CN117809348A - Security face comparison search system - Google Patents
Security face comparison search system Download PDFInfo
- Publication number
- CN117809348A CN117809348A CN202311808254.1A CN202311808254A CN117809348A CN 117809348 A CN117809348 A CN 117809348A CN 202311808254 A CN202311808254 A CN 202311808254A CN 117809348 A CN117809348 A CN 117809348A
- Authority
- CN
- China
- Prior art keywords
- face
- iris
- module
- information
- comparison search
- 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
- 238000001514 detection method Methods 0.000 claims abstract description 65
- 238000000034 method Methods 0.000 claims abstract description 65
- 239000013598 vector Substances 0.000 claims description 55
- 230000006870 function Effects 0.000 claims description 20
- 238000004891 communication Methods 0.000 claims description 18
- 230000003044 adaptive effect Effects 0.000 claims description 15
- 238000013139 quantization Methods 0.000 claims description 15
- 238000004364 calculation method Methods 0.000 claims description 13
- 238000013500 data storage Methods 0.000 claims description 12
- 238000013523 data management Methods 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 10
- 238000012217 deletion Methods 0.000 claims description 9
- 230000037430 deletion Effects 0.000 claims description 9
- 239000000284 extract Substances 0.000 claims description 9
- 238000013475 authorization Methods 0.000 claims description 8
- 238000000605 extraction Methods 0.000 claims description 7
- 238000012795 verification Methods 0.000 claims description 6
- 230000002708 enhancing effect Effects 0.000 claims description 5
- 238000013135 deep learning Methods 0.000 claims description 4
- 238000011156 evaluation Methods 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 230000003287 optical effect Effects 0.000 claims description 4
- 238000012549 training Methods 0.000 claims description 4
- 238000010586 diagram Methods 0.000 description 5
- 230000005540 biological transmission Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005315 distribution function Methods 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 238000004321 preservation Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000005428 wave function Effects 0.000 description 1
Abstract
The invention discloses a security face comparison search system, which relates to the field of face recognition, and is characterized in that a face region is positioned through a face detection algorithm, face features are extracted, iris features are extracted through an iris detection algorithm, face information and iris information acquired by a camera are compared with face information and iris information acquired in advance through a search matching algorithm, a comparison search result and an alarm signal are output, and the accuracy of face comparison is improved; the method comprises the steps of sending an alarm short message to a user through a front-end server, generating a log file for storing comparison search results through log record, displaying face information, iris information and the comparison search results in the form of images and texts through a PC end application program, encrypting the face information and the iris information through an asymmetric encryption algorithm, authenticating and authorizing the identity of a visitor through a dynamic password mode, and protecting the safety of the face information and the iris information.
Description
Technical Field
The invention relates to the field of face recognition, in particular to a security face comparison search system.
Background
Because the traditional security monitoring system is difficult to meet the requirements of real-time performance, accuracy and the like, the conditions of missing report, false report and the like are easy to occur, and therefore, certain influence is generated on security management. The development of the face recognition technology provides a high-efficiency and accurate solution for security monitoring, and the security face comparison search system is a security system based on the face recognition technology, and has the main functions of comparing and searching face images shot by a camera, thereby realizing the identification of personnel.
On one hand, the traditional security face comparison search system has limited accuracy, and factors such as illumination, angle, expression, makeup and the like can influence the accuracy of face recognition, so that the accuracy of the security face comparison search system is reduced; on the other hand, the traditional security face comparison search system lacks a data protection function and has a data leakage risk, so the invention discloses a security face comparison search system, and aims to provide a security face comparison search system capable of improving face comparison accuracy and protecting data safety.
Disclosure of Invention
Aiming at the defects of the prior art, the invention discloses a security face comparison search system, wherein a face recognition module recognizes a face region in face information through a face detection algorithm and extracts face features, the face recognition module extracts iris features through an iris detection algorithm, and a comparison search module compares the face information and iris information acquired by a camera with the face information and iris information acquired in advance through a retrieval matching algorithm and outputs comparison search results and alarm signals, so that the accuracy of face comparison is improved; the alarm module sends an alarm short message to a user through a front-end server to realize the alarm function when comparison search fails, the recording module generates a log file for storing comparison search results through log recording to realize the recording of the comparison search results, the interface display module displays face information, iris information and the comparison search results in the form of images and texts through a PC end application program, the data encryption module realizes the encryption of the face information and the iris information through an asymmetric encryption algorithm, and the identity authentication module realizes the identity authentication and authorization of visitors through a dynamic password mode to protect the safety of the face information and the iris information.
The invention adopts the following technical scheme:
a security face comparison search system, the system comprising:
the camera is used for collecting face information and iris information, and the camera collects the face information and the iris information through the optical objective lens and the infrared light source;
the data storage module is used for storing face information and iris information, and the data storage module stores the face information and the iris information acquired by the camera and acquired in advance by adopting a database;
the face recognition module is used for processing the face information and the iris information, extracting face features and iris features, recognizing a face area in the face information through a face detection algorithm, extracting the face features, and extracting the iris features through an iris detection algorithm;
the comparison search module is used for comparing the face information and the iris information acquired by the camera with the face information and the iris information acquired in advance through a search matching algorithm and outputting comparison search results and alarm signals;
the alarm and recording module is used for realizing an alarm function when the face comparison fails and recording comparison search results, and comprises a receiving port, an alarm module and a recording module, wherein the output end of the receiving port is connected with the input ends of the alarm module and the recording module, and the output end of the alarm module is connected with the input end of the recording module;
The interface display module is used for displaying the face information, the iris information and the comparison search result acquired by the camera and acquired in advance in a form of images and texts;
the data protection module is used for protecting the safety of the face information and the iris information, and comprises a data encryption module and an identity authentication module, wherein the data encryption module is used for encrypting the face information and the iris information through an asymmetric encryption algorithm, the identity authentication module is used for authenticating and authorizing the identity of a visitor in a dynamic password mode, and the data encryption module and the identity authentication module independently work in parallel;
the output end of the camera is connected with the input end of the data protection module, the output end of the data protection module is connected with the input end of the data storage module, the output end of the data storage module is connected with the input ends of the face recognition module and the interface display module, the output end of the face recognition module is connected with the input end of the comparison search module, the output end of the comparison search module is connected with the input end of the alarm and recording module, and the output end of the alarm and recording module is connected with the input end of the interface display module.
As a further technical scheme of the invention, the face detection algorithm processes the face picture through the Haar cascade classifier to realize the identification of the face region and the extraction of the face features, and the workflow of the face detection algorithm comprises:
step one, collecting images;
the face detection algorithm receives face pictures acquired by a camera through an API interface and divides the face pictures into positive samples and negative samples, wherein the positive samples are pictures containing faces, and the negative samples are pictures not containing faces;
step two, haar characteristic value extraction;
the face detection algorithm performs denoising, enhancing and normalizing operations on the face picture through an image processing function, and obtains a Haar characteristic value through a Haar characteristic value calculation formula and a Haar characteristic template, wherein the Haar characteristic value calculation formula is as follows:
in equation (1), S (x, y) is the Haar eigenvalue at position (x, y), w i Is the weight of rectangular region i, Σf (x i ,y i ) Is the gray sum of pixel points in a black rectangular area, sigma g (x i ,y i ) The gray sums of pixel points in the white rectangular area are obtained, i is the ordering of the rectangular areas, and n is the number of the rectangular areas;
training a classifier;
The face detection algorithm trains the classifier by using extracted Haar features and an adaptive lifting method, effective features are learned from the existing Haar features, the adaptive lifting method trains the weak classifier to construct a strong classifier based on the weight of the adaptive adjustment Haar features, and the calculation formula of the adaptive lifting method is as follows:
in the formula (2), sign is a de-sign function, M is the number of adaptive iterations, M is the iterative ordering, G m (x) For the m-th iteration weak classifier classification result, N is the number of Haar features, k is the sequence of Haar features,for the weighting coefficients of the kth Haar feature of the mth iteration, I (G m (x k ) Classifying the correct function, gamma, for the kth Haar feature of the mth iteration m Regularization term for the mth iteration;
step four, detecting a face area;
the face detection algorithm adopts a cascading mode to connect the strong classifiers to form a cascading classifier, the cascading classifier eliminates the area which is not a face through the primary classifier, the detection rate of the area which is not a face is reduced, the face detection algorithm carries out image scanning on the input face picture to be detected through a sliding window mode, and the face area is found out;
step five, extracting face feature vectors;
The face detection algorithm processes the well positioned face area by adopting a deep learning method, extracts face features, and normalizes the face features by adopting an Euclidean distance method to obtain face feature vectors.
As a further technical scheme of the invention, the iris detection algorithm decomposes an original iris image into components based on a two-dimensional Gabor filter, the iris detection algorithm adopts a real part and an imaginary part of the two-dimensional Gabor filter to describe an iris, and carries out polarity quantization on a filtering result to obtain an iris feature code, and a calculation formula of the iris detection algorithm is as follows:
in the formula (3), H (x, y) is the output result of the two-dimensional Gabor filter, alpha is the effective width of the two-dimensional Gabor filter, beta is the effective length of the two-dimensional Gabor filter, u 0 And v 0 Determining the frequency and direction of the modulation term, (x) 0 ,y 0 ) For the position in the iris image, when the real part and the imaginary part of the output result are both positive, the polarity quantization value is 11; when the real part is positive and the imaginary part is negative, the polarity quantization value is 10; the real part is positive, the negative imaginary part is positive, and the polarity quantization value is 01; and when the real part and the imaginary part are both negative, the polarity quantization value is 00, so that the iris feature code is obtained.
As a further technical solution of the present invention, the workflow of the search matching algorithm includes:
Step 1, collecting face characteristic information;
the retrieval matching algorithm receives the feature vector of the face or the iris from the face recognition module through an API interface;
step 2, searching a database;
the retrieval matching algorithm calculates the distance between the feature vector of the face or iris to be compared and the feature vector of the face or iris acquired in advance through a Euclidean distance method, and determines the face or iris similar to the face or iris to be compared, wherein the Euclidean distance method has the following formula:
in the formula (4), L is the distance of the feature vector, δ j A, a is a weight coefficient of the j-th element of the face or iris characteristic vector to be compared j For the j-th element of the face or iris characteristic vector to be compared, sigma j B, weighting coefficient of jth element of face or iris characteristic vector acquired in advance j Z is the number of elements of the face or iris feature vector, j is the sequence of the elements of the face or iris feature vector;
step 3, similarity evaluation;
the search matching algorithm calculates the similarity between the face or iris to be compared and the searched face or iris through a cosine similarity method, the cosine similarity method removes the influence of the dimensional difference between vectors on the result through L2 norm normalization, and the cosine similarity method has the following formula:
In the formula (5), D is the similarity between the face or iris to be compared and the retrieved face or iris, A is the feature vector of the face or iris to be compared, and B is the feature vector of the retrieved face or iris;
step 4, judging matching;
the search matching algorithm compares the calculated similarity with a set similarity threshold value through a comparison method, when the similarity is larger than the threshold value, the matching is successful, when the similarity is smaller than the threshold value, the matching is failed, and an alarm signal is output according to the matching result and the search result is compared.
As a further technical scheme of the invention, the receiving port receives the alarm signal of the comparison search module and the comparison search result through the wireless communication protocol, the alarm module sends the alarm short message to the user through the front-end server to realize the warning function when the comparison search fails, and the recording module generates the log file for storing the comparison search result through log recording to realize the recording of the comparison search result.
As a further technical scheme of the invention, the PC application program comprises a connection module, a data reading module, a data display module and a data management module, wherein the connection module establishes a connection between the PC application program and a database through a wireless communication protocol, the data reading module acquires face information, iris information and comparison search result data through inquiring an API interface, the data display module displays the face information, iris information and comparison search result data on the interface in an image and text mode, and the data management module regularly executes backup and deletion operations on the face information, iris information and comparison search result through a timing task.
As a further technical scheme of the invention, the asymmetric encryption algorithm executes encryption operation based on RSA algorithm, the asymmetric encryption algorithm processes the face information and the iris information through public keys and private keys to realize encryption and decryption of the face information and the iris information, the dynamic password executes identity authentication operation through an identity authentication platform, the identity authentication platform comprises a dynamic password generation module, a visitor interface and a verification authorization module, the dynamic password generation module generates and stores the dynamic password based on a dynamic password algorithm of time and sends the dynamic password to a mobile phone of a visitor in a short message mode, the visitor interface realizes dynamic password input by a user through a touch screen, and the verification authorization module compares the dynamic password input by the user with the dynamic password stored by the identity authentication platform through a comparison circuit and authorizes the visitor according to an identity authentication result.
Has the positive beneficial effects that:
the invention discloses a security face comparison search system, wherein a face recognition module recognizes a face region in face information through a face detection algorithm and extracts face features, the face recognition module extracts iris features through an iris detection algorithm, and a comparison search module compares the face information and iris information acquired by a camera with the face information and iris information acquired in advance through a search matching algorithm and outputs comparison search results and alarm signals, so that the accuracy of face comparison is improved; the alarm module sends an alarm short message to a user through a front-end server to realize the alarm function when comparison search fails, the recording module generates a log file for storing comparison search results through log recording to realize the recording of the comparison search results, the interface display module displays face information, iris information and the comparison search results in the form of images and texts through a PC end application program, the data encryption module realizes the encryption of the face information and the iris information through an asymmetric encryption algorithm, and the identity authentication module realizes the identity authentication and authorization of visitors through a dynamic password mode to protect the safety of the face information and the iris information.
Drawings
FIG. 1 is a schematic diagram of the overall architecture of a security face comparison search system of the present invention;
FIG. 2 is a workflow diagram of a face detection algorithm in a security face comparison search system of the present invention;
FIG. 3 is a workflow diagram of a search matching algorithm in a security face comparison search system according to the present invention;
FIG. 4 is a schematic diagram of a PC end application program architecture in a security face comparison search system according to the present invention;
fig. 5 is a schematic diagram of an architecture of an identity authentication platform in a security face comparison search system according to the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. 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.
A security face comparison search system, the system comprising:
the camera is used for collecting face information and iris information, and the camera collects the face information and the iris information through the optical objective lens and the infrared light source;
The data storage module is used for storing face information and iris information, and the data storage module stores the face information and the iris information acquired by the camera and acquired in advance by adopting a database;
the face recognition module is used for processing the face information and the iris information, extracting face features and iris features, recognizing a face area in the face information through a face detection algorithm, extracting the face features, and extracting the iris features through an iris detection algorithm;
the comparison search module is used for comparing the face information and the iris information acquired by the camera with the face information and the iris information acquired in advance through a search matching algorithm and outputting comparison search results and alarm signals;
the alarm and recording module is used for realizing an alarm function when the face comparison fails and recording comparison search results, and comprises a receiving port, an alarm module and a recording module, wherein the output end of the receiving port is connected with the input ends of the alarm module and the recording module, and the output end of the alarm module is connected with the input end of the recording module;
The interface display module is used for displaying the face information, the iris information and the comparison search result acquired by the camera and acquired in advance in a form of images and texts;
the data protection module is used for protecting the safety of the face information and the iris information, and comprises a data encryption module and an identity authentication module, wherein the data encryption module is used for encrypting the face information and the iris information through an asymmetric encryption algorithm, the identity authentication module is used for authenticating and authorizing the identity of a visitor in a dynamic password mode, and the data encryption module and the identity authentication module independently work in parallel;
the output end of the camera is connected with the input end of the data protection module, the output end of the data protection module is connected with the input end of the data storage module, the output end of the data storage module is connected with the input ends of the face recognition module and the interface display module, the output end of the face recognition module is connected with the input end of the comparison search module, the output end of the comparison search module is connected with the input end of the alarm and recording module, and the output end of the alarm and recording module is connected with the input end of the interface display module.
In a specific embodiment, a camera acquires face information and iris information in real time through an optical objective lens and an infrared light source, data are transmitted to a data protection module, the data protection module encrypts the face information and the iris information through a public key of an asymmetric encryption algorithm, a private key decrypts the encrypted face information and the iris information are transmitted to a face recognition module, the face recognition module recognizes a face area in the face information through a face detection algorithm and extracts face features through a deep learning method, meanwhile, the face recognition module extracts iris features through an iris detection algorithm, a search matching algorithm searches the face information and the iris information acquired in advance in a database according to the face information and the iris information acquired by the camera, a face most similar to the face acquired at present is determined, the alarm module sends the alarm signal to the user in the form of alarm short message through front end service, generates a log file through log record to store the compared search result, is convenient for subsequent inquiry and analysis, and a visitor completes identity authentication through an identity authentication platform so as to obtain the authority of reading face information and iris information in a database, then links the database with a PC end application program through a connecting module, reads the face information and iris information in the database through a data reading module, finally displays the face information, iris information and received compared search result data on an interface through a data display module, and periodically transmits the face information through a data management module, and carrying out data backup and deletion on iris information and comparison search results so as to restore and release the storage space in the future.
In the above embodiment, the face detection algorithm processes the face picture through the Haar cascade classifier to realize the recognition of the face region and the extraction of the face feature, and the workflow of the face detection algorithm includes:
step one, collecting images;
the face detection algorithm receives face pictures acquired by a camera through an API interface and divides the face pictures into positive samples and negative samples, wherein the positive samples are pictures containing faces, and the negative samples are pictures not containing faces;
step two, haar characteristic value extraction;
the face detection algorithm performs denoising, enhancing and normalizing operations on the face picture through an image processing function, and obtains a Haar characteristic value through a Haar characteristic value calculation formula and a Haar characteristic template, wherein the Haar characteristic value calculation formula is as follows:
in equation (1), S (x, y) is the Haar eigenvalue at position (x, y), w i Is the weight of rectangular region i, Σf (x i ,y i ) Is the gray sum of pixel points in a black rectangular area, sigma g (x i ,y i ) The gray sums of pixel points in the white rectangular area are obtained, i is the ordering of the rectangular areas, and n is the number of the rectangular areas;
training a classifier;
the face detection algorithm trains the classifier by using extracted Haar features and an adaptive lifting method, effective features are learned from the existing Haar features, the adaptive lifting method trains the weak classifier to construct a strong classifier based on the weight of the adaptive adjustment Haar features, and the calculation formula of the adaptive lifting method is as follows:
In the formula (2), sign is a de-sign function, M is the number of adaptive iterations, M is the iterative ordering, G m (x) For the m-th iteration weak classifier classification result, N is the number of Haar features, k is the sequence of Haar features,for the weighting coefficients of the kth Haar feature of the mth iteration, I (G m (x k ) Classifying the correct function, gamma, for the kth Haar feature of the mth iteration m Regularization term for the mth iteration;
step four, detecting a face area;
the face detection algorithm adopts a cascading mode to connect the strong classifiers to form a cascading classifier, the cascading classifier eliminates the area which is not a face through the primary classifier, the detection rate of the area which is not a face is reduced, the face detection algorithm carries out image scanning on the input face picture to be detected through a sliding window mode, and the face area is found out;
step five, extracting face feature vectors;
the face detection algorithm processes the well positioned face area by adopting a deep learning method, extracts face features, and normalizes the face features by adopting an Euclidean distance method to obtain face feature vectors.
In a specific embodiment, an application platform of the face detection algorithm may be built, and in the working process of the platform, a hardware platform may be built, for example, by building the following components: programmable controller, memory, classifier, bus and display, the following is a more detailed description of the embodiments:
And (3) a programmable controller: the programmable controller can realize the basic operations of carrying out arithmetic and logic operation on data and reading and storing the data in the memory.
A memory: the memory is a hardware device that is mainly used to store the algorithm codes and data of the algorithms.
A classifier: the classifier is mainly used for detecting whether the image acquired by the camera contains the face features or the face patterns.
Bus: the bus is used for connecting all the components of the equipment, and communication and coordination among all the components are realized through data transmission, control and address information.
A display: the display is used for displaying the running state and parameters of the equipment algorithm and providing a window for a user to observe and adjust the parameters in time.
Under the cooperation of the hardware components, the application of a face detection algorithm can be realized according to specific application requirements and system requirements, and the accuracy of face region identification and feature extraction is improved in practical application, so that the accuracy of security comparison search system is improved.
In a specific embodiment, the face detection algorithm is mainly used for detecting a face area in a given image and extracting a feature vector of the face, and the specific working steps are as follows: firstly, a face detection algorithm can call an API interface to receive a face picture acquired by a camera, and in the receiving process, the face picture is classified into a positive sample and a negative sample so as to train a classifier subsequently, wherein the positive sample is usually an image containing a target or a feature, for example, an image containing a face, and the negative sample is an image not containing a target or a feature, usually some background images or other non-target images; secondly, removing noise in a face picture by a face detection algorithm through a wavelet denoising method, enhancing the face picture through a histogram equalization method, and carrying out normalization operation on the picture acquired by a camera through a face alignment method; then, a face detection algorithm builds a plurality of weak classifiers based on an application platform, training the weak classifiers by adopting a Haar characteristic and a self-adaptive lifting method, and combining the weak classifiers into one strong classifier; finally, the face detection algorithm forms a cascade classifier through a plurality of strong classifiers, the cascade classifier adopts a step-by-step classification mode, the detection rate is improved by reducing the detection rate of non-face areas, namely, the areas which cannot be faces are rapidly eliminated through a primary classifier, then the detection rate is improved through a high-level classifier, the determination of the face areas in the picture is realized, meanwhile, the face characteristics are normalized through the Euclidean distance method, the differences of face orientation, scale, brightness and the like can be effectively removed, and accordingly more accurate face characteristic vectors are obtained. In this embodiment, a face detection algorithm and a face recognition algorithm are adopted to process face images acquired by a camera respectively, and the processing results in the same time are recorded in table 1
Table 1 statistical table of treatment effect
The comparison finds that the total quantity, speed and accuracy of the processed pictures of the face detection algorithm are far greater than those of the face detection algorithm, and the practicability and effectiveness of the face detection algorithm are proved.
In the above embodiment, the iris detection algorithm decomposes the original iris image into components based on a two-dimensional Gabor filter, the iris detection algorithm describes the iris by using the real part and the imaginary part of the two-dimensional Gabor filter, and performs polarity quantization on the filtering result to obtain an iris feature code, and the calculation formula of the iris detection algorithm is as follows:
in the formula(3) Wherein H (x, y) is the output result of the two-dimensional Gabor filter, alpha is the effective width of the two-dimensional Gabor filter, beta is the effective length of the two-dimensional Gabor filter, u 0 And v 0 Determining the frequency and direction of the modulation term, (x) 0 ,y 0 ) For the position in the iris image, when the real part and the imaginary part of the output result are both positive, the polarity quantization value is 11; when the real part is positive and the imaginary part is negative, the polarity quantization value is 10; the real part is positive, the negative imaginary part is positive, and the polarity quantization value is 01; and when the real part and the imaginary part are both negative, the polarity quantization value is 00, so that the iris feature code is obtained.
In a specific embodiment, the principle of a two-dimensional Gabor filter is based on the product of a Morse wave function and a Gaussian distribution function, and can be used for image processing in the frequency domain or the time domain. In iris recognition, a two-dimensional Gabor filter may be used to extract texture and morphological features of the iris. The method comprises the following specific steps: firstly, receiving iris images acquired by a camera; secondly, converting an iris image into a gray level image, extracting the edge of the image through a Canny edge detection algorithm, tracking the edge by using an edge tracking algorithm, separating and positioning the inner edge and the outer edge of the iris, enhancing the iris image by adopting a histogram equalization method, normalizing the iris image acquired by a camera based on ellipse fitting, and unifying the parameters such as the size, the shape, the area and the like of the iris to a certain standard; finally, describing the iris by using the real part and the imaginary part of a two-dimensional Gabor filter, performing filtering processing on an iris image by using the iris, and then quantifying the polarity of the filtered result into 2-bit binary numbers, namely respectively comparing the filtered result with 0 to obtain binary values of 0 or 1, and combining the binary values into 2-bit binary numbers, wherein the polarity quantifying rule is as follows: when the real part and the imaginary part are both positive quantized values of 11; the real part is 10 when the positive imaginary part is negative; the real part is negative and the imaginary part is positive and is 01; and when the real part and the imaginary part are negative, the value is 00, so that an iris characteristic code is obtained and is used as the characteristic description of the iris.
In the above embodiment, the workflow of the search matching algorithm includes:
step 1, collecting face characteristic information;
the retrieval matching algorithm receives the feature vector of the face or the iris from the face recognition module through an API interface;
step 2, searching a database;
the retrieval matching algorithm calculates the distance between the feature vector of the face or iris to be compared and the feature vector of the face or iris acquired in advance through a Euclidean distance method, and determines the face or iris similar to the face or iris to be compared, wherein the Euclidean distance method has the following formula:
in the formula (4), L is the distance of the feature vector, δ j A, a is a weight coefficient of the j-th element of the face or iris characteristic vector to be compared j For the j-th element of the face or iris characteristic vector to be compared, sigma j B, weighting coefficient of jth element of face or iris characteristic vector acquired in advance j Z is the number of elements of the face or iris feature vector, j is the sequence of the elements of the face or iris feature vector;
step 3, similarity evaluation;
the search matching algorithm calculates the similarity between the face or iris to be compared and the searched face or iris through a cosine similarity method, the cosine similarity method removes the influence of the dimensional difference between vectors on the result through L2 norm normalization, and the cosine similarity method has the following formula:
In the formula (5), D is the similarity between the face or iris to be compared and the retrieved face or iris, A is the feature vector of the face or iris to be compared, and B is the feature vector of the retrieved face or iris;
step 4, judging matching;
the search matching algorithm compares the calculated similarity with a set similarity threshold value through a comparison method, when the similarity is larger than the threshold value, the matching is successful, when the similarity is smaller than the threshold value, the matching is failed, and an alarm signal is output according to the matching result and the search result is compared.
In a specific embodiment, an application platform for retrieving a matching algorithm may be built, and during the working process of the platform, a hardware platform may be built, for example, by building the following components: multicore CPU, memory, network connections, bus and display, the following is a more detailed description of the embodiments:
multicore CPU: the multi-core CPU is mainly used for data processing and algorithm operation and provides enough calculation resource support.
Memory: the memory is used for storing data to be processed, algorithm codes and intermediate results, and provides the capability of quick reading and writing.
Network connection: the network connection is used for transmitting data and operation results, and can be a wired or wireless network, so that the real-time performance and stability of the camera data and the database are ensured.
Bus: the bus is used for connecting all the components of the equipment, and communication and coordination among all the components are realized through data transmission, control and address information.
A display: the display is used for displaying the running state and parameters of the equipment algorithm and providing a window for a user to observe and adjust the parameters in time.
Under the cooperation of the hardware components, the application of the retrieval matching algorithm can be realized according to specific application requirements and system requirements, and the accuracy of face retrieval comparison is improved in practical application, so that the accuracy of security comparison search systems is improved.
In a specific embodiment, the search matching algorithm uses a face feature vector and an iris feature vector to perform comparison and search, and finds a face which is acquired in advance in a database and is most similar to the face information acquired at present, and the specific steps are as follows: firstly, designing an API interface meeting the requirements according to the requirements of a search matching algorithm and the characteristic vector output format of a face recognition module, wherein the search matching algorithm interacts with the face recognition module through the API interface, and transmits the characteristic vector of a face or an iris to a comparison search module; secondly, carrying out face retrieval, matching the received feature vector of the face or the iris with the feature vector which is pre-acquired and stored in the database, calculating the distance between the feature vector of the current face or the iris and the feature vector which is pre-acquired in the database by using a Euclidean distance method, and when the distance is smaller, indicating that the two feature vectors are more similar, and further finding out the face with the most similar face information; then, the search matching algorithm carries out similarity evaluation on the searched most similar human face through a cosine similarity method, and the cosine similarity method calculates the similarity between the currently collected human face and the searched human face according to the characteristic vector of the currently collected human face or the iris and the characteristic vector of the searched human face or the iris; and finally, setting a threshold value of similarity, judging whether the currently acquired face is matched with a certain face in the database, comparing the calculated similarity with the threshold value by a search matching algorithm through a comparison method, if the similarity exceeds the threshold value, judging that the matching is successful, recording a matching result, if the similarity is lower than the threshold value, judging that the matching is failed, outputting an alarm signal of the matching failure, and recording the matching result.
In the above embodiment, the receiving port receives the alarm signal of the comparison search module and the comparison search result through the wireless communication protocol, the alarm module sends the alarm short message to the user through the front end server to realize the warning function when the comparison search fails, and the recording module generates the log file for storing the comparison search result through log recording to realize the recording of the comparison search result.
In a specific embodiment, firstly, a receiving port of an alarm and record module receives a comparison failure alarm signal and a recorded comparison search result through a wireless communication protocol; secondly, sending alarm signals with failed comparison to a designated user in a short message mode through a front-end server, wherein the specific steps are as follows: 1. and after receiving the alarm request of the comparison search module, the front-end server analyzes and forwards the request to the alarm module. 2. Sending a short message: after receiving the alarm request, the alarm module sends a short message to the user according to a preset mode. 3. And (5) feeding back an alarm result: the alarm module needs to monitor and feed back the sending result, ensures successful sending and timely informs related personnel; and finally, the recording module generates a log file for storing the comparison search result through log recording, so as to realize the recording of the comparison search result. The specific implementation steps are as follows: 1. design log format: the recording module needs to design a proper log format to ensure that recorded contents are rich and clear, and is easy to check and inquire. Typically, the log format includes content such as time stamps, search keywords, search results, and the like. 2. Setting a storage path: the recording module needs to set a storage path, and stores the log file generated by comparing the search results into the designated path so as to facilitate subsequent inquiry. 3. Generating a log file: and the comparison searching module transmits the result to the recording module when the searching is finished, and the recording module generates a log file according to the appointed log format and stores the file into a set storage path. 4. Maintaining a log: to avoid the log files becoming too large, the log files need to be cleaned and backed up periodically, and each log file is named and marked.
In the above embodiment, the PC-side application includes a connection module, a data reading module, a data display module and a data management module, where the connection module establishes a connection between the PC-side application and the database through a wireless communication protocol, the data reading module obtains face information, iris information and comparison search result data through querying an API interface, the data display module displays the face information, iris information and comparison search result data on an interface in an image and text mode, and the data management module periodically executes backup and deletion operations on the face information, iris information and comparison search result through a timing task.
In a specific embodiment, first, the process of the connection module establishing contact between the PC-side application program and the database through the wireless communication protocol needs to involve the following steps: 1. selecting a wireless communication protocol: and selecting Wi-Fi wireless communication protocols according to different application scenes and requirements. 2. Configuration of communication parameters: according to the characteristics and requirements of Wi-Fi wireless communication protocol, the communication parameters are configured. 3. Establishing connection: the connection module establishes connection between the PC end application program and the database through a Wi-Fi wireless communication protocol so as to transmit and exchange data; secondly, the data reading module reads face information, iris information and comparison search result data from the database and the comparison search module by calling an API interface and a data reading code, and then the data display module processes and loads the image data and the comparison search result data obtained from the database and the comparison search module and displays the image data and the comparison search result data on an interface in the form of images and texts; finally, the data management module regularly executes backup and deletion operations on the face information, the iris information and the comparison search result through the timing tasks as follows: 1. determining a backup period and a deletion period: and determining the backup period and the deletion period of the face information, the iris information and the comparison search result according to the actual needs and the data volume. 2. Timing task setting: the timing tasks of backup and deletion operations on face information, iris information and comparison search results are set through a programming language or a third party scheduling tool so as to ensure automation and on-time execution. 3. Data backup operation: in the backup period, the data management module needs to backup the face information, iris information and comparison search result data in the database and store the backup data in a specified backup path. 4. Data deletion operation: in the deleting period, the data management module needs to delete the face information, the iris information and the comparison search result data which are out of date or are not needed any more so as to release the storage space, reduce the storage cost and improve the system efficiency.
In the above embodiment, the asymmetric encryption algorithm performs encryption operation based on the RSA algorithm, the asymmetric encryption algorithm processes the face information and the iris information through the public key and the private key, so as to implement encryption and decryption of the face information and the iris information, the dynamic password performs identity authentication operation through the identity authentication platform, the identity authentication platform includes a dynamic password generating module, a visitor interface and a verification authorization module, the dynamic password generating module generates and stores the dynamic password based on the dynamic password algorithm of time and sends the dynamic password to the mobile phone of the visitor in the form of a short message, the visitor interface implements user input of the dynamic password through the touch screen, and the verification authorization module compares the dynamic password input by the user with the dynamic password stored by the identity authentication platform through the comparison circuit and authorizes the visitor according to the identity authentication result.
In a specific embodiment, the specific steps of the asymmetric encryption algorithm are as follows: 1. a pair of public and private keys are generated using the RSA algorithm, wherein the public key is used to encrypt data and the private key is used to decrypt data. The public-private key is a pair of large primes that are associated with each other. 2. Encrypting the data using the public key: and encrypting the data needing to be encrypted by using the generated public key, and generating the encrypted data. The encrypted data can only be decrypted using the private key. 3. Transmitting the encrypted data through a communication channel: the encrypted data is transmitted to a receiving party through a communication channel. 4. Decrypting the data using the private key: the receiving side decrypts the encrypted data by using the generated private key and obtains the original content of the data. In the working process of the specific embodiment, the data encryption module encrypts the face information and the iris information through an asymmetric encryption algorithm, and ensures the safety and privacy of data; the dynamic password generating module generates and stores the dynamic password based on a time dynamic password algorithm and sends the dynamic password to a mobile phone of a visitor in a short message mode, and the method comprises the following specific implementation steps: 1. time synchronization: the dynamic password algorithm is time-based and therefore needs to be time-synchronized with the visitor's cell phone to ensure that the generated dynamic password is consistent with the password on the cell phone. 2. Dynamic password algorithm: a time-based dynamic password algorithm is selected, and configured and arranged at an authentication platform to generate a unique, one-time, time-sensitive password. 3. Dynamic password preservation: the identity authentication platform needs to store the dynamic password in a database and set an expiration date and expiration time to ensure safety and accuracy. 4. And sending a short message: the generated dynamic password is sent to the mobile phone of the visitor in the form of a short message. And then the visitor inputs the password through the touch screen, the identity authentication platform verifies the password through the comparison circuit, the dynamic password input by the visitor is compared with the dynamic password stored by the identity authentication platform, if the password is up, the visitor is authorized to access, and if the password is inconsistent, the visitor is not allowed to access.
While specific embodiments of the present invention have been described above, it will be understood by those skilled in the art that these specific embodiments are by way of example only, and that various omissions, substitutions, and changes in the form and details of the methods and systems described above may be made by those skilled in the art without departing from the spirit and scope of the invention. For example, it is within the scope of the present invention to combine the above-described method steps to perform substantially the same function in substantially the same way to achieve substantially the same result. Accordingly, the scope of the invention is limited only by the following claims.
Claims (7)
1. A security face comparison search system is characterized in that: the system comprises:
the camera is used for collecting face information and iris information, and the camera collects the face information and the iris information through the optical objective lens and the infrared light source;
the data storage module is used for storing face information and iris information, and the data storage module stores the face information and the iris information acquired by the camera and acquired in advance by adopting a database;
the face recognition module is used for processing the face information and the iris information, extracting face features and iris features, recognizing a face area in the face information through a face detection algorithm, extracting the face features, and extracting the iris features through an iris detection algorithm;
The comparison search module is used for comparing the face information and the iris information acquired by the camera with the face information and the iris information acquired in advance through a search matching algorithm and outputting comparison search results and alarm signals;
the alarm and recording module is used for realizing an alarm function when the face comparison fails and recording comparison search results, and comprises a receiving port, an alarm module and a recording module, wherein the output end of the receiving port is connected with the input ends of the alarm module and the recording module, and the output end of the alarm module is connected with the input end of the recording module;
the interface display module is used for displaying the face information, the iris information and the comparison search result acquired by the camera and acquired in advance in a form of images and texts;
the data protection module is used for protecting the safety of the face information and the iris information, and comprises a data encryption module and an identity authentication module, wherein the data encryption module is used for encrypting the face information and the iris information through an asymmetric encryption algorithm, the identity authentication module is used for authenticating and authorizing the identity of a visitor in a dynamic password mode, and the data encryption module and the identity authentication module independently work in parallel;
The output end of the camera is connected with the input end of the data protection module, the output end of the data protection module is connected with the input end of the data storage module, the output end of the data storage module is connected with the input ends of the face recognition module and the interface display module, the output end of the face recognition module is connected with the input end of the comparison search module, the output end of the comparison search module is connected with the input end of the alarm and recording module, and the output end of the alarm and recording module is connected with the input end of the interface display module.
2. The security face comparison search system of claim 1, wherein: the face detection algorithm processes the face picture through the Haar cascade classifier to realize the identification of a face region and the extraction of face features, and the workflow of the face detection algorithm comprises the following steps:
step one, collecting images;
the face detection algorithm receives face pictures acquired by a camera through an API interface and divides the face pictures into positive samples and negative samples, wherein the positive samples are pictures containing faces, and the negative samples are pictures not containing faces;
Step two, haar characteristic value extraction;
the face detection algorithm performs denoising, enhancing and normalizing operations on the face picture through an image processing function, and obtains a Haar characteristic value through a Haar characteristic value calculation formula and a Haar characteristic template, wherein the Haar characteristic value calculation formula is as follows:
in equation (1), S (x, y) is the Haar eigenvalue at position (x, y), w i Is the weight of rectangular region i, Σf (x i ,y i ) Is the gray sum of pixel points in a black rectangular area, sigma g (x i ,y i ) The gray sums of pixel points in the white rectangular area are obtained, i is the ordering of the rectangular areas, and n is the number of the rectangular areas;
training a classifier;
the face detection algorithm trains the classifier by using extracted Haar features and an adaptive lifting method, effective features are learned from the existing Haar features, the adaptive lifting method trains the weak classifier to construct a strong classifier based on the weight of the adaptive adjustment Haar features, and the calculation formula of the adaptive lifting method is as follows:
in the formula (2), sign is a de-sign function, M is the number of adaptive iterations, M is the iterative ordering, G m (x) For the m-th iteration weak classifier classification result, N is the number of Haar features, k is the sequence of Haar features, For the mth iteration kthWeighting coefficients of the Haar features, I (G m (x k ) Classifying the correct function, gamma, for the kth Haar feature of the mth iteration m Regularization term for the mth iteration;
step four, detecting a face area;
the face detection algorithm adopts a cascading mode to connect the strong classifiers to form a cascading classifier, the cascading classifier eliminates the area which is not a face through the primary classifier, the detection rate of the area which is not a face is reduced, the face detection algorithm carries out image scanning on the input face picture to be detected through a sliding window mode, and the face area is found out;
step five, extracting face feature vectors;
the face detection algorithm processes the well positioned face area by adopting a deep learning method, extracts face features, and normalizes the face features by adopting an Euclidean distance method to obtain face feature vectors.
3. The security face comparison search system of claim 1, wherein: the iris detection algorithm is based on a two-dimensional Gabor filter to decompose an original iris image into components, the iris detection algorithm adopts the real part and the imaginary part of the two-dimensional Gabor filter to describe the iris, and carries out polarity quantization on a filtering result to obtain an iris feature code, and the calculation formula of the iris detection algorithm is as follows:
In the formula (3), H (x, y) is the output result of the two-dimensional Gabor filter, alpha is the effective width of the two-dimensional Gabor filter, beta is the effective length of the two-dimensional Gabor filter, u 0 And v 0 Determining the frequency and direction of the modulation term, (x) 0 ,y 0 ) For the position in the iris image, when the real part and the imaginary part of the output result are both positive, the polarity quantization value is 11; when the real part is positive and the imaginary part is negative, the polarity quantization value is 10; the real part is positive, the negative imaginary part is positive, and the polarity quantization value is 01;and when the real part and the imaginary part are both negative, the polarity quantization value is 00, so that the iris feature code is obtained.
4. The security face comparison search system of claim 1, wherein: the workflow of the search matching algorithm comprises the following steps:
step 1, collecting face characteristic information;
the retrieval matching algorithm receives the feature vector of the face or the iris from the face recognition module through an API interface;
step 2, searching a database;
the retrieval matching algorithm calculates the distance between the feature vector of the face or iris to be compared and the feature vector of the face or iris acquired in advance through a Euclidean distance method, and determines the face or iris similar to the face or iris to be compared, wherein the Euclidean distance method has the following formula:
In the formula (4), L is the distance of the feature vector, δ j A, a is a weight coefficient of the j-th element of the face or iris characteristic vector to be compared j For the j-th element of the face or iris characteristic vector to be compared, sigma j B, weighting coefficient of jth element of face or iris characteristic vector acquired in advance j Z is the number of elements of the face or iris feature vector, j is the sequence of the elements of the face or iris feature vector;
step 3, similarity evaluation;
the search matching algorithm calculates the similarity between the face or iris to be compared and the searched face or iris through a cosine similarity method, the cosine similarity method removes the influence of the dimensional difference between vectors on the result through L2 norm normalization, and the cosine similarity method has the following formula:
in the formula (5), D is the similarity between the face or iris to be compared and the retrieved face or iris, A is the feature vector of the face or iris to be compared, and B is the feature vector of the retrieved face or iris;
step 4, judging matching;
the search matching algorithm compares the calculated similarity with a set similarity threshold value through a comparison method, when the similarity is larger than the threshold value, the matching is successful, when the similarity is smaller than the threshold value, the matching is failed, and an alarm signal is output according to the matching result and the search result is compared.
5. The security face comparison search system of claim 1, wherein: the receiving port receives the alarm signal of the incoming word comparison search module and the comparison search result through a wireless communication protocol, the alarm module sends an alarm short message to a user through a front-end server to achieve the alarm function when the comparison search fails, and the recording module generates a log file for storing the comparison search result through log recording to achieve the recording of the comparison search result.
6. The security face comparison search system of claim 1, wherein: the PC application program comprises a connection module, a data reading module, a data display module and a data management module, wherein the connection module establishes a connection between the PC application program and a database through a wireless communication protocol, the data reading module acquires face information, iris information and comparison search result data through an inquiry API interface, the data display module displays the face information, the iris information and the comparison search result data on the interface in an image and text mode, and the data management module regularly executes backup and deletion operations on the face information, the iris information and the comparison search result through a timing task.
7. The security face comparison search system of claim 1, wherein: the asymmetric encryption algorithm performs encryption operation based on an RSA algorithm, the asymmetric encryption algorithm processes face information and iris information through a public key and a private key to achieve encryption and decryption of the face information and the iris information, the dynamic password performs identity authentication operation through an identity authentication platform, the identity authentication platform comprises a dynamic password generation module, a visitor interface and a verification authorization module, the dynamic password generation module generates and stores a dynamic password based on a dynamic password algorithm of time and sends the dynamic password to a mobile phone of a visitor in a short message mode, the visitor interface achieves user input of the dynamic password through a touch screen, and the verification authorization module compares the dynamic password input by a user with the dynamic password stored by the identity authentication platform through a comparison circuit and authorizes the visitor according to an identity authentication result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311808254.1A CN117809348A (en) | 2023-12-26 | 2023-12-26 | Security face comparison search system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311808254.1A CN117809348A (en) | 2023-12-26 | 2023-12-26 | Security face comparison search system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117809348A true CN117809348A (en) | 2024-04-02 |
Family
ID=90424475
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311808254.1A Pending CN117809348A (en) | 2023-12-26 | 2023-12-26 | Security face comparison search system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117809348A (en) |
-
2023
- 2023-12-26 CN CN202311808254.1A patent/CN117809348A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11288504B2 (en) | Iris liveness detection for mobile devices | |
| Joseph et al. | RETRACTED ARTICLE: A multimodal biometric authentication scheme based on feature fusion for improving security in cloud environment | |
| JP7060619B2 (en) | Biometric identification system and method | |
| US8542886B2 (en) | System for secure face identification (SCIFI) and methods useful in conjunction therewith | |
| Chakraborty et al. | An overview of face liveness detection | |
| US11651623B2 (en) | Methods and apparatus for outdoor access control using biometric verification | |
| Smith-Creasey et al. | Continuous face authentication scheme for mobile devices with tracking and liveness detection | |
| Babaei et al. | Face Recognition Application for Automatic Teller Machines (ATM) | |
| CN104836991A (en) | Camera with privacy protection function | |
| EP3837634A1 (en) | Methods and apparatus for facial recognition | |
| KR20140089810A (en) | CCTV Environment based Security Management System for Face Recognition | |
| CN108491785A (en) | A kind of artificial intelligence image identification attack defending system | |
| Agarwal et al. | Iris sensor identification in multi-camera environment | |
| CN113095212A (en) | Face recognition method and system based on local sorting algorithm encryption | |
| US11017212B2 (en) | Methods and apparatus for biometric verification | |
| CN108512651A (en) | A kind of artificial intelligence image identification attack defense method, system and storage medium | |
| CN117809348A (en) | Security face comparison search system | |
| Anika et al. | Multi image retrieval for kernel-based automated intruder detection | |
| CN109450878B (en) | Biological feature recognition method, device and system | |
| CN113190858A (en) | Image processing method, system, medium and device based on privacy protection | |
| CN112418179A (en) | Face recognition and collection system based on public place and privacy protection method | |
| CN110598531A (en) | Method and system for recognizing electronic seal based on face of mobile terminal | |
| Shankar et al. | Steganalysis of minor embedded JPEG image in transform and spatial domain system using SVM-PSO | |
| CN109993140B (en) | On-site tattoo evidence obtaining handheld device, system and control method | |
| US11017214B1 (en) | Methods and apparatus for biometric verification |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination |