CN111553208A - Identity recognition method, system, device and medium based on image of people and certificate integration - Google Patents
Identity recognition method, system, device and medium based on image of people and certificate integration Download PDFInfo
- Publication number
- CN111553208A CN111553208A CN202010294075.0A CN202010294075A CN111553208A CN 111553208 A CN111553208 A CN 111553208A CN 202010294075 A CN202010294075 A CN 202010294075A CN 111553208 A CN111553208 A CN 111553208A
- Authority
- CN
- China
- Prior art keywords
- image
- face image
- target face
- target
- space vector
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 33
- 230000010354 integration Effects 0.000 title abstract description 8
- 239000013598 vector Substances 0.000 claims abstract description 54
- 230000001815 facial effect Effects 0.000 claims description 26
- 238000012545 processing Methods 0.000 claims description 26
- 238000013527 convolutional neural network Methods 0.000 claims description 20
- 238000001514 detection method Methods 0.000 claims description 12
- 238000004590 computer program Methods 0.000 claims description 9
- 238000003062 neural network model Methods 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000012549 training Methods 0.000 description 14
- 238000012552 review Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 3
- 238000002790 cross-validation Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 239000002699 waste material Substances 0.000 description 2
- 238000003491 array Methods 0.000 description 1
- 238000012550 audit Methods 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/751—Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Human Computer Interaction (AREA)
- Biomedical Technology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Image Analysis (AREA)
- Collating Specific Patterns (AREA)
Abstract
The invention discloses a method, a system, equipment and a medium for identifying an identity of an image based on the integration of a witness and a certificate. The identity recognition method based on the image of the people and the certificate unification comprises the following steps: acquiring an original image of a testimony of a witness; acquiring a first target face image and a second target face image in the original image; extracting a first feature space vector corresponding to the first target face image and a second feature space vector corresponding to the second target face image; calculating a Euclidean distance between the first feature space vector and the second feature space vector; judging whether the Euclidean distance is smaller than a first preset threshold value or not, and if so, determining that the identity recognition is passed; if not, determining that the identity identification does not pass. The invention can improve the auditing efficiency and the auditing accuracy of the image with the integration of the human evidence and the certificate, and solves the problems of low auditing efficiency and low accuracy caused by manual auditing by using a manual intelligent method.
Description
Technical Field
The invention relates to the field of image processing, in particular to a method, a system, equipment and a medium for identifying an identity of an image based on people and evidence integration.
Background
The human face is used as a stable biological characteristic with obvious individual difference and is widely applied to identity identification. At present, the face recognition technology is already applied to railway automatic ticket checking and gate entering systems, mobile terminal login, payment authentication of financial systems and the like. However, in actual project development, it is found that the multitask face detection algorithm under the complex background has the problems of high false detection rate, low recognition accuracy of low-resolution faces and the like, and a CNN (Convolutional Neural network) network can further improve the high accuracy and robustness of face recognition through structure optimization.
Currently, face detection and recognition technology is well developed and is in large-scale commercial use. However, the quality of the face under the unconstrained condition may be much lower than that of the standard data set, the recognition accuracy is limited by various factors, and the face detection and recognition under the practical application still face huge challenges.
The image of people and evidence unification, such as "handheld certificate photo", as the material of auditing carries out manual audit and wastes time and energy, and manual judgement has individual difference, leads to the higher complaint rate because of erroneous judgement produces.
Disclosure of Invention
The invention aims to overcome the defects of time and labor waste, individual difference of manual judgment and inaccurate judgment in the prior art due to the adoption of manual examination and verification, and provides a method, a system, equipment and a medium for identifying the identity of a human-evidence integrated image.
The invention solves the technical problems through the following technical scheme:
the invention provides an identity recognition method based on a person-certificate-integrated image, which comprises the following steps:
s1, acquiring an original image of a person and a certificate;
s2, acquiring a first target face image and a second target face image in the original image;
the first target face image corresponds to a first real face outside the identification photo, and the second target face image corresponds to a second real face in the identification photo;
s3, extracting a first feature space vector corresponding to the first target face image and a second feature space vector corresponding to the second target face image;
s4, calculating the Euclidean distance between the first characteristic space vector and the second characteristic space vector;
s5, judging whether the Euclidean distance is smaller than a first preset threshold value or not, and if yes, determining that identity recognition is passed; if not, determining that the identity identification does not pass.
Preferably, step S2 includes the following steps:
acquiring an image pyramid of a set level according to the original image, and selecting an image of a target level in the image pyramid as a first image;
and inputting the first image into a multitask convolution neural network model by adopting a sliding window, and detecting the first target face image and the second target face image in the first image through the multitask convolution neural network model.
Preferably, step S2 further includes the following steps:
detecting a first face key point corresponding to the first target face image and a second face key point corresponding to the second target face image through the multitask convolutional neural network model;
adjusting the first target facial image to be aligned with the standard facial image according to the first facial key points and standard facial key points corresponding to the standard facial image;
adjusting the second target facial image to be aligned with the standard facial image according to the second facial keypoints and standard facial keypoints corresponding to the standard facial image;
preferably, when the original image includes a non-real face, step S2 further includes:
acquiring a third target face image corresponding to a non-real face in the original image;
deleting the third target face image from the first, second, and third target face images using a face classifier.
Preferably, step S2 further includes the following steps:
judging whether the resolution of the first target face image is smaller than a second preset threshold value or not, if so, performing super-resolution processing on the first target face image;
judging whether the resolution of the second target face image is smaller than a second preset threshold value or not, if so, performing super-resolution processing on the second target face image;
and/or, step S3 includes:
and extracting a first feature space vector corresponding to the first target face image and a second feature space vector corresponding to the second target face image by adopting a regularization method.
The invention also provides an identity recognition system based on the image of people and evidence integration, which comprises:
the first acquisition module is used for acquiring an original image of the combination of the testimony and the witness;
the second acquisition module is used for acquiring a first target face image and a second target face image in the original image;
the first target face image corresponds to a first real face outside the identification photo, and the second target face image corresponds to a second real face in the identification photo;
an extraction module, configured to extract a first feature space vector corresponding to the first target face image and a second feature space vector corresponding to the second target face image;
a calculation module, configured to calculate a euclidean distance between the first feature space vector and the second feature space vector;
the judging module is used for judging whether the Euclidean distance is smaller than a first preset threshold value or not, and if yes, determining that the identity recognition is passed; if not, determining that the identity identification does not pass.
Preferably, the second obtaining module includes:
the selecting unit is used for acquiring an image pyramid with a set level according to the original image and selecting an image of a target level in the image pyramid as a first image;
a first detection unit, configured to input the first image into a multitask convolutional neural network model by using a sliding window, and detect the first target face image and the second target face image in the first image through the multitask convolutional neural network model.
Preferably, the second obtaining module further comprises:
a second detection unit, configured to detect a first face key point corresponding to the first target face image and a second face key point corresponding to the second target face image through the multitask convolutional neural network model;
a first adjusting unit, configured to adjust the first target face image to be aligned with a standard face image according to the first face key point and a standard face key point corresponding to the standard face image;
a second adjusting unit configured to adjust the second target face image to be aligned with a standard face image according to the second face key point and a standard face key point corresponding to the standard face image;
preferably, when the original image includes a non-real face, the second obtaining module further includes:
the acquisition unit is used for acquiring a third target face image corresponding to a non-real face in the original image;
a deleting unit configured to delete the third target face image from the first target face image, the second target face image, and the third target face image using a face classifier.
Preferably, the second obtaining module further comprises:
the first judging unit is used for judging whether the resolution of the first target face image is detected to be smaller than a second preset threshold value or not, and if so, performing super-resolution processing on the first target face image;
the second judging unit is used for judging whether the resolution of the second target face image is detected to be smaller than a second preset threshold value or not, and if so, performing super-resolution processing on the second target face image;
the extraction module comprises:
and the extracting unit is used for extracting a first feature space vector corresponding to the first target face image and a second feature space vector corresponding to the second target face image by adopting a regularization method.
The invention also provides electronic equipment which comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, wherein the identity recognition method based on the image integrated with the human body and the certificate is realized when the processor executes the computer program.
The invention also provides a computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, implements the steps of the aforementioned method for identifying an identity based on a unified image of a person and a certificate.
The positive progress effects of the invention are as follows:
the method comprises the steps of obtaining a first target face image and a second target face image in an original image; extracting a first feature space vector corresponding to the first target face image and a second feature space vector corresponding to the second target face image; calculating a Euclidean distance between the first feature space vector and the second feature space vector; judging whether the Euclidean distance is smaller than a first preset threshold value or not, and if so, determining that the identity recognition is passed; if not, determining that the identity identification does not pass. Compared with the existing manual review which is time-consuming and labor-consuming and has individual difference in manual review, the complaint rate caused by misjudgment is higher, and the method can effectively improve the review efficiency and accuracy.
Drawings
Fig. 1 is a flowchart of a method for identifying an identity based on a combined image of a person and a certificate according to embodiment 1 of the present invention;
fig. 2 is a schematic block diagram of an identity recognition system based on a unified image of a witness in accordance with embodiment 2 of the present invention;
fig. 3 is a block diagram of a second obtaining module according to embodiment 2 of the present invention;
fig. 4 is a schematic structural diagram of an electronic device according to embodiment 3 of the present invention.
Detailed Description
The invention is further illustrated by the following examples, which are not intended to limit the scope of the invention.
Example 1
As shown in fig. 1, the embodiment discloses an identity recognition method based on a person-certificate-integrated image, which includes the following steps:
s101, acquiring an original image of a person and certificate in one;
step S102, acquiring an image pyramid with a set level according to the original image, and selecting an image of a target level in the image pyramid as a first image;
in this example, use is made ofThe image pyramid is obtained from the original image by the scale factor, and the side length of the minimum image in the image pyramid is not less than 20 pixels.
Step S103, inputting the first image into a multitask convolution neural network model by adopting a sliding window, and detecting a first target face image, a second target face image and a third target face image corresponding to an unreal human face in the original image in the first image by the multitask convolution neural network model;
in this embodiment, the facial image is detected by using a Focal MTCNN (a multitask convolutional neural network) model.
The size of the sliding window is 12 × 12, but other sizes may be selected according to actual circumstances.
The first target facial image corresponds to a first real face outside the identification photo, and the second target facial image corresponds to a second real face in the identification photo;
in this embodiment, the multitask convolutional neural network is obtained by using a very large scale data set and a Focal local (Loss function) training, and Batch Normalization is added to the multitask convolutional neural network.
Step S104, deleting the third target face image from the first target face image, the second target face image and the third target face image by adopting a face classifier.
In this embodiment, an AdaBoost + Haar face classifier provided by OpenCV (Open source Computer Vision Library) is used, and the false detection face is deleted step by a cascaded two classifier.
Step S105, judging whether the resolution of the first target face image is detected to be smaller than a second preset threshold value, if so, executing step S106; if not, the super resolution processing is not required to be performed on the first target face image, and the step S109 is directly performed;
step S106, performing super-resolution processing on the first target face image;
step S107, judging whether the resolution of the second target face image is detected to be smaller than a second preset threshold value, if so, executing step S108; if not, the second target face image does not need to be subjected to super-resolution processing;
step S108, performing super-resolution processing on the second target face image;
in this embodiment, a Super-Resolution technology based on an Enhanced Super-Resolution general network (Enhanced Super-Resolution generated countermeasure network) is adopted, so that the face recognition accuracy is improved.
Specifically, when it is detected that the diagonal length of the first target face image is smaller than 80 pixels, and the resolution of the first target face image is smaller than the second preset threshold, the × 4 super-resolution processing is performed, and of course, the × n super-resolution processing may be performed according to an actual situation.
Step S109, detecting a first face key point corresponding to the first target face image and a second face key point corresponding to the second target face image through the multitask convolution neural network model;
step S110, adjusting the first target face image to be aligned with the standard face image according to the first face key point and a standard face key point corresponding to the standard face image;
step S111, adjusting the first target face image to be aligned with the standard face image according to the second face key point and a standard face key point corresponding to the standard face image;
in the present embodiment, the coordinates of 5 key points in a standard face image with a resolution of 112 × 112 are shown in the following table:
key points | X coordinate | Y coordinate |
Center of left eye | 38.2946 | 51.6963 |
Center of right eye | 73.5318 | 51.5014 |
Nose tip | 56.0252 | 71.7366 |
Left mouth corner | 41.5493 | 92.3655 |
Right mouth corner | 70.7300 | 92.2041 |
The super-resolution processing is to amplify the coordinates of the key points of the face in the same proportion.
Step S112, training and extracting a first feature space vector corresponding to the first target face image and a second feature space vector corresponding to the second target face image by adopting a regularization method;
among other things, regularization methods include, but are not limited to, DropBlock ResNet-34.
In this embodiment, the feature space vector is the second-to-last layer of the network, and can be obtained directly from the trained model.
Step S113, calculating the Euclidean distance between the first characteristic space vector and the second characteristic space vector;
step S114, judging whether the Euclidean distance is smaller than a first preset threshold value, if so, executing step S115; if not, go to step S116;
step S115, determining that the identity identification is passed;
and step S116, determining that the identification fails.
In this embodiment, when the accuracy is guaranteed to reach 0.95 and the recall rate reaches 0.8, the first preset threshold may be set to 1.3.
The first set threshold and the second set threshold can be adjusted according to actual conditions, and the optimal threshold is determined by adopting cross validation.
In addition, the trained model is obtained by training a super-large scale data set MS-Celeb-1M, and the trained model is verified by using a natural scene unconstrained face recognition data set LFW. Where the training samples are pre-processed using random luminance, contrast and saturation, specifically, the image RGB channel is subtracted by 127.5 per pixel value and divided by 128. And reduces the learning rate on a schedule during training.
The following is illustrated with reference to specific examples:
firstly, acquiring an original image of a handheld certificate photo, acquiring an image of a target level in a pyramid based on the original image, inputting the image of the target level in the pyramid into a multitask convolutional neural network model by using a sliding window, detecting a face image of the handheld certificate photo in the image and key points corresponding to the face image as well as key points corresponding to the face image of a person in the certificate photo by using Focal MTCNN, deleting a false detection face by using an AdaBoost + Haar face classifier, judging whether the resolution of the face image of the handheld certificate photo and the face image of the person in the certificate photo are lower than a given threshold value or not, if so, performing super-resolution processing, if not, directly aligning the face image of the handheld certificate photo and the face image of the person in the certificate photo with a standard face image respectively, and extracting feature space vectors of the face image of the handheld certificate photo and the face image of the person in the certificate photo by using DropBlock Net-34 training, finally, whether the identity recognition is passed or not is determined by comparing the Euclidean distance calculated by the characteristic space vector with a set threshold value.
In the identity recognition method based on the image of the combination of the person and the certificate, a first target face image and a second target face image in the original image are obtained; extracting a first feature space vector corresponding to the first target face image and a second feature space vector corresponding to the second target face image; calculating a Euclidean distance between the first feature space vector and the second feature space vector; judging whether the Euclidean distance is smaller than a first preset threshold value or not, and if so, determining that the identity recognition is passed; if not, determining that the identity identification does not pass. Compared with the existing manual review which is time-consuming and labor-consuming and has individual difference, the original image complaint rate of the testimony integration caused by misjudgment is higher, and the invention can improve the review efficiency and accuracy.
Example 2
As shown in fig. 2, the present embodiment discloses an identification system based on an image integrated by people and certificates, which includes:
the first acquisition module 1 is used for acquiring an original image of a person and certificate combination;
a second obtaining module 2, configured to obtain a first target face image and a second target face image in the original image;
the first target face image corresponds to a first real face outside the identification photo, and the second target face image corresponds to a second real face in the identification photo;
an extraction module 3, configured to extract a first feature space vector corresponding to the first target face image and a second feature space vector corresponding to the second target face image by using a regularization method;
a calculation module 4, configured to calculate a euclidean distance between the first feature space vector and the second feature space vector;
the judging module 5 is used for judging whether the Euclidean distance is smaller than a first preset threshold value, and if yes, determining that the identity identification is passed; if not, determining that the identity identification does not pass.
In this embodiment, when the accuracy is guaranteed to reach 0.95 and the recall rate reaches 0.8, the first preset threshold may be set to 1.3.
The first set threshold and the second set threshold can be adjusted according to actual conditions, and the optimal threshold is determined by adopting cross validation.
As shown in fig. 3, the second obtaining module 2 in this embodiment includes:
a selecting unit 21, configured to obtain an image pyramid of a set level according to the original image, and select an image of a target level in the image pyramid as a first image;
in the examples, use is made ofThe image pyramid is obtained from the original image by the scale factor, and the side length of the minimum image in the image pyramid is not less than 20 pixels.
A first detecting unit 22, configured to input the first image into a multitask convolutional neural network model by using a sliding window, and detect, through the multitask convolutional neural network model, a first target face image in the first image, a second target face image, and a third target face image corresponding to an unreal human face in the original image;
in the embodiment, the facial image is detected by adopting a Focal MTCNN model.
The first target facial image corresponds to a first real face outside the identification photo, and the second target facial image corresponds to a second real face in the identification photo;
in this embodiment, the multitask convolutional neural network is obtained by using a very large scale data set and a Focal local training, and Batch Normalization is added to the multitask convolutional neural network.
A deleting unit 23 configured to delete the third target face image from the first target face image, the second target face image, and the third target face image using a face classifier.
In this embodiment, the false detection face is deleted step by using an AdaBoost + Haar face classifier provided by OpenCV and using a cascaded two-classifier.
A first judging unit 24, configured to judge whether the resolution of the first target face image is smaller than a second preset threshold, and if so, perform super-resolution processing on the first target face image; if not, performing super-resolution processing on the first target face image is not needed;
a second judging unit 25, configured to judge whether the resolution of the second target face image is smaller than a second preset threshold, and if so, perform super-resolution processing on the second target face image; if not, the second target face image does not need to be subjected to super-resolution processing;
in the embodiment, the super-resolution technology based on the ESRGAN is adopted, so that the face recognition precision is improved.
Specifically, when it is detected that the diagonal length of the first target face image is smaller than 80 pixels, and the resolution of the first target face image is smaller than the second preset threshold, the × 4 super-resolution processing is performed, and of course, the × n super-resolution processing may be performed according to an actual situation.
A second detecting unit 26, configured to detect a first face key point corresponding to the first target face image and a second face key point corresponding to the second target face image through the multitask convolutional neural network model;
a first adjusting unit 27 for adjusting the first target face image to be aligned with a standard face image according to the first face key points and standard face key points corresponding to the standard face image;
a second adjusting unit 28, configured to adjust the second target face image to be aligned with the standard face image according to the first face key points and standard face key points corresponding to the standard face image.
In the present embodiment, the coordinates of 5 key points in a standard face image with a resolution of 112 × 112 are shown in the following table:
key points | X coordinate | Y coordinate |
Center of left eye | 38.2946 | 51.6963 |
Center of right eye | 73.5318 | 51.5014 |
Nose tip | 56.0252 | 71.7366 |
Left mouth corner | 41.5493 | 92.3655 |
Right mouth corner | 70.7300 | 92.2041 |
The super-resolution processing is to amplify the coordinates of the key points of the face in the same proportion.
In addition, the trained model is obtained by training a super-large scale data set MS-Celeb-1M, and the trained model is verified by using a natural scene unconstrained face recognition data set LFW. Where the training samples are pre-processed using random luminance, contrast and saturation, specifically, the image RGB channel is subtracted by 127.5 per pixel value and divided by 128. And reduces the learning rate on a schedule during training.
The following is illustrated with reference to specific examples:
firstly, acquiring an original image of a handheld certificate photo, acquiring an image of a target level in a pyramid based on the original image, inputting the image of the target level in the pyramid into a multitask convolutional neural network model by using a sliding window, detecting a face image of the handheld certificate photo in the image and key points corresponding to the face image as well as key points corresponding to the face image of a person in the certificate photo by using Focal MTCNN, deleting a false detection face by using an AdaBoost + Haar face classifier, judging whether the resolution of the face image of the handheld certificate photo and the face image of the person in the certificate photo are lower than a given threshold value or not, if so, performing super-resolution processing, if not, directly aligning the face image of the handheld certificate photo and the face image of the person in the certificate photo with a standard face image respectively, and extracting feature space vectors of the face image of the handheld certificate photo and the face image of the person in the certificate photo by using DropBlock Net-34 training, finally, whether the identity recognition is passed or not is determined by comparing the Euclidean distance calculated by the characteristic space vector with a set threshold value.
In this embodiment, the feature space vector is the second-to-last layer of the network, and can be obtained directly from the trained model. The trained model is obtained by training a super-large-scale data set MS-Celeb-1M, and the trained model is verified by using a natural scene unconstrained face recognition data set LFW. Where the training samples are pre-processed using random luminance, contrast and saturation, specifically, the image RGB channel is subtracted by 127.5 per pixel value and divided by 128. And reduces the learning rate on a schedule during training.
The identity recognition system based on the image integrated with the person and the certificate disclosed by the embodiment obtains a first target face image and a second target face image in the original image; extracting a first feature space vector corresponding to the first target face image and a second feature space vector corresponding to the second target face image; calculating a Euclidean distance between the first feature space vector and the second feature space vector; judging whether the Euclidean distance is smaller than a first preset threshold value or not, and if so, determining that the identity recognition is passed; if not, determining that the identity identification does not pass. Compared with the existing manual review which is time-consuming and labor-consuming and has individual difference, the original image complaint rate of the testimony integration caused by misjudgment is higher, and the invention can improve the review efficiency and accuracy.
Example 3
Fig. 4 is a schematic structural diagram of an electronic device according to embodiment 3 of the present invention. The electronic device comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, wherein the identity recognition method based on the unified image of the person and the certificate provided by the embodiment 1 is realized when the processor executes the program. The electronic device 30 shown in fig. 4 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiment of the present invention.
As shown in fig. 4, the electronic device 30 may be embodied in the form of a general purpose computing device, which may be, for example, a server device. The components of the electronic device 30 may include, but are not limited to: the at least one processor 31, the at least one memory 32, and a bus 33 connecting the various system components (including the memory 32 and the processor 31).
The bus 33 includes a data bus, an address bus, and a control bus.
The memory 32 may include volatile memory, such as Random Access Memory (RAM)321 and/or cache memory 322, and may further include Read Only Memory (ROM) 323.
The processor 31 executes various functional applications and data processing, such as the identification method based on the unified image of the person and certificate provided in embodiment 1 of the present invention, by running the computer program stored in the memory 32.
The electronic device 30 may also communicate with one or more external devices 34 (e.g., keyboard, pointing device, etc.). Such communication may be through input/output (I/O) interfaces 35. Also, model-generating device 30 may also communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network, such as the Internet) via network adapter 36. As shown, network adapter 36 communicates with the other modules of model-generating device 30 via bus 33. It should be understood that although not shown in the figures, other hardware and/or software modules may be used in conjunction with the model-generating device 30, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID (disk array) systems, tape drives, and data backup storage systems, etc.
It should be noted that although in the above detailed description several units/modules or sub-units/modules of the electronic device are mentioned, such a division is merely exemplary and not mandatory. Indeed, the features and functionality of two or more of the units/modules described above may be embodied in one unit/module according to embodiments of the invention. Conversely, the features and functions of one unit/module described above may be further divided into embodiments by a plurality of units/modules.
Example 4
The present embodiment provides a computer-readable storage medium, on which a computer program is stored, which when executed by a processor implements the steps of the method for identifying an identity based on a unified image for identification and witness provided in embodiment 1.
More specific examples, among others, that the readable storage medium may employ may include, but are not limited to: a portable disk, a hard disk, random access memory, read only memory, erasable programmable read only memory, optical storage device, magnetic storage device, or any suitable combination of the foregoing.
In a possible implementation manner, the present invention can also be implemented in the form of a program product, which includes program code for causing a terminal device to execute the steps of implementing the identification method based on the unified image of human license provided in embodiment 1 when the program product runs on the terminal device.
Where program code for carrying out the invention is written in any combination of one or more programming languages, the program code may be executed entirely on the user device, partly on the user device, as a stand-alone software package, partly on the user device and partly on a remote device or entirely on the remote device.
While specific embodiments of the invention have been described above, it will be appreciated by those skilled in the art that this is by way of example only, and that the scope of the invention is defined by the appended claims. Various changes and modifications to these embodiments may be made by those skilled in the art without departing from the spirit and scope of the invention, and these changes and modifications are within the scope of the invention.
Claims (12)
1. An identity recognition method based on a person-certificate-integrated image is characterized by comprising the following steps:
s1, acquiring an original image of a person and a certificate;
s2, acquiring a first target face image and a second target face image in the original image;
the first target face image corresponds to a first real face outside the identification photo, and the second target face image corresponds to a second real face in the identification photo;
s3, extracting a first feature space vector corresponding to the first target face image and a second feature space vector corresponding to the second target face image;
s4, calculating the Euclidean distance between the first characteristic space vector and the second characteristic space vector;
s5, judging whether the Euclidean distance is smaller than a first preset threshold value or not, and if yes, determining that identity recognition is passed; if not, determining that the identity identification does not pass.
2. The method for identifying an identity based on a unified witness image as claimed in claim 1, wherein the step S2 comprises the steps of:
acquiring an image pyramid of a set level according to the original image, and selecting an image of a target level in the image pyramid as a first image;
and inputting the first image into a multitask convolution neural network model by adopting a sliding window, and detecting the first target face image and the second target face image in the first image through the multitask convolution neural network model.
3. The method for identifying a person-license-based image according to claim 2, wherein the step S2 further comprises the steps of:
detecting a first face key point corresponding to the first target face image and a second face key point corresponding to the second target face image through the multitask convolutional neural network model;
adjusting the first target facial image to be aligned with the standard facial image according to the first facial key points and standard facial key points corresponding to the standard facial image;
adjusting the second target facial image to be aligned with the standard facial image according to the second facial keypoints and standard facial keypoints corresponding to the standard facial image.
4. The method for identifying an identity based on a person-license-united image as claimed in claim 1, wherein when the original image includes an unreal face, the step S2 further includes:
acquiring a third target face image corresponding to a non-real face in the original image;
deleting the third target face image from the first, second, and third target face images using a face classifier.
5. The method for identifying an identity based on a unified image of a witness as claimed in any one of claims 1 to 3, wherein the step S2 further comprises the steps of:
judging whether the resolution of the first target face image is smaller than a second preset threshold value or not, if so, performing super-resolution processing on the first target face image;
judging whether the resolution of the second target face image is smaller than a second preset threshold value or not, if so, performing super-resolution processing on the second target face image;
and/or, step S3 includes:
and extracting a first feature space vector corresponding to the first target face image and a second feature space vector corresponding to the second target face image by adopting a regularization method.
6. The utility model provides an identification system based on image of testimony of a witness unification which characterized in that, identification system includes:
the first acquisition module is used for acquiring an original image of the combination of the testimony and the witness;
the second acquisition module is used for acquiring a first target face image and a second target face image in the original image;
the first target face image corresponds to a first real face outside the identification photo, and the second target face image corresponds to a second real face in the identification photo;
an extraction module, configured to extract a first feature space vector corresponding to the first target face image and a second feature space vector corresponding to the second target face image;
a calculation module, configured to calculate a euclidean distance between the first feature space vector and the second feature space vector;
the judging module is used for judging whether the Euclidean distance is smaller than a first preset threshold value or not, and if yes, determining that the identity recognition is passed; if not, determining that the identity identification does not pass.
7. The people-witness-based image identification system as claimed in claim 6, wherein the second acquisition module comprises:
the selecting unit is used for acquiring an image pyramid with a set level according to the original image and selecting an image of a target level in the image pyramid as a first image;
a first detection unit, configured to input the first image into a multitask convolutional neural network model by using a sliding window, and detect the first target face image and the second target face image in the first image through the multitask convolutional neural network model.
8. The people-witness-based image identification system as claimed in claim 7, wherein the second obtaining module further comprises:
a second detection unit, configured to detect a first face key point corresponding to the first target face image and a second face key point corresponding to the second target face image through the multitask convolutional neural network model;
a first adjusting unit, configured to adjust the first target face image to be aligned with a standard face image according to the first face key point and a standard face key point corresponding to the standard face image;
and the second adjusting unit is used for adjusting the second target face image to be aligned with the standard face image according to the first face key points and standard face key points corresponding to the standard face image.
9. The system for identifying a person-card-integrated-image-based person, according to claim 6, wherein when the original image includes an unreal face, the second obtaining module further includes:
the acquisition unit is used for acquiring a third target face image corresponding to a non-real face in the original image;
a deleting unit configured to delete the third target face image from the first target face image, the second target face image, and the third target face image using a face classifier.
10. The system for identifying a person-license-based image according to any one of claims 6 to 8, wherein the second obtaining module further comprises:
the first judging unit is used for judging whether the resolution of the first target face image is detected to be smaller than a second preset threshold value or not, and if so, performing super-resolution processing on the first target face image;
the second judging unit is used for judging whether the resolution of the second target face image is detected to be smaller than a second preset threshold value or not, and if so, performing super-resolution processing on the second target face image;
the extraction module comprises:
and the extracting unit is used for extracting a first feature space vector corresponding to the first target face image and a second feature space vector corresponding to the second target face image by adopting a regularization method.
11. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor,
the processor, when executing the computer program, implements the method for identifying identity based on unified witness image of any one of claims 1 to 5.
12. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method for identifying an identity based on a unified image of a person and certificate of any one of claims 1 to 5.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010294075.0A CN111553208A (en) | 2020-04-15 | 2020-04-15 | Identity recognition method, system, device and medium based on image of people and certificate integration |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010294075.0A CN111553208A (en) | 2020-04-15 | 2020-04-15 | Identity recognition method, system, device and medium based on image of people and certificate integration |
Publications (1)
Publication Number | Publication Date |
---|---|
CN111553208A true CN111553208A (en) | 2020-08-18 |
Family
ID=72005791
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010294075.0A Pending CN111553208A (en) | 2020-04-15 | 2020-04-15 | Identity recognition method, system, device and medium based on image of people and certificate integration |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111553208A (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107423700A (en) * | 2017-07-17 | 2017-12-01 | 广州广电卓识智能科技有限公司 | The method and device of testimony verification |
CN107944395A (en) * | 2017-11-27 | 2018-04-20 | 浙江大学 | A kind of method and system based on neutral net verification testimony of a witness unification |
CN109255299A (en) * | 2018-08-13 | 2019-01-22 | 北京市商汤科技开发有限公司 | Identity identifying method and device, electronic equipment and storage medium |
CN110414454A (en) * | 2019-07-31 | 2019-11-05 | 南充折衍智能光电科技有限公司 | A kind of testimony of a witness unification identifying system based on machine vision |
-
2020
- 2020-04-15 CN CN202010294075.0A patent/CN111553208A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107423700A (en) * | 2017-07-17 | 2017-12-01 | 广州广电卓识智能科技有限公司 | The method and device of testimony verification |
CN107944395A (en) * | 2017-11-27 | 2018-04-20 | 浙江大学 | A kind of method and system based on neutral net verification testimony of a witness unification |
CN109255299A (en) * | 2018-08-13 | 2019-01-22 | 北京市商汤科技开发有限公司 | Identity identifying method and device, electronic equipment and storage medium |
CN110414454A (en) * | 2019-07-31 | 2019-11-05 | 南充折衍智能光电科技有限公司 | A kind of testimony of a witness unification identifying system based on machine vision |
Non-Patent Citations (1)
Title |
---|
王荣生: "基于深度学习的人脸检测和识别关键技术研究与实现" * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111191539B (en) | Certificate authenticity verification method and device, computer equipment and storage medium | |
CN110826519B (en) | Face shielding detection method and device, computer equipment and storage medium | |
CN103065134B (en) | A kind of fingerprint identification device and method with information | |
CN110569878B (en) | Photograph background similarity clustering method based on convolutional neural network and computer | |
WO2018028546A1 (en) | Key point positioning method, terminal, and computer storage medium | |
CN110197146B (en) | Face image analysis method based on deep learning, electronic device and storage medium | |
CN111950329A (en) | Target detection and model training method and device, computer equipment and storage medium | |
WO2019033572A1 (en) | Method for detecting whether face is blocked, device and storage medium | |
WO2019033574A1 (en) | Electronic device, dynamic video face recognition method and system, and storage medium | |
CN112364827B (en) | Face recognition method, device, computer equipment and storage medium | |
CN111191532A (en) | Face recognition method and device based on construction area and computer equipment | |
CN113449704B (en) | Face recognition model training method and device, electronic equipment and storage medium | |
CN111160202A (en) | AR equipment-based identity verification method, AR equipment-based identity verification device, AR equipment-based identity verification equipment and storage medium | |
CN110795714A (en) | Identity authentication method and device, computer equipment and storage medium | |
WO2021151319A1 (en) | Card edge detection method, apparatus, and device, and readable storage medium | |
CN110910445A (en) | Object size detection method and device, detection equipment and storage medium | |
CN111222433A (en) | Automatic face auditing method, system, equipment and readable storage medium | |
CN110443184A (en) | ID card information extracting method, device and computer storage medium | |
CN112560683A (en) | Method and device for identifying copied image, computer equipment and storage medium | |
CN113378764A (en) | Video face acquisition method, device, equipment and medium based on clustering algorithm | |
CN111626132A (en) | Model generation method, face recognition method, system, device and medium | |
CN111950546A (en) | License plate recognition method and device, computer equipment and storage medium | |
CN109241942B (en) | Image processing method and device, face recognition equipment and storage medium | |
CN111008635A (en) | OCR-based multi-bill automatic identification method and system | |
CN113642428B (en) | Face living body detection method and device, electronic 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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200818 |