to the handwritten signature to be detected; if the number of connected components is greater than the number of standard connected components, and / or the number of stroke intersections is less than the number of standard intersections, then it is determined that there is a stroke disconnection in the handwritten signature to be detected; otherwise, it is determined that there is no stroke disconnection in the handwritten signature to be detected.

[0044] Specifically, the handwritten signature to be detected is processed into a binary image, the binary image is thinned, a single-pixel-wide stroke structure is extracted, the starting point and ending point of each stroke in the stroke structure are identified as breakpoints, and the intersection points between strokes are identified. The starting points, ending points, and intersection points are used as nodes, and the stroke segments between necklace nodes are used as edges. Based on the nodes and edges, a stroke graph structure is constructed, the number of independent connected regions in the stroke graph structure is counted to obtain the number of connected components. At the same time, the number of connected components (standard number of connected components) and the standard number of intersection points of the standard signature corresponding to the real name of the handwritten signature to be detected are counted in the above manner. If the number of connected components of the handwritten signature to be detected is greater than the standard number of connected components, it is determined that there is a break between strokes in the handwritten signature to be detected. Or if the number of intersection points of the handwritten signature to be detected is less than the standard number of intersection points, it is determined that there is a break between strokes in the handwritten signature to be detected. Otherwise, if the number of connected components of the handwritten signature to be detected is less than or equal to the standard number of connected components, and the number of intersection points of the handwritten signature to be detected is greater than or equal to the standard number of intersection points, it is determined that there is no break between strokes in the handwritten signature to be detected. For example, the standard number of connected components of the character "王" is 1, and the standard number of intersection points is 3. If the topmost stroke "一" in "王" is disconnected from the stroke "丨", its corresponding number of connected components is 2, and the number of intersection points is 2. In the embodiment of the present invention, the break detection between strokes is performed by comprehensively analyzing the connected components and intersection points of the signature, which can improve the break detection accuracy.

[0045] 205. Determine the stroke connection status detection result of the handwritten signature to be detected based on the in-stroke breakpoint detection result and the inter-stroke connection detection result.

[0046] Specifically, a reasonable in-stroke breakpoint detection score is set for the in-stroke breakpoint detection result, and a reasonable inter-stroke connection detection score is set for the inter-stroke connection detection result. If the number of breakpoints in a stroke is越多, the corresponding in-stroke breakpoint detection score is越低. If the number of breaks between strokes is越多, the corresponding inter-stroke connection detection score is越低. Then, weight coefficients for the in-stroke breakpoint detection score and the inter-stroke connection detection score are set according to actual requirements. Based on the weight coefficients, the in-stroke breakpoint detection score and the inter-stroke connection detection score are weighted and summed to obtain the stroke connection status detection result of the handwritten signature to be detected, that is, the stroke connection status detection score.

[0047] 206. Perform stroke integrity detection on the handwritten signature to be detected to obtain the stroke integrity detection result.

[0048] It should be noted that there seem to be some incorrect or unclear expressions in the original text such as "越多" which should be corrected for a more accurate translation. I have translated it as presented while keeping those inaccuracies in the original for reference.In this embodiment of the invention, to ensure the comprehensiveness of the standardization detection, it is also necessary to perform stroke integrity detection on the handwritten signature to be detected, that is, to detect whether some strokes are missing in the handwritten signature. Based on this, step 206 specifically includes: constructing a preset signature library, wherein the method for constructing the preset signature library includes: obtaining complete signature images with complete strokes for various characters; dividing the complete signature images into multiple grids and determining the gray mean, gray standard deviation, and stroke density of each grid; concatenating the gray mean, gray standard deviation, and stroke density into a standard feature vector, and constructing the preset signature library from the complete signature images and their corresponding standard feature vectors; determining the matching complete signature image that matches the handwritten signature to be detected in the preset signature library; and determining the target signature of the handwritten signature to be detected. The feature vector is used to calculate the similarity distance between the handwritten signature to be detected and the matching complete signature image based on the feature vector to be detected and the standard feature vector corresponding to the matching complete signature image. It is then determined whether the similarity distance is greater than a preset similarity threshold. If so, the handwritten signature to be detected is considered complete; otherwise, it is considered incomplete. The method for determining the preset similarity threshold includes: for the same sample text, obtaining different styles of complete signatures and different styles of incomplete signatures; calculating the complete similarity distance between each pair of complete signatures and the incomplete similarity distance between each incomplete signature and each complete signature; and determining the preset similarity threshold based on the complete similarity distance and the incomplete similarity distance.

[0049] Specifically, a complete signature image matching the handwritten signature to be detected can be found in a pre-defined signature database using character segmentation. Simultaneously, the feature vector to be detected for the handwritten signature is determined. This determination involves: segmenting the image corresponding to the handwritten signature into multiple grids and determining the mean gray level, standard deviation gray level, and stroke density of each grid; concatenating the mean gray level, standard deviation gray level, and stroke density to form the feature vector to be detected. Alternatively, a pre-trained feature extraction model can be used to extract the feature vector to be detected. Then, based on the feature vector to be detected and the standard feature vector corresponding to the complete signature, the cosine similarity between the handwritten signature to be detected and its matched complete signature is calculated. At the same time, the evaluation criteria for the cosine similarity, i.e., a pre-defined similarity threshold, also needs to be determined. Specifically, this threshold is determined by: determining the complete stroke similarity distance between each pair of complete stroke signatures i. The incomplete stroke similarity distance between each incomplete stroke signature and each complete stroke signature j Then, the preset similarity threshold is determined according to the following formula. :

[0050] Where n is the total number of complete stroke similarity distances, and m is the total number of incomplete stroke similarity distances. Further, if the cosine similarity between the handwritten signature to be detected and its matching complete signature is greater than the preset similarity threshold t, the handwritten signature to be detected is determined to be complete; otherwise, it is determined that the handwritten signature to be detected is missing strokes. In another embodiment of the present invention, the stroke completeness score corresponding to the stroke completeness detection result of the handwritten signature to be detected can be reasonably set according to the degree of difference between the cosine similarity and the preset similarity threshold t. For example, if the cosine similarity is greater than the preset similarity threshold t, the larger the difference between the two, the higher the stroke completeness score; if the cosine similarity is less than the preset similarity threshold t, the larger the difference between the two, the lower the stroke completeness score. The embodiments of the present invention determine the preset similarity threshold by obtaining complete and incomplete signatures of different styles of the same text sample, comprehensively covering various possible states of strokes, thereby making the subsequently determined preset similarity threshold more reasonable and accurate.

[0051] 207. Based on the stroke connection status detection results and stroke integrity detection results, determine the standardization of the handwritten signature to be detected.

[0052] In this embodiment of the invention, after obtaining the stroke connection state detection result and the stroke integrity detection result, it is necessary to comprehensively analyze the stroke connection state detection result and the stroke integrity detection result to determine the standardization of the handwritten signature to be detected. Based on this, step 207 specifically includes: determining the connection state feature vector corresponding to the stroke connection state detection result and the integrity feature vector corresponding to the stroke integrity detection result; performing cross processing on the connection state feature vector and the integrity feature vector to obtain the stroke cross feature vector; and inputting the stroke cross feature vector into a preset standardization detection model for detection to obtain the standardization detection result of the handwritten signature to be detected. Specifically, the method for cross-processing the connection state feature vector and the integrity feature vector includes: performing feature-level cross-processing on the connection state feature vector and the integrity feature vector to obtain a feature cross vector; performing element-level cross-processing on the connection state feature vector and the integrity feature vector to obtain an element cross vector; performing low-order cross-processing on the connection state feature vector and the integrity feature vector to obtain a low-order cross vector; and combining the feature cross vector, the element cross vector, and the low-order cross vector to obtain the stroke cross feature vector.

[0053] Specifically, feature extraction models, such as CNN models, are used to extract the connection state feature vector corresponding to the stroke connection state detection result and the integrity feature vector corresponding to the stroke integrity detection result, respectively. Then, to fully utilize the relationships between data, extract more latent features, and simultaneously handle both high-order and low-order processing to make data utilization more efficient and the subsequent prediction results more accurate, meeting the needs of practical application scenarios, it is necessary to perform cross-processing on the connection state feature vector and the integrity feature vector. For example, if the connection state feature vector is (a1, a2) and the integrity feature vector is (b1, b2), the specific cross-processing method includes: performing feature-level cross-processing between different feature vectors, that is, performing a Hadamard product on all elements of the vectors and then weighting the result accordingly. A convolution transformation is then performed on w1 to obtain the feature cross vector f(w1×(a1×b1,a2×b2)). Simultaneously, element-wise crosses are performed on all feature vectors; that is, after performing a Hadamard product on each element of the vectors, different weight values ​​w2 and w3 are assigned to each product result, followed by a linear transformation, resulting in the element cross vector f(w2×a1×b1,w3×a2×b2). Furthermore, all feature vectors undergo low-order cross processing, and the cross-processed result is assigned a weight coefficient w4, followed by a linear transformation, resulting in the low-order cross vector f(w4(a1,a2,b1,b2)). Finally, the feature cross vectors, element cross vectors, and low-order cross vectors are combined, such as by horizontal concatenation, to obtain the stroke cross feature vector. It should be noted that the above examples are merely illustrative and do not limit the scope of this application's embodiments. Therefore, by cross-processing the connection state feature vector and integrity feature vector, different features can be automatically or explicitly combined to generate new feature combinations. These combined features may contain complex nonlinear relationships between the original features, enabling the model to capture more refined and richer information in the data. This means it can fully utilize the relationships between various data points, extract more latent features, and simultaneously handle both high-order and low-order processing, making data utilization more efficient and resulting in more accurate data trend analysis, thus meeting the needs of practical applications. Furthermore, by inputting the stroke cross-feature vector into a preset standardization detection model, the model can directly output the standardization detection results and standardization value of the handwritten signature.In this process, to improve the accuracy of standardization detection, it is necessary to pre-train and construct a preset standardization detection model. Based on this, the method includes: constructing a preset initial standardization detection model; obtaining a sample feature dataset, wherein the sample feature dataset includes sample connectivity feature vectors corresponding to sample stroke connectivity detection results and sample integrity feature vectors corresponding to sample stroke integrity detection results for multiple sample handwritten signatures with annotation information, wherein the annotation information is the actual standardization of the sample handwritten signature; dividing the sample feature dataset into a training feature set and a test feature set, using the training feature set to train the preset initial standardization detection model, and using the test feature set to test the trained preset initial standardization detection model, and finally using the trained preset initial standardization detection model that meets the test conditions as the preset standardization detection model.

[0054] Specifically, during model training, a pre-defined initial canonical detection model is first constructed. Then, a sample feature dataset is downloaded from the network. Ensure the dataset contains all necessary files. Convert the labeled files to a format understandable by the pre-defined initial canonical detection model. Finally, train and test the model. Specifically, the data feature set can be divided first: using randomness or a specific strategy (such as stratified sampling), the sample data feature set is divided into a training feature set and a test feature set. The model is then trained using the training feature set, and tested using the test feature set to evaluate its performance on unseen data. Metrics such as mCP, precision, and recall on the test feature set are calculated and recorded. If the model performance does not meet requirements, it can return to the training phase for further iterations or adjustments. This process yields a pre-defined canonical detection model that meets the requirements.

[0055] According to another handwritten signature standardization detection method provided by the present invention, compared with the current method of detecting signature standardization by calculating the similarity between the handwritten signature and the user's registered name, the present invention, after determining that the handwritten signature is the user's real registered name, further performs stroke connection state detection and stroke integrity detection on the handwritten signature, which can detect the handwritten signature from multiple aspects, ensuring the comprehensiveness of handwritten signature detection, thereby improving the detection accuracy of handwritten signature standardization.

[0056] Furthermore, as Figure 1 In specific implementation, embodiments of the present invention provide a device for detecting the standardization of handwritten signatures, such as... Figure 3 As shown, the device includes: an acquisition unit 31, a judgment unit 32, a detection unit 33, and a determination unit 34.

[0057] The acquisition unit 31 can be used to acquire the handwritten signature to be detected of the target user and the writing process data when writing the handwritten signature to be detected.

[0058] The judgment unit 32 can be used to determine whether the handwritten signature to be detected matches the real name of the target user.

[0059] The detection unit 33 can be used to perform stroke connection state detection on the handwritten signature to be detected based on the writing process data if the handwritten signature to be detected matches the real name, and to obtain the stroke connection state detection result, and to perform stroke integrity detection on the handwritten signature to be detected, and to obtain the stroke integrity detection result.

[0060] The determining unit 34 can be used to determine the standardization of the handwritten signature to be detected based on the stroke connection state detection result and the stroke integrity detection result.

[0061] In specific application scenarios, writing process data includes writing trajectory information and the number of stroke intersections; for example, to detect the stroke connection state of the handwritten signature to be detected, Figure 4 As shown, the detection unit 33 includes a breakpoint detection module 331, a chain break detection module 332, and a first determination module 333.

[0062] The breakpoint detection module 331 can be used to detect breakpoints within the strokes of the handwritten signature to be detected based on the writing trajectory information, and obtain the breakpoint detection result within the strokes.

[0063] The broken link detection module 332 can be used to detect the broken links between strokes of the handwritten signature to be detected based on the number of stroke intersections, and obtain the end-connection detection result between strokes.

[0064] The first determining module 333 can be used to determine the stroke connection status detection result of the handwritten signature to be detected based on the stroke breakpoint detection result and the stroke connection detection result.

[0065] In specific application scenarios, in order to detect breakpoints within strokes of a handwritten signature to be detected, the breakpoint detection module 331 can be used to identify each stroke to be detected in the handwritten signature to be detected, taking any one of the strokes to be detected as a target stroke to be detected, and determining multiple trajectory points at equal intervals on the writing trajectory corresponding to the target stroke to be detected; based on the writing trajectory information, determining the position coordinates and writing timestamp of each trajectory point, and determining the spatial distance and time difference between each pair of adjacent trajectory points; if the spatial distance of any trajectory point is greater than a preset distance threshold, and the corresponding time difference is greater than a preset time threshold, then it is determined that there is a breakpoint in the target stroke to be detected; otherwise, it is determined that there is no breakpoint in the target stroke to be detected.

[0066] In specific application scenarios, in order to detect the breaks in the strokes of the handwritten signature to be detected, the broken stroke detection module 332 can be used to construct the stroke graph structure of the handwritten signature to be detected, determine the number of connected components in the stroke graph structure, and obtain the number of standard connected components in the standard stroke graph structure of the standard signature and the number of standard intersections between each stroke in the standard signature, wherein the standard signature is a compliance signature of the real name corresponding to the handwritten signature to be detected; if the number of connected components is greater than the number of standard connected components, and / or the number of stroke intersections is less than the number of standard intersections, then it is determined that there are breaks in the strokes of the handwritten signature to be detected; otherwise, it is determined that there are no breaks in the strokes of the handwritten signature to be detected.

[0067] In specific application scenarios, in order to perform stroke integrity detection on the handwritten signature to be detected, the detection unit 33 also includes a construction module 334 and a calculation module 335.

[0068] The construction module 334 can be used to construct a preset signature library. The method for constructing the preset signature library includes: obtaining complete signature images of multiple characters with complete strokes; dividing the complete signature images into multiple grids and determining the gray mean, gray standard deviation and stroke density of each grid; concatenating the gray mean, gray standard deviation and stroke density into a standard feature vector, and constructing the preset signature library from the complete signature images and their corresponding standard feature vectors.

[0069] The first determining module 333 can be used to determine a matching complete signature image that matches the handwritten signature to be detected in the preset signature library.

[0070] The calculation module 335 can be used to determine the feature vector to be detected of the handwritten signature to be detected, calculate the similarity distance between the handwritten signature to be detected and the matching complete signature image based on the feature vector to be detected and the standard feature vector corresponding to the matching complete signature image, and determine whether the similarity distance is greater than a preset similarity threshold. If it is, the handwritten signature to be detected is determined to have complete strokes; otherwise, the handwritten signature to be detected is determined to have incomplete strokes. The method for determining the preset similarity threshold includes: for the same sample text, obtaining complete signatures and incomplete signatures of different styles; calculating the complete similarity distance between each pair of complete signatures, and calculating the incomplete similarity distance between each incomplete signature and each complete signature; and determining the preset similarity threshold based on the complete similarity distance and the incomplete similarity distance.

[0071] In specific application scenarios, in order to perform stroke integrity detection on the handwritten signature to be detected, the detection unit 33 also includes a detection module 336.

[0072] The detection module 336 can be used to input the handwritten signature to be detected into a preset integrity detection model for detection, and obtain the stroke integrity detection result of the handwritten signature to be detected. The preset integrity detection model is pre-built based on a sample dataset, which includes sample handwritten signatures with annotation information, and the annotation information is the actual stroke integrity of the sample handwritten signature.

[0073] In specific application scenarios, in order to determine the standardization of the handwritten signature to be detected, the determining unit 34 includes a second determining module 341 and a summing module 342.

[0074] The second determining module 341 can be used to determine the stroke connection state score and its corresponding connection state weight coefficient corresponding to the stroke connection state detection result, and to determine the stroke integrity score and its corresponding integrity weight coefficient corresponding to the stroke integrity detection result.

[0075] The summation module 342 can be used to perform a weighted summation of the stroke connection state score and the stroke integrity score based on the connection state weight coefficient and the integrity weight coefficient to obtain a comprehensive score, and determine the standardization of the handwritten signature to be detected based on the comprehensive score.

[0076] In specific application scenarios, in order to determine whether the handwritten signature to be detected matches the real name of the target user, the judgment unit 32 includes a character segmentation module 321, a matching module 322, and a judgment module 323.

[0077] The character segmentation module 321 can be used to segment the handwritten signature to be detected to obtain multiple segmented characters to be detected, and to segment the real name of the target user to obtain multiple real segmented characters.

[0078] The matching module 322 can be used to perform similarity matching between each of the characters to be detected and the real characters at the corresponding character positions to obtain a similarity matching score for each character position.

[0079] The judgment module 323 can be used to determine the position weight of each character position, and based on the position weight, to perform a weighted summation of each similarity matching score to obtain a comprehensive score, and based on the comprehensive score, to determine whether the handwritten signature to be detected matches the real name of the target user.

[0080] It should be noted that other corresponding descriptions of the functional modules involved in the handwritten signature standardization detection device provided in this embodiment of the invention can be found in [reference]. Figure 1 The corresponding description of the method shown will not be repeated here.

[0081] Based on the above, Figure 1 Accordingly, this embodiment of the invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, performs the following steps: acquiring a handwritten signature to be detected from a target user and writing process data during the writing of the handwritten signature; determining whether the handwritten signature to be detected matches the real name of the target user; if the handwritten signature to be detected matches the real name, performing stroke connection state detection on the handwritten signature to be detected based on the writing process data to obtain a stroke connection state detection result, and performing stroke integrity detection on the handwritten signature to be detected to obtain a stroke integrity detection result; and determining the standardization of the handwritten signature to be detected based on the stroke connection state detection result and the stroke integrity detection result.

[0082] Based on the above, Figure 1 The method shown and as Figure 3 The embodiment of the device shown in the invention also provides a physical structure diagram of a computer device, such as... Figure 5 As shown, the computer device includes: a processor 41, a memory 42, and a computer program stored in the memory 42 and executable on the processor. Both the memory 42 and the processor 41 are mounted on a bus 43. When the processor 41 executes the program, it performs the following steps: acquiring a handwritten signature to be detected from a target user and writing process data during the writing of the handwritten signature; determining whether the handwritten signature to be detected matches the real name of the target user; if the handwritten signature to be detected matches the real name, performing stroke connection state detection on the handwritten signature to be detected based on the writing process data to obtain a stroke connection state detection result, and performing stroke integrity detection on the handwritten signature to be detected to obtain a stroke integrity detection result; and determining the standardization of the handwritten signature to be detected based on the stroke connection state detection result and the stroke integrity detection result.

[0083] Through the technical solution of this invention, this invention determines whether the handwritten signature to be detected is the user's registered real name. After determining that the handwritten signature is the user's registered real name, it further performs stroke connection status detection and stroke integrity detection on the handwritten signature. This allows for detection of the handwritten signature from multiple aspects, ensuring the comprehensiveness of the handwritten signature detection and thus improving the accuracy of handwritten signature standardization detection.

[0084] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those presented herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0085] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for detecting the standardization of handwritten signatures, characterized in that, include: Obtain the handwritten signature to be detected from the target user and the writing process data when writing the handwritten signature to be detected; Determine whether the handwritten signature to be detected matches the real name of the target user; If the handwritten signature to be detected matches the real name, then the handwritten signature to be detected is subjected to stroke connection state detection based on the writing process data to obtain stroke connection state detection result, and the handwritten signature to be detected is subjected to stroke integrity detection to obtain stroke integrity detection result. Based on the stroke connection state detection results and the stroke integrity detection results, the standardization of the handwritten signature to be detected is determined.

2. The method according to claim 1, characterized in that, The writing process data includes writing trajectory information and the number of stroke intersections; Based on the writing process data, the stroke connection state of the handwritten signature to be detected is detected to obtain the stroke connection state detection result, including: Based on the writing trajectory information, breakpoint detection within the strokes is performed on the handwritten signature to be detected to obtain the breakpoint detection result within the strokes. Based on the number of stroke intersections, the handwritten signature to be detected is subjected to stroke disconnection detection to obtain the stroke end-connection detection result. Based on the intra-stroke breakpoint detection results and the inter-stroke connection detection results, the stroke connection status detection results of the handwritten signature to be detected are determined.

3. The method according to claim 2, characterized in that, Based on the writing trajectory information, breakpoint detection is performed within the strokes of the handwritten signature to be detected, resulting in breakpoint detection results, including: In the handwritten signature to be detected, each stroke to be detected is identified, and any stroke to be detected in each stroke to be detected is taken as a target stroke to be detected. Multiple trajectory points are determined at equal intervals on the writing trajectory corresponding to the target stroke to be detected. Based on the writing trajectory information, the position coordinates and writing timestamp of each trajectory point are determined, and the spatial distance and time difference between each pair of adjacent trajectory points are determined. If the spatial distance of any trajectory point is greater than a preset distance threshold and the corresponding time difference is greater than a preset time threshold, then it is determined that there is a breakpoint in the target stroke to be detected; otherwise, it is determined that there is no breakpoint in the target stroke to be detected. Based on the number of stroke intersections, the handwritten signature to be detected is subjected to stroke disconnection detection to obtain stroke end-connection detection results, including: Construct the stroke graph structure of the handwritten signature to be detected, determine the number of connected components in the stroke graph structure, and obtain the number of standard connected components in the standard stroke graph structure of the standard signature and the number of standard intersections between each stroke in the standard signature, wherein the standard signature is a compliance signature of the real name corresponding to the handwritten signature to be detected. If the number of connected components is greater than the number of standard connected components, and / or the number of stroke intersections is less than the number of standard intersections, then it is determined that there are breaks between strokes in the handwritten signature to be detected; otherwise, it is determined that there are no breaks between strokes in the handwritten signature to be detected.

4. The method according to claim 1, characterized in that, Perform stroke integrity detection on the handwritten signature to be detected, and obtain the stroke integrity detection result, including: Constructing a preset signature library, wherein the method for constructing the preset signature library includes: obtaining complete signature images of various characters and their corresponding strokes; dividing the complete signature images into multiple grids and determining the gray mean, gray standard deviation, and stroke density of each grid; concatenating the gray mean, gray standard deviation, and stroke density into a standard feature vector, and constructing the preset signature library from the complete signature images and their corresponding standard feature vectors; Determine a matching complete signature image from the preset signature library that matches the handwritten signature to be detected; The method involves determining the feature vector to be detected for the handwritten signature to be detected, calculating the similarity distance between the handwritten signature to be detected and the matching complete signature image based on the feature vector to be detected and the standard feature vector corresponding to the matching complete signature image, and determining whether the similarity distance is greater than a preset similarity threshold. If it is, the handwritten signature to be detected is determined to have complete strokes; otherwise, the handwritten signature to be detected is determined to have incomplete strokes. The method for determining the preset similarity threshold includes: obtaining different styles of complete stroke signatures and different styles of incomplete stroke signatures for the same sample text; calculating the complete stroke similarity distance between each pair of complete stroke signatures and calculating the incomplete stroke similarity distance between each incomplete stroke signature and each complete stroke signature; and determining the preset similarity threshold based on the complete stroke similarity distance and the incomplete stroke similarity distance.

5. The method according to claim 1, characterized in that, Perform stroke integrity detection on the handwritten signature to be detected, and obtain the stroke integrity detection result, including: The handwritten signature to be detected is input into a preset integrity detection model for detection to obtain the stroke integrity detection result of the handwritten signature to be detected. The preset integrity detection model is pre-built based on a sample dataset, which includes sample handwritten signatures with annotation information, and the annotation information is the actual stroke integrity of the sample handwritten signature.

6. The method according to claim 1, characterized in that, Based on the stroke connection state detection results and the stroke integrity detection results, the standardization of the handwritten signature to be detected is determined, including: Determine the stroke connection state score and its corresponding connection state weight coefficient corresponding to the stroke connection state detection result, and determine the stroke integrity score and its corresponding integrity weight coefficient corresponding to the stroke integrity detection result. Based on the connection state weight coefficient and the integrity weight coefficient, the stroke connection state score and the stroke integrity score are weighted and summed to obtain a comprehensive score, and the standardization of the handwritten signature to be detected is determined based on the comprehensive score.

7. The method according to claim 1, characterized in that, Determining whether the handwritten signature to be detected matches the real name of the target user includes: The handwritten signature to be detected is segmented into multiple segments to be detected, and the real name of the target user is segmented into multiple real segments. Each of the characters to be detected and the corresponding real characters at their respective positions are matched for similarity to obtain a similarity score for each character position. The position weight of each character position is determined. Based on the position weight, the similarity matching scores of each character are weighted and summed to obtain a comprehensive score. Based on the comprehensive score, it is determined whether the handwritten signature to be detected matches the real name of the target user.

8. A device for detecting the standardization of handwritten signatures, characterized in that, include: The acquisition unit is used to acquire the handwritten signature to be detected of the target user and the writing process data when writing the handwritten signature to be detected; A judgment unit is used to determine whether the handwritten signature to be detected matches the real name of the target user; The detection unit is configured to, if the handwritten signature to be detected matches the real name, perform stroke connection state detection on the handwritten signature to be detected based on the writing process data to obtain a stroke connection state detection result, and perform stroke integrity detection on the handwritten signature to be detected to obtain a stroke integrity detection result. The determining unit is used to determine the standardization of the handwritten signature to be detected based on the stroke connection state detection result and the stroke integrity detection result.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images