CN113837287B - Certificate abnormal information identification method, device, equipment and medium - Google Patents
Certificate abnormal information identification method, device, equipment and medium Download PDFInfo
- Publication number
- CN113837287B CN113837287B CN202111130685.8A CN202111130685A CN113837287B CN 113837287 B CN113837287 B CN 113837287B CN 202111130685 A CN202111130685 A CN 202111130685A CN 113837287 B CN113837287 B CN 113837287B
- Authority
- CN
- China
- Prior art keywords
- feature vector
- vector
- feature
- foreground
- characteristic
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Processing (AREA)
Abstract
The invention relates to an artificial intelligence technology and discloses a certificate abnormal information identification method which is used for improving the accuracy of certificate picture abnormal information identification. The method provided by the invention comprises the following steps: acquiring a certificate picture, carrying out image recognition on the certificate picture to obtain a foreground region and a background region corresponding to the certificate picture, carrying out feature calculation on the foreground region and the background region respectively based on a preset feature processing mode and position coding information to obtain a foreground feature vector and a background feature vector, carrying out feature value calculation on the foreground feature vector and the background feature vector to obtain a foreground feature value and a background feature value, and calculating the difference between the foreground feature value and the background feature value to obtain a feature difference value; comparing the characteristic difference value with a preset threshold value to obtain a comparison result, if the characteristic difference value is not larger than the preset threshold value, confirming that the certificate picture has no abnormal information, and if the characteristic difference value is larger than the preset threshold value, analyzing based on a morphological characteristic algorithm to determine the abnormal type.
Description
Technical Field
The present invention relates to the field of image processing, and in particular, to a method and apparatus for identifying abnormal information of a certificate, a computer device, and a storage medium.
Background
When information identification is performed on a certificate, abnormal information such as scratches or light lines of the certificate generally have a certain influence on an information detection result, and errors such as inconsistent information identification are usually caused, so that economic influence can be caused in some identification scenes.
The existing mode is mainly based on the convolutional neural network to detect abnormal information such as scratches or light lines, but because some characteristics of the convolutional neural network, after the final accuracy of the classification network reaches a certain standard, the abnormal information learned by the classification network is easy to bring larger errors, so that the accuracy is lowered, and the requirement on a data set is higher.
In order to meet the high requirement of the data set, a semantic segmentation method is provided for carrying out segmentation processing on abnormal information such as scratches on certificates, but the method brings about some problems in implementation. The improvement of the precision of the existing popular semantic segmentation method is mainly based on the coding mode of downsampling coding, but the problem is that the pooling operation in the downsampling process can enlarge the receptive field and integrate the context information, so that the classification accuracy is improved. But at the same time, the pooling operation weakens the position information, so that when the abnormal information of the identification certificate is identified, only local features can be noticed, and the global is ignored, thereby leading to lower accuracy of the abnormal information identification.
Therefore, the conventional method has the problem that the accuracy of identification is low when the certificate picture is abnormally identified.
Disclosure of Invention
The embodiment of the invention provides a certificate abnormal information identification method, a device, computer equipment and a storage medium, so as to improve the accuracy of certificate picture abnormal information identification.
A method for identifying abnormal information of a certificate, comprising:
acquiring a certificate picture, and extracting position coding information from the certificate picture;
performing image recognition on the certificate picture to obtain a foreground region and a background region corresponding to the certificate picture, wherein the foreground region is a character region of the certificate picture, and the background region is a residual region except the foreground region in the certificate picture;
based on a preset feature processing mode and the position coding information, respectively carrying out feature calculation on the foreground region and the background region to obtain a foreground feature vector and a background feature vector;
based on a preset characteristic value calculation mode, carrying out characteristic value calculation on the foreground characteristic vector to obtain a corresponding foreground characteristic value, carrying out characteristic value calculation on the background characteristic vector to obtain a corresponding background characteristic value, and calculating a difference value between the foreground characteristic value and the background characteristic value to obtain a characteristic difference value;
Comparing the characteristic difference value with a preset threshold value to obtain a comparison result;
and if the comparison result is that the characteristic difference value is not larger than the preset threshold value, confirming that the certificate picture has no abnormal information, and if the comparison result is that the characteristic difference value is larger than the preset threshold value, analyzing the foreground characteristic vector and the background characteristic vector based on a morphological characteristic algorithm, and determining an abnormal type.
A certificate anomaly information recognition device comprising:
the picture acquisition module is used for acquiring a certificate picture and extracting position coding information from the certificate picture;
the image recognition module is used for carrying out image recognition on the certificate picture to obtain a foreground region and a background region corresponding to the certificate picture, wherein the foreground region is a character region of the certificate picture, and the background region is a residual region except the foreground region in the certificate picture;
the feature processing module is used for respectively carrying out feature calculation on the foreground region and the background region based on a preset feature processing mode and the position coding information to obtain a foreground feature vector and a background feature vector;
the characteristic value acquisition module is used for carrying out characteristic value calculation on the foreground characteristic vector based on a preset characteristic value calculation mode to obtain a corresponding foreground characteristic value, carrying out characteristic value calculation on the background characteristic vector to obtain a corresponding background characteristic value, and calculating a difference value between the foreground characteristic value and the background characteristic value to obtain a characteristic difference value;
The comparison result acquisition module is used for comparing the characteristic difference value with a preset threshold value to obtain a comparison result;
and the comparison module is used for confirming that the certificate picture does not have abnormal information if the comparison result is that the characteristic difference value is not greater than the preset threshold value, and analyzing the foreground characteristic vector and the background characteristic vector based on a morphological characteristic algorithm to determine the abnormal type if the comparison result is that the characteristic difference value is greater than the preset threshold value.
A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of the credential anomaly information identification method described above when the computer program is executed.
A computer readable storage medium storing a computer program which, when executed by a processor, implements the steps of the credential anomaly information identification method described above.
According to the certificate abnormal information identification method, device, computer equipment and storage medium, a certificate picture is obtained, position coding information is extracted from the certificate picture, image identification is carried out on the certificate picture, a foreground area and a background area corresponding to the certificate picture are obtained, characteristic calculation is carried out on the foreground area and the background area respectively based on a preset characteristic processing mode and the position coding information, a foreground characteristic vector and a background characteristic vector are obtained, characteristic value calculation is carried out on the foreground characteristic vector based on a preset characteristic value calculation mode, a corresponding foreground characteristic value is obtained, characteristic value calculation is carried out on the background characteristic vector, a corresponding background characteristic value is obtained, a difference value is obtained, the characteristic difference value is compared with a preset threshold value, a comparison result is obtained, if the characteristic difference value is not larger than the preset threshold value, the foreground characteristic vector and the background characteristic vector are analyzed based on a morphological characteristic algorithm, an abnormal type is determined, the foreground area and the background area of the certificate picture are identified, the position of the foreground characteristic image and the background area are also processed according to the preset characteristic algorithm, and the local characteristic value is compared with the background characteristic value, and the attention rate is improved, and the image is processed according to the characteristic value is processed in a global mode.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the description of the embodiments of the present invention will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic view of an application environment of a method for identifying abnormal information of a document according to an embodiment of the present invention;
FIG. 2 is a flow chart of a method for identifying abnormal information of a document according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a certificate abnormal information identifying apparatus according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of a computer device in accordance with an embodiment of the invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The certificate abnormal information identification method provided by the application can be applied to an application environment as shown in fig. 1, wherein computer equipment communicates with a server through a network. The computer device may be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices, among others. The server may be implemented as a stand-alone server or as a server cluster composed of a plurality of servers.
The server may be an independent server, or may be a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (Content Delivery Network, CDN), and basic cloud computing services such as big data and artificial intelligence platforms.
In one embodiment, as shown in fig. 2, a method for identifying abnormal information of a certificate is provided, and the method is applied to the server in fig. 1, and is illustrated as an example, and includes the following steps S101 to S106:
s101, acquiring a certificate picture, and extracting position coding information from the certificate picture.
In step S101, the types of the certificates include, but are not limited to, identity cards, wedding cards, and professional-grade certificates.
The method for acquiring the certificate picture includes, but is not limited to, camera shooting, website interception and the like.
The position coding information refers to position information corresponding to the pixel points on the certificate picture, for example, the position coding information of the first row and the first column of the pixel points of the certificate picture is (1, 1).
The certificate picture is obtained, and the position coding information is extracted from the certificate picture, so that when the abnormal information of the certificate picture is identified later, the position information of all pixel points on the certificate picture can be reserved, and the abnormal information of the certificate picture can be positioned during subsequent analysis.
S102, carrying out image recognition on the certificate picture to obtain a foreground region and a background region corresponding to the certificate picture, wherein the foreground region is a character region of the certificate picture, and the background region is the rest region except the foreground region in the certificate picture.
In step S102, the above image recognition refers to a technique of processing, analyzing and understanding the image by a computer to recognize targets and objects of various modes, and the implementation method of the image recognition includes, but is not limited to, a mode recognition algorithm and a support vector machine algorithm.
The foreground region refers to a character region of the certificate picture, and the character region comprises a name and name content corresponding to the certificate picture, a head portrait corresponding to the certificate picture, an address and address information content corresponding to the certificate picture and the like.
The background area refers to all areas except the character area of the certificate picture.
The identification area of the certificate picture can be divided into the foreground area and the background area by carrying out image identification on the certificate picture, so that the two areas can be respectively subjected to characteristic processing, the identification of the abnormal information of the certificate picture is focused on local characteristics and global characteristics, and the accuracy of the identification of the abnormal information of the certificate picture is further effectively improved.
And S103, respectively carrying out feature calculation on the foreground region and the background region based on a preset feature processing mode and position coding information to obtain a foreground feature vector and a background feature vector.
In step S103, the above-described feature calculation method includes, but is not limited to, a scale-invariant feature transform algorithm, a DB algorithm (Differentiable Binarization, differentiable binarization processing algorithm).
The preset feature processing mode refers to a mode of processing the foreground region and the background region according to a preset feature processing method.
Preferably, the foreground region and the background region are processed by adopting a semantic segmentation method.
The foreground region, the background region and the position coding information are subjected to feature processing to obtain corresponding foreground feature vectors and background feature vectors so as to process the feature vectors later, so that the identification of the abnormal information of the certificate picture focuses on local features and global features, and the accuracy of the identification of the abnormal information of the certificate picture is improved effectively.
S104, based on a preset characteristic value calculation mode, carrying out characteristic value calculation on the foreground characteristic vector to obtain a corresponding foreground characteristic value, carrying out characteristic value calculation on the background characteristic vector to obtain a corresponding background characteristic value, and calculating a difference value between the foreground characteristic value and the background characteristic value to obtain a characteristic difference value.
In step S104, the above-mentioned preset feature value calculation method is a method for calculating feature values of the foreground feature vector and the background feature vector according to a preset feature value calculation method.
Preferably, the step of the preset feature value calculation mode is to calculate a feature polynomial of the corresponding feature vector, and calculate all the roots of the feature equation, that is, all the feature values, and optionally one of the feature values is used as the corresponding feature value.
The foreground characteristic value and the background characteristic value are calculated, the characteristic states of the foreground region and the background region can be obtained, the difference value of the foreground region and the background region can be quantized according to the two characteristic values, if the difference between the two learned characteristic values is large, the existence of abnormal information of the certificate picture is indicated, so that the abnormal information is identified, and the abnormal information is effectively identified according to the identification basis.
S105, comparing the characteristic difference value with a preset threshold value to obtain a comparison result.
In step S105, the preset threshold is a threshold of a gap between the background feature value and the foreground feature value obtained after training the plurality of normal document pictures.
Through the preset threshold value, whether the certificate picture acquired in the step S101 has abnormal information or not can be rapidly determined, and corresponding processing is performed aiming at whether the certificate picture has abnormal information or not, so that the accuracy of identifying the abnormal information of the certificate picture is improved.
And S106, if the comparison result is that the feature difference value is not greater than a preset threshold value, confirming that the certificate picture has no abnormal information, and if the comparison result is that the feature difference value is greater than the preset threshold value, analyzing the foreground feature vector and the background feature vector based on a morphological feature algorithm, and determining the abnormal type.
In step S106, the morphological feature algorithm refers to a series of image processing algorithms for processing image shape features. Such morphological feature algorithms include, but are not limited to, binary morphological algorithms, gray scale mathematical morphological algorithms.
Such anomaly types include, but are not limited to, scratches and striae.
Preferably, when the contrast result is that the feature difference value is larger than a preset threshold value, a gray level mathematical morphology algorithm is adopted to reconstruct the foreground feature vector and the background feature vector, the image of the abnormal information is deepened and amplified, when the result identified by the gray level mathematical morphology algorithm is a linear result, the abnormal type is determined to be a scratch, and when the result identified by the gray level mathematical morphology algorithm is a nonlinear result, the abnormal type is determined to be a light line.
And judging whether the certificate picture has abnormal information or not through a comparison result, and determining the abnormal type when the abnormal information exists, so that the accuracy rate of identifying the abnormal information of the certificate picture is improved.
According to the identification method for the abnormal information of the certificate, the certificate picture is obtained, the position coding information is extracted from the certificate picture, the image identification is carried out on the certificate picture, the foreground area and the background area corresponding to the certificate picture are obtained, the foreground area and the background area are respectively subjected to characteristic calculation based on the preset characteristic processing mode and the position coding information, the foreground characteristic vector and the background characteristic vector are obtained, the foreground characteristic vector is subjected to characteristic value calculation based on the preset characteristic value calculation mode, the corresponding foreground characteristic value is obtained, the background characteristic vector is subjected to characteristic value calculation, the corresponding background characteristic value is obtained, the difference value of the foreground characteristic value and the background characteristic value is calculated, the characteristic difference value is obtained, the characteristic difference value is compared with the preset threshold value, the comparison result is obtained, if the characteristic difference value is not larger than the preset threshold value, the abnormal information does not exist in the certificate picture, the abnormal type is determined based on the morphological characteristic algorithm, the foreground characteristic vector and the background characteristic vector are identified, the foreground area and the background area of the certificate picture are identified, the foreground characteristic value and the background area are processed based on the position coding information, the local characteristic value is also accurately compared with the local characteristic value, and the attention rate of the certificate is improved.
In some optional implementations of the present embodiment, step S102 includes the following steps S201 to S203:
s201, recognizing the character positions of the evidence pictures to obtain recognition results.
S202, generating at least one text box according to the recognition result.
S203, taking the areas corresponding to all the text boxes as foreground areas corresponding to the certificate pictures, and taking all the areas except the foreground areas of the certificate pictures as background areas.
For the above step S201, the above recognition result includes, but is not limited to, character position, face head portrait.
Methods of identifying character positions in a document picture include, but are not limited to, CTPN text detection algorithms (Connectionist Text Proposal Network), morphological operations, maximum stable extremum area text detection algorithms.
For the step S202, the shape of the text box includes a rectangle or an irregular pattern, where the irregular pattern is a pattern formed around the position of the attached character. Based on the above-described text box generation method according to the recognition result, the method includes, but is not limited to, a DB algorithm (Differentiable Binarization, differentiable binarization processing algorithm), CTPN text detection algorithm (Connectionist Text Proposal Network).
Through the steps, the character positions in the certificate picture are identified, and the character positions in the certificate picture are detected so as to facilitate the construction of a text box at the back, so that the region division is carried out on the certificate picture, and the accuracy of the region division is improved.
In some optional implementations of the present embodiment, step S103 includes the following steps S301 to S305:
s301, extracting features of pixel points of a foreground region to obtain a first feature vector and position coding information corresponding to the first feature vector.
S302, extracting features of pixel points of the background area to obtain a second feature vector and position coding information corresponding to the second feature vector.
S303, carrying out downsampling processing on the first feature vector and the position coding information corresponding to the first feature vector, and carrying out downsampling processing on the second feature vector and the position coding information corresponding to the second feature vector to obtain a first downsampled feature vector and a second downsampled feature vector, wherein the first downsampled feature vector and the second downsampled feature vector are high-dimensional feature vectors containing the position information.
S304, respectively carrying out coding compression processing on the first downsampled feature vector and the second downsampled feature vector to obtain a first recoded feature vector and a second recoded feature vector.
S305, respectively carrying out up-sampling operation on the first recoding feature vector and the second recoding feature vector to obtain a foreground feature vector and a background feature vector.
For the step S301, the first feature vector refers to a feature vector corresponding to the foreground region.
For the step S302, the second feature vector refers to a feature vector corresponding to the background area.
The downsampling process refers to a process of reducing the feature of the image in step S303. The downsampling algorithms described above include, but are not limited to, a swin-transformer based algorithm, a transformer based algorithm.
Preferably, the downsampling is performed using a swin-transformer based algorithm.
Specifically, step S303 further includes steps S3031 to S3033:
s3031, multiplying the first eigenvector of the same convolution layer by the position coding information corresponding to the first eigenvector to obtain the first coding eigenvector.
S3032, multiplying the second feature vector of the same convolution layer with the position coding information corresponding to the second feature vector to obtain a second coding feature vector.
S3033, based on the maximum pooling layer, the first coding feature vector and the second coding feature vector are respectively subjected to downsampling processing to obtain a first downsampled feature vector and a second downsampled feature vector.
In step S3033, the maximum pooling layer refers to a pooling layer for reducing the size of the encoded feature vector.
And the calculation speed of the first coding feature vector and the second coding feature vector is improved based on the maximum pooling layer, and meanwhile, the robustness of the extracted features is improved.
For the above step S304, specifically, step S304 further includes steps S3041 to S3044:
s3041, based on a preset cutting mode, cutting the first downsampling feature vector and the second downsampling feature vector respectively to obtain a first feature vector block set corresponding to the first downsampling feature vector and a second feature vector block set corresponding to the second downsampling feature vector, wherein the first feature vector block set comprises at least two feature vector blocks, and the second feature vector block set comprises at least two feature vector blocks.
S3042, selecting one feature vector block from the first feature vector block set at will as a first feature vector block, and performing coding processing on the first feature vector block to obtain a corresponding first coding vector block.
S3043, selecting one characteristic vector block from the second characteristic vector block set at will as a second characteristic vector block, and carrying out coding processing on the second characteristic vector block to obtain a corresponding second coding vector block.
S3044, respectively fusing and compressing all the first coding vector blocks and all the second coding vector blocks to obtain corresponding first recoding feature vectors and second recoding feature vectors.
In step S3041, the predetermined cutting method refers to a method for cutting the feature vector. The preset cutting modes include, but are not limited to, a two-part cutting method and a three-part cutting method.
In step S3042, each first feature vector block in the first feature vector block set is subjected to encoding processing. Preferably, the encoding process is an encoding process based on a swin-transformer algorithm.
In step S3043, each of the second feature vector blocks in the second feature vector block set is subjected to encoding processing. Preferably, the encoding process is an encoding process based on a swin-transformer algorithm.
In step S3044, the above-described fusion and compression process is preferably a fusion and compression process based on a swin-transformer algorithm. The feature vectors are cut in a preset cutting mode, and each cut feature vector block is subjected to coding operation, so that the high-dimensional feature vectors can be further processed, the processing precision of the feature vectors is improved, and the accuracy of identifying abnormal information is further improved.
Specifically, step S3044 includes the following steps a to c:
a. and respectively carrying out point multiplication calculation on all the first coding vector blocks and all the second coding vector blocks, and dividing the result after the point multiplication calculation by a preset constant value to obtain a first coding vector matrix and a second coding vector matrix.
b. And respectively performing attention computation on the first coding vector matrix, the second coding vector matrix and the query feature matrix to obtain a first attention feature matrix and a second attention feature matrix, wherein the query feature matrix is a preset feature matrix.
c. And carrying out normalization processing on the first attention feature matrix and the first coding vector matrix to obtain a corresponding first recoding feature vector, and carrying out normalization processing on the second attention feature matrix and the second coding vector matrix to obtain a corresponding second recoding feature vector.
In step a, the dot product calculation is performed on all the first encoding vector blocks and all the second encoding vector blocks, and the result obtained after the dot product calculation is divided by a preset constant value, so that errors caused by overlarge dot product results can be prevented.
In step b, the above-mentioned attention calculation refers to a process of acquiring a target matrix requiring important attention by scanning global features.
In step c, the normalization process includes averaging and variance of the first attention feature matrix and the first encoding vector matrix, and the second attention feature matrix and the second encoding vector matrix, where the normalization process may unify the first recoding feature vector and the second recoding feature vector and prevent the overfitting phenomenon.
Through the dot multiplication processing, the attention calculation and the normalization processing, the fitting phenomenon is prevented from occurring in the encoding process, so that the accuracy of the obtained first recoding feature vector and the second recoding feature vector is improved, the subsequent calculated amount is effectively reduced, and the complexity of identifying abnormal information of the evidence picture is reduced.
For the above step S305, the above up-sampling refers to an operation procedure of restoring the resolution of the feature vector to the resolution of the original picture. The upsampling algorithm includes, but is not limited to, a swin-transformer based algorithm, a transformer based algorithm.
Preferably, the upsampling is performed using a swin-transformer based algorithm.
In step S305, specifically, step S305 further includes steps S3051 to S3052:
s3051, respectively inputting the first recoding feature vector and the second recoding feature vector into convolution layers with corresponding scales, and performing cross-layer connection to obtain the corresponding first cross-layer feature vector and second cross-layer feature vector.
S3052, based on a deconvolution algorithm, up-sampling operation is respectively carried out on the first cross-layer feature vector and the second cross-layer feature vector, and a foreground feature vector corresponding to the first cross-layer feature vector and a background feature vector corresponding to the second cross-layer feature vector are obtained.
For the step S3051, the cross-layer connection refers to a process of connecting vectors corresponding to discontinuous convolution layers. For example, when the convolution layer a, the convolution layer B and the convolution layer C exist, the convolution layer a is directly connected to the convolution layer B, the convolution layer B is connected to the convolution layer C, and the cross-layer connection is a process of connecting the feature vector in the convolution layer a with the feature vector in the convolution layer C.
For the above step S3052, the above deconvolution algorithm refers to the inverse process of the convolution algorithm.
Through the steps, more picture information can be reserved for the extracted foreground feature vector and background feature vector, so that the accuracy of identifying the abnormal information of the certificate picture can be improved when the foreground feature vector and the background feature vector are analyzed later.
It should be understood that the sequence number of each step in the foregoing embodiment does not mean that the execution sequence of each process should be determined by the function and the internal logic, and should not limit the implementation process of the embodiment of the present invention.
In one embodiment, a certificate abnormal information identification device is provided, and the certificate abnormal information identification device corresponds to the certificate abnormal information identification method in the embodiment one by one. As shown in fig. 3, the certificate abnormal information identifying apparatus includes a picture obtaining module 11, an image identifying module 12, a feature processing module 13, a feature value obtaining module 14, a comparison result obtaining module 15, and a comparison module 16. The functional modules are described in detail as follows:
the picture acquisition module 11 is used for acquiring a certificate picture and extracting position coding information from the certificate picture.
The image recognition module 12 is configured to perform image recognition on the document image to obtain a foreground area and a background area corresponding to the document image, where the foreground area is a character area of the document image, and the background area is a remaining area of the document image except the foreground area.
The feature processing module 13 is configured to perform feature computation on the foreground region and the background region based on a preset feature processing manner and position coding information, so as to obtain a foreground feature vector and a background feature vector.
The feature value obtaining module 14 is configured to perform feature value calculation on the foreground feature vector based on a preset feature value calculation mode to obtain a corresponding foreground feature value, perform feature value calculation on the background feature vector to obtain a corresponding background feature value, and calculate a difference value between the foreground feature value and the background feature value to obtain a feature difference value.
And the comparison result obtaining module 15 is used for comparing the characteristic difference value with a preset threshold value to obtain a comparison result.
And the comparison module 16 is configured to confirm that no abnormal information exists in the certificate picture if the comparison result is that the feature difference is not greater than the preset threshold, and analyze the foreground feature vector and the background feature vector based on the morphological feature algorithm to determine the abnormal type if the comparison result is that the feature difference is greater than the preset threshold.
In one embodiment, the image recognition module 12 further includes:
the identification result acquisition unit is used for identifying the character positions of the certificate pictures to obtain identification results.
And the text box acquisition unit is used for generating at least one text box according to the identification result.
And the area determining unit is used for taking the areas corresponding to all the text boxes as foreground areas corresponding to the certificate pictures and taking all the areas except the foreground areas of the certificate pictures as background areas.
In one embodiment, the feature processing module 13 further includes:
the first feature vector acquisition unit is used for extracting features of the pixel points of the foreground region to obtain a first feature vector and position coding information corresponding to the first feature vector.
And the second feature vector acquisition unit is used for extracting features of the pixel points of the background area to obtain a second feature vector and position coding information corresponding to the second feature vector.
The downsampling unit is used for downsampling the first feature vector and the position coding information corresponding to the first feature vector, and the second feature vector and the position coding information corresponding to the second feature vector to obtain a first downsampled feature vector and a second downsampled feature vector, wherein the first downsampled feature vector and the second downsampled feature vector are high-dimensional feature vectors containing the position information.
And the encoding unit is used for respectively carrying out encoding compression processing on the first downsampled feature vector and the second downsampled feature vector to obtain a first recoded feature vector and a second recoded feature vector.
And the up-sampling unit is used for respectively carrying out up-sampling operation on the first recoding feature vector and the second recoding feature vector to obtain a foreground feature vector and a background feature vector.
In one embodiment, the downsampling unit further comprises:
and the first coding feature vector acquisition unit is used for multiplying the first feature vector of the same convolution layer with the position coding information corresponding to the first feature vector to obtain a first coding feature vector.
And the second coding feature vector acquisition unit is used for multiplying the second feature vector of the same convolution layer with the position coding information corresponding to the second feature vector to obtain a second coding feature vector.
And the downsampling processing unit is used for respectively downsampling the first coding feature vector and the second coding feature vector based on the maximum pooling layer to obtain a first downsampled feature vector and a second downsampled feature vector.
In one embodiment, the encoding unit further comprises:
the cutting unit is used for respectively cutting the first downsampling feature vector and the second downsampling feature vector based on a preset cutting mode to obtain a first feature vector block set corresponding to the first downsampling feature vector and a second feature vector block set corresponding to the second downsampling feature vector, wherein the first feature vector block set comprises at least two feature vector blocks, and the second feature vector block set comprises at least two feature vector blocks.
The first coding vector block acquisition unit is used for arbitrarily selecting one feature vector block from the first feature vector block set as a first feature vector block, and carrying out coding processing on the first feature vector block to obtain a corresponding first coding vector block.
The second coding vector block acquisition unit is used for arbitrarily selecting one characteristic vector block from the second characteristic vector block set as a second characteristic vector block, and carrying out coding processing on the second characteristic vector block to obtain a corresponding second coding vector block.
And the compression unit is used for respectively fusing and compressing all the first coding vector blocks and all the second coding vector blocks to obtain corresponding first recoding feature vectors and second recoding feature vectors.
In one embodiment, the compression unit further comprises:
and the dot multiplication unit is used for respectively carrying out dot multiplication calculation on all the first coding vector blocks and all the second coding vector blocks, and dividing the result of the dot multiplication calculation by a preset constant value to obtain a first coding vector matrix and a second coding vector matrix.
And the attention calculating unit is used for respectively carrying out attention calculation on the first coding vector matrix, the second coding vector matrix and the query feature matrix to obtain a first attention feature matrix and a second attention feature matrix, wherein the query feature matrix is a preset feature matrix.
The normalization processing unit is used for performing normalization processing on the first attention feature matrix and the first coding vector matrix to obtain a corresponding first recoding feature vector, and performing normalization processing on the second attention feature matrix and the second coding vector matrix to obtain a corresponding second recoding feature vector.
In one embodiment, the upsampling unit further comprises:
the connection unit is used for inputting the first recoding feature vector and the second recoding feature vector into the convolution layers with corresponding scales respectively, and performing cross-layer connection to obtain the corresponding first cross-layer feature vector and second cross-layer feature vector.
The deconvolution unit is used for respectively carrying out up-sampling operation on the first cross-layer feature vector and the second cross-layer feature vector based on a deconvolution algorithm to obtain a foreground feature vector corresponding to the first cross-layer feature vector and a background feature vector corresponding to the second cross-layer feature vector.
The meaning of "first" and "second" in the above modules/units is merely to distinguish different modules/units, and is not used to limit which module/unit has higher priority or other limiting meaning. Furthermore, the terms "comprises," "comprising," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or modules is not necessarily limited to those steps or modules that are expressly listed or inherent to such process, method, article, or apparatus, but may include other steps or modules that may not be expressly listed or inherent to such process, method, article, or apparatus, and the partitioning of such modules by means of any other means that may be implemented by such means.
The specific limitation of the certificate abnormal information identification apparatus may be referred to as limitation of the certificate abnormal information identification method hereinabove, and will not be described herein. The above-described respective modules in the certificate abnormal information identifying apparatus may be implemented in whole or in part by software, hardware, and a combination thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.
In one embodiment, a computer device is provided, which may be a server, the internal structure of which may be as shown in fig. 4. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database of the computer device is used for storing data involved in the identification method of the abnormal information of the certificate. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program, when executed by a processor, implements a method for identifying credential anomaly information.
In one embodiment, a computer device is provided that includes a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor executes the computer program to implement the steps of the credential anomaly information identification method of the above-described embodiments, such as steps S101-S106 shown in fig. 2 and other extensions of the method and related steps. Alternatively, the processor may implement the functions of the respective modules/units of the certificate authority information identifying apparatus in the above-described embodiment, such as the functions of the modules 11 to 16 shown in fig. 3, when executing the computer program. In order to avoid repetition, a description thereof is omitted.
The processor may be a central processing unit (Central Processing Unit, CPU), but may also be other general purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), off-the-shelf programmable gate arrays (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like that is a control center of the computer device, connecting various parts of the overall computer device using various interfaces and lines.
The memory may be used to store the computer program and/or modules, and the processor may implement various functions of the computer device by running or executing the computer program and/or modules stored in the memory, and invoking data stored in the memory. The memory may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program (such as a sound playing function, an image playing function, etc.) required for at least one function, and the like; the storage data area may store data (such as audio data, video data, etc.) created according to the use of the cellular phone, etc.
The memory may be integrated in the processor or may be provided separately from the processor.
In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, which when executed by a processor, implements the steps of the credential anomaly information identification method in the above-described embodiment, such as steps S101-S106 shown in fig. 2, and other extensions of the method and extensions of related steps. Alternatively, the computer program when executed by the processor realizes the functions of the respective modules/units of the certificate abnormal information identifying apparatus in the above-described embodiment, such as the functions of the modules 11 to 16 shown in fig. 2. In order to avoid repetition, a description thereof is omitted.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile memory can include Read Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), memory bus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-described division of the functional units and modules is illustrated, and in practical application, the above-described functional distribution may be performed by different functional units and modules according to needs, i.e. the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-described functions.
The above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention, and are intended to be included in the scope of the present invention.
Claims (8)
1. A certificate anomaly information identification method, characterized by comprising:
acquiring a certificate picture, and extracting position coding information from the certificate picture;
Recognizing the character positions of the certificate pictures to obtain recognition results;
generating at least one text box according to the identification result;
taking all the areas corresponding to the text boxes as foreground areas corresponding to the certificate pictures and taking all the areas except the foreground areas of the certificate pictures as background areas, wherein the foreground areas are character areas of the certificate pictures and the background areas are the rest areas except the foreground areas in the certificate pictures;
extracting features of pixel points of a foreground region to obtain a first feature vector and position coding information corresponding to the first feature vector;
extracting features of pixel points of the background area to obtain a second feature vector and position coding information corresponding to the second feature vector;
the first feature vector and the position coding information corresponding to the first feature vector are subjected to downsampling processing, and a first downsampled feature vector and a second downsampled feature vector are obtained, wherein the first downsampled feature vector and the second downsampled feature vector are high-dimensional feature vectors containing position information;
Respectively carrying out coding compression processing on the first downsampled feature vector and the second downsampled feature vector to obtain a first recoded feature vector and a second recoded feature vector;
respectively carrying out up-sampling operation on the first recoding feature vector and the second recoding feature vector to obtain a foreground feature vector and a background feature vector;
based on a preset characteristic value calculation mode, carrying out characteristic value calculation on the foreground characteristic vector to obtain a corresponding foreground characteristic value, carrying out characteristic value calculation on the background characteristic vector to obtain a corresponding background characteristic value, and calculating a difference value between the foreground characteristic value and the background characteristic value to obtain a characteristic difference value;
comparing the characteristic difference value with a preset threshold value to obtain a comparison result;
and if the comparison result is that the characteristic difference value is not larger than the preset threshold value, confirming that the certificate picture has no abnormal information, and if the comparison result is that the characteristic difference value is larger than the preset threshold value, analyzing the foreground characteristic vector and the background characteristic vector based on a morphological characteristic algorithm, and determining an abnormal type.
2. The method of claim 1, wherein the step of downsampling the first and second feature vectors with the corresponding position-coded information to obtain first and second downsampled feature vectors comprises:
Multiplying a first feature vector of the same convolution layer with position coding information corresponding to the first feature vector to obtain a first coding feature vector;
multiplying a second feature vector of the same convolution layer with position coding information corresponding to the second feature vector to obtain a second coding feature vector;
and based on a maximum pooling layer, respectively carrying out downsampling processing on the first coding feature vector and the second coding feature vector to obtain a first downsampled feature vector and a second downsampled feature vector.
3. The method of claim 1, wherein the step of encoding and compressing the first downsampled feature vector and the second downsampled feature vector to obtain a first recoded feature vector and a second recoded feature vector, respectively, comprises:
based on a preset cutting mode, respectively cutting a first downsampling feature vector and a second downsampling feature vector to obtain a first feature vector block set corresponding to the first downsampling feature vector and a second feature vector block set corresponding to the second downsampling feature vector, wherein the first feature vector block set comprises at least two feature vector blocks, and the second feature vector block set comprises at least two feature vector blocks;
Randomly selecting one feature vector block from the first feature vector block set as a first feature vector block, and carrying out coding processing on the first feature vector block to obtain a corresponding first coding vector block;
randomly selecting one feature vector block from the second feature vector block set as a second feature vector block, and carrying out coding processing on the second feature vector block to obtain a corresponding second coding vector block;
and respectively fusing and compressing all the first coding vector blocks and all the second coding vector blocks to obtain corresponding first recoding feature vectors and second recoding feature vectors.
4. A method according to claim 3, wherein the step of fusing and compressing all the first coding vector blocks and all the second coding vector blocks to obtain corresponding first recoding feature vectors and second recoding feature vectors comprises:
respectively performing point multiplication calculation on all the first coding vector blocks and all the second coding vector blocks, and dividing the result of the point multiplication calculation by a preset constant value to obtain a first coding vector matrix and a second coding vector matrix;
Respectively carrying out attention computation on the first coding vector matrix, the second coding vector matrix and the query feature matrix to obtain a first attention feature matrix and a second attention feature matrix, wherein the query feature matrix is a preset feature matrix;
and normalizing the first attention feature matrix and the first coding vector matrix to obtain a corresponding first recoding feature vector, and normalizing the second attention feature matrix and the second coding vector matrix to obtain a corresponding second recoding feature vector.
5. The method of claim 1, wherein the step of upsampling the first recoded feature vector and the second recoded feature vector to obtain a foreground feature vector and a background feature vector, respectively, comprises:
inputting the first recoding feature vector and the second recoding feature vector into convolution layers with corresponding scales respectively, and performing cross-layer connection to obtain a corresponding first cross-layer feature vector and a corresponding second cross-layer feature vector;
and respectively carrying out up-sampling operation on the first cross-layer feature vector and the second cross-layer feature vector based on a deconvolution algorithm to obtain a foreground feature vector corresponding to the first cross-layer feature vector and a background feature vector corresponding to the second cross-layer feature vector.
6. A certificate abnormality information identifying apparatus, comprising:
the picture acquisition module is used for acquiring a certificate picture and extracting position coding information from the certificate picture;
recognizing the character positions of the certificate pictures to obtain recognition results;
generating at least one text box according to the identification result;
taking all the areas corresponding to the text boxes as foreground areas corresponding to the certificate pictures and taking all the areas except the foreground areas of the certificate pictures as background areas, wherein the foreground areas are character areas of the certificate pictures and the background areas are the rest areas except the foreground areas in the certificate pictures;
the feature processing module is used for extracting features of pixel points of the foreground region to obtain a first feature vector and position coding information corresponding to the first feature vector;
extracting features of pixel points of the background area to obtain a second feature vector and position coding information corresponding to the second feature vector;
the first feature vector and the position coding information corresponding to the first feature vector are subjected to downsampling processing, and a first downsampled feature vector and a second downsampled feature vector are obtained, wherein the first downsampled feature vector and the second downsampled feature vector are high-dimensional feature vectors containing position information;
Respectively carrying out coding compression processing on the first downsampled feature vector and the second downsampled feature vector to obtain a first recoded feature vector and a second recoded feature vector;
respectively carrying out up-sampling operation on the first recoding feature vector and the second recoding feature vector to obtain a foreground feature vector and a background feature vector;
the characteristic value acquisition module is used for carrying out characteristic value calculation on the foreground characteristic vector based on a preset characteristic value calculation mode to obtain a corresponding foreground characteristic value, carrying out characteristic value calculation on the background characteristic vector to obtain a corresponding background characteristic value, and calculating a difference value between the foreground characteristic value and the background characteristic value to obtain a characteristic difference value;
the comparison result acquisition module is used for comparing the characteristic difference value with a preset threshold value to obtain a comparison result;
and the comparison module is used for confirming that the certificate picture does not have abnormal information if the comparison result is that the characteristic difference value is not greater than the preset threshold value, and analyzing the foreground characteristic vector and the background characteristic vector based on a morphological characteristic algorithm to determine the abnormal type if the comparison result is that the characteristic difference value is greater than the preset threshold value.
7. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the steps of the credential anomaly information identification method of any one of claims 1 to 5.
8. A computer-readable storage medium storing a computer program, wherein the computer program when executed by a processor implements the steps of the credential anomaly information identification method of any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111130685.8A CN113837287B (en) | 2021-09-26 | 2021-09-26 | Certificate abnormal information identification method, device, equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111130685.8A CN113837287B (en) | 2021-09-26 | 2021-09-26 | Certificate abnormal information identification method, device, equipment and medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113837287A CN113837287A (en) | 2021-12-24 |
| CN113837287B true CN113837287B (en) | 2023-08-29 |
Family
ID=78970183
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111130685.8A Active CN113837287B (en) | 2021-09-26 | 2021-09-26 | Certificate abnormal information identification method, device, equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113837287B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107688784A (en) * | 2017-08-23 | 2018-02-13 | 福建六壬网安股份有限公司 | A kind of character identifying method and storage medium based on further feature and shallow-layer Fusion Features |
| CN109460765A (en) * | 2018-09-25 | 2019-03-12 | 平安科技（深圳）有限公司 | Driving license is taken pictures recognition methods, device and the electronic equipment of image in natural scene |
| CN110751041A (en) * | 2019-09-19 | 2020-02-04 | 平安科技（深圳）有限公司 | Certificate authenticity verification method, system, computer equipment and readable storage medium |
| CN111242074A (en) * | 2020-01-20 | 2020-06-05 | 佛山科学技术学院 | Certificate photo background replacement method based on image processing |
| CN111666932A (en) * | 2020-05-27 | 2020-09-15 | 平安科技（深圳）有限公司 | Document auditing method and device, computer equipment and storage medium |
-
2021
- 2021-09-26 CN CN202111130685.8A patent/CN113837287B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107688784A (en) * | 2017-08-23 | 2018-02-13 | 福建六壬网安股份有限公司 | A kind of character identifying method and storage medium based on further feature and shallow-layer Fusion Features |
| CN109460765A (en) * | 2018-09-25 | 2019-03-12 | 平安科技（深圳）有限公司 | Driving license is taken pictures recognition methods, device and the electronic equipment of image in natural scene |
| CN110751041A (en) * | 2019-09-19 | 2020-02-04 | 平安科技（深圳）有限公司 | Certificate authenticity verification method, system, computer equipment and readable storage medium |
| CN111242074A (en) * | 2020-01-20 | 2020-06-05 | 佛山科学技术学院 | Certificate photo background replacement method based on image processing |
| CN111666932A (en) * | 2020-05-27 | 2020-09-15 | 平安科技（深圳）有限公司 | Document auditing method and device, computer equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113837287A (en) | 2021-12-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10789465B2 (en) | Feature extraction and matching for biometric authentication | |
| CN110414507B (en) | License plate recognition method and device, computer equipment and storage medium | |
| CN110569721A (en) | Recognition model training method, image recognition method, device, equipment and medium | |
| CN110287971B (en) | Data verification method, device, computer equipment and storage medium | |
| CN110738236B (en) | Image matching method and device, computer equipment and storage medium | |
| CN106447592B (en) | Online personalization service per feature descriptor | |
| CN109492642B (en) | License plate recognition method, license plate recognition device, computer equipment and storage medium | |
| CN111191568A (en) | Method, device, equipment and medium for identifying copied image | |
| CN113435330B (en) | Video-based micro-expression recognition method, device, equipment and storage medium | |
| CN113496208B (en) | Video scene classification method and device, storage medium and terminal | |
| CN112668462B (en) | Vehicle damage detection model training, vehicle damage detection method, device, equipment and medium | |
| CN109308704B (en) | Background eliminating method, device, computer equipment and storage medium | |
| CN113435594A (en) | Security detection model training method, device, equipment and storage medium | |
| CN111047496A (en) | Threshold determination method, watermark detection device and electronic equipment | |
| CN111680181A (en) | Abnormal object identification method and terminal equipment | |
| CN112052702A (en) | Method and device for identifying two-dimensional code | |
| CN113837287B (en) | Certificate abnormal information identification method, device, equipment and medium | |
| CN108364024B (en) | Image matching method and device, computer equipment and storage medium | |
| CN116258873A (en) | Position information determining method, training method and device of object recognition model | |
| CN113344961B (en) | Image background segmentation method, device, computing equipment and storage medium | |
| CN113255700B (en) | Image feature map processing method and device, storage medium and terminal | |
| CN115424001A (en) | Scene similarity estimation method and device, computer equipment and storage medium | |
| CN112686847B (en) | Identification card image shooting quality evaluation method and device, computer equipment and medium | |
| Agarwal et al. | A maximum relevancy and minimum redundancy feature selection approach for median filtering forensics | |
| CN111914779A (en) | Table text detection method and device, computer equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |