KR102596462B1 - Access control authentication system and method capable of measuring heart rate using multi-modal sensor - Google Patents

Abstract

The access control authentication system capable of measuring heart rate using a multi-modal sensor according to an embodiment of the present invention is capable of acquiring facial images for each user requiring facial recognition in the state of wearing and not wearing the accessory, , video acquisition means that transmits the corresponding videos to the authentication server so that they can be used for authentication; And through a communication network, images of individuals requiring face recognition are provided from image acquisition means in both the state of wearing and not wearing accessories, the recognition elements necessary for face authentication are extracted from the facial area in the image, and based on the extracted recognition elements It includes an authentication server that performs authentication, but in the case where accessories are not worn, an additional face side image is taken and provided, and the recognition elements are extracted and used for face authentication, and a radar sensor located in close proximity to the entrant is used to identify the entrant. It is characterized by being able to recognize breathing and estimate heart rate.

Description

Access control authentication system and method capable of measuring heart rate using multi-modal sensor}

The present invention relates to an access control authentication system and method, and more specifically, to enable recognition and access authentication by extracting recognition elements for a limited area of the face even while wearing accessories, and to check the health status of the entrant. This relates to an access control authentication system and method capable of measuring heart rate using a multi-modal sensor capable of measuring heart rate.

In general, access control that requires control of access in buildings or offices within buildings is achieved through access managers, access card reading, and biometrics such as fingerprints or iris. The access control method by access managers requires visitors to individually collect access information. There is a very cumbersome problem because it has to be recorded.

When people look at photos or videos, they can easily recognize people, objects, scenes, and visual details. The goal of object recognition technology is to teach computers to do things that humans can naturally do, such as the ability to understand what is contained in an image.

In other words, object recognition, which allows computers to analyze and interpret visual information on behalf of humans, which receives the most information, is a computer vision technology that identifies objects in images or videos, and is calculated through machine learning algorithms such as deep learning. It is a core technology that becomes

In particular, object recognition using recent machine learning algorithms is a technology that is being used in various fields such as video surveillance, facial recognition, robot control, Internet of Things (IoT), autonomous driving, manufacturing, and security.

However, most facial recognition technologies extract feature points such as the eyes, nose, and mouth of the face and perform facial recognition based on the extracted feature points. In cases where infectious diseases are prevalent, such as the recent spread of the coronavirus, wearing accessories is mandatory. , Face recognition is difficult when wearing such accessories, making existing facial recognition technology using facial data when not wearing accessories useless.

On the other hand, if the visitor requires authentication and can estimate the health status of the visitor, the safety of the visitor can be guaranteed. For this purpose, if a heart rate estimation algorithm using various sensors is applied, there will be an advantage of being able to check the health status every time one enters and exits, and in recent Internet environments, multi-modal (Multi-modal), a human-centered interface technology that communicates with each other through various input elements such as sensor information, will be used. Modal) With the development of technology, its value is increasing.

Therefore, in order to solve the above-mentioned problem, additional recognition elements and feature points such as ears and forehead are extracted to facilitate face recognition even while wearing accessories, and technologies such as side/front image synthesis are used to facilitate face recognition. There is a need for research on an access control authentication system and method that can measure heart rate using a multi-modal sensor that enables safe access authentication.

Republic of Korea Patent No. 10-1946756 (registered on January 31, 2019)

The purpose of the present invention is to extract recognition elements such as ears and forehead to facilitate face recognition even while wearing accessories, and to measure heart rate using a multi-modal sensor that can perform face recognition and access authentication based on the extracted recognition elements. The goal is to provide a measurable access control authentication system and method.

Other purposes are to perform face recognition and authenticate entry and exit by using technologies such as side/frontal images of the face or image synthesis according to whether or not accessories are worn, and to recognize breathing and heart rate by collecting transmission and reception signals using radio waves from a radar sensor. The aim is to provide an access control authentication system and method capable of measuring heart rate using a multi-modal sensor capable of measurement.

The access control authentication system capable of measuring heart rate using a multi-modal sensor according to an embodiment of the present invention is capable of acquiring facial images for each user requiring facial recognition in the state of wearing and not wearing the accessory, , an image acquisition means that transmits the corresponding images to the authentication server and provides them so that they can be used for authentication; Authentication that receives images of individuals requiring face recognition from an image acquisition means through a communication network, both wearing and not wearing accessories, extracts the recognition elements necessary for face authentication, and performs authentication based on the extracted recognition elements. Including a server, the authentication server receives the face side image that is additionally acquired when acquiring the face image, extracts recognition elements, and uses it for face authentication, and monitors the visitor's breathing using a radar sensor located close to the visitor. It is characterized by being able to recognize and estimate heart rate.

The authentication server includes an extraction unit that extracts the recognition elements necessary for face recognition and authentication from the personal image provided from the image acquisition means; An authentication unit that performs face authentication of an individual using data previously stored in a database to recognize the recognition elements and face images extracted from the extraction unit; And it further includes a database that stores face data including recognition elements to be used for authentication of access users for each individual, and provides the authentication unit upon request for authentication.

The recognition element includes features of facial parts, unique features, and distances between parts, and is at least one of the distance between the eyes and nose, the size and shape of the mouth, the unique features of the ears, and the unique features of the forehead. It is characterized by including.

When extracting recognition elements from a face image while wearing an accessory, the extraction unit extracts additional recognition elements from the face side image and the front face image. When extracting additional recognition elements, the distance between the eyes and ears, ear size, earlobe, and ear are extracted. It is characterized by additional extraction of at least one of shape, forehead wrinkle condition, forehead size, and forehead shape.

The authentication server synthesizes the side and front images of the individual's face image among the individual face images obtained from the image acquisition means and uses them for authentication, or combines the face image with the individual wearing the accessory and the face image without the accessory. A synthesis unit that synthesizes and uses it for authentication; and a learning unit that performs learning to improve the recognition rate when authenticating an individual by using a machine learning algorithm based on face data obtained for each individual.

In the above, the radar sensor uses a continuous wave-based radar sensor, and the continuous wave-based radar sensor transmits and receives continuous waves to recognize changes in human body movement due to biological signals, collects the transmitted and received signals, and uses frequency, time, and phase differences. It is characterized by being able to estimate heart rate by recognizing breathing, which is biometric information.

In the access control authentication method capable of measuring heart rate using a multi-modal sensor using the authentication server according to an embodiment of the present invention, the image acquisition means is obtained by taking a facial image for each individual when wearing an accessory and when not wearing an accessory. steps; The authentication server extracts recognition elements from facial images received from the image acquisition means when wearing an accessory and when not wearing an accessory; Storing and registering face data images for each individual, including extracted recognition elements, in a database; and performing personal authentication by matching the facial data image stored and registered in the database with the newly acquired image.

The authentication server acquires facial images from the front and side when the accessory is worn from the image acquisition means, respectively, and performs synthesis of the front and side images using a synthesis algorithm; registering the synthesized image data in a database to perform access authentication of a user with the corresponding face; And upon access authentication, it further includes performing additional secondary authentication after primary face authentication by the access authentication.

The present invention additionally extracts recognition elements such as ears and forehead to facilitate face recognition even when wearing accessories, and utilizes technologies such as side/front image synthesis to enable face recognition even when accessories such as masks are not worn. It has the advantage of preventing the spread of infectious diseases and helping to strengthen security.

In addition, the image synthesis algorithm is used to synthesize facial side/frontal images or images according to the presence or absence of accessories for each individual, making it easy to identify and recognize individuals regardless of whether they are wearing accessories, and furthermore, enabling safe access authentication. there is.

In addition, as synthetic images are stored and managed in a database, the number of face data required for individual recognition is reduced, data storage space is secured, and DB management is easy even with a small capacity.

Figure 1 is a block diagram showing the configuration of an access control authentication system capable of measuring heart rate using a multi-modal sensor according to an embodiment of the present invention.
FIG. 2 is a block diagram showing the internal configuration of the authentication server of FIG. 1 in detail.
Figure 3 is a flowchart showing the extraction and authentication process of recognition elements from a face image in an access control authentication method capable of measuring heart rate using a multi-modal sensor according to an embodiment of the present invention.
Figure 4 is a flowchart showing an image synthesis and additional authentication process for authentication of a face image in an access control authentication method capable of measuring heart rate using a multi-modal sensor according to an embodiment of the present invention.

Hereinafter, specific embodiments of the present invention will be described in detail with reference to the drawings. However, the spirit of the present invention is not limited to the presented embodiments, and those skilled in the art who understand the spirit of the present invention may add, change, or delete other components within the scope of the same spirit, or create other degenerative inventions or this invention. Other embodiments that are included within the scope of the invention can be easily proposed, but this will also be said to be included within the scope of the invention of the present application. In addition, components having the same function within the scope of the same idea shown in the drawings of each embodiment will be described using the same reference numerals.

Figure 1 is a block diagram showing the configuration of an access control authentication system capable of measuring heart rate using a multi-modal sensor according to an embodiment of the present invention, and Figure 2 is a block diagram showing the internal structure of the authentication server of Figure 1 in detail. .

As shown in Figure 1, the access control authentication system capable of measuring heart rate using a multi-modal sensor of the present invention includes an authentication server (connected to an image acquisition means 100 and an access control device 400 and a communication network 200) 300).

The image acquisition means 100 can acquire facial images for each user requiring face recognition in a state of wearing and not wearing an accessory, and transmit the corresponding images to the authentication server 300 to use them for authentication. It is provided so that it can be done.

Here, accessories may include, for example, masks, glasses, sunglasses, various accessories, face shields, etc.

Additionally, the image acquisition means 100 may be a camera with a built-in object recognition algorithm. Furthermore, the image acquisition means 100 builds a database 360 to store images for each user for authentication on behalf of the authentication server 300, extracts recognition elements for personal identification from the images, and Based on this, the entire process of face authentication can also be performed.

In addition, the image acquisition means 100 uses AdaBoost, Support Vector Machine (SVM), Linear Disciminant Analysis (LDA), and Principal Component Analysis to increase the recognition rate when shooting with a camera. Algorithms such as Analusis (PCA) may be built-in.

These algorithmic techniques all identify the recognition target area based on appearance, detect the area around the face using a model trained by a set of captured images to be used for training, and overcome various surrounding constraints through training. As a result, face recognition accuracy and reliability can be increased, and recognition elements can be better extracted through the authentication server 300 and used for authentication based on this.

The access control device 400 is a device for allowing access only to authenticated users after performing face authentication from the authentication server 300. It may be provided to include an image acquisition means 100, and may be equipped with a gate for access control, etc. may be included.

The authentication server 300 receives images of individuals requiring face recognition from the image acquisition means 100 through the communication network 200 in both states wearing and without accessories, extracts and extracts recognition element data required for face authentication. Authentication is performed based on the recognized recognition elements, and the authentication results are transmitted to the access control device 400 so that access can be determined.

In particular, the authentication server 300 extracts additional recognition elements mainly from the ears and forehead among the parts of the face that do not cover the accessories to facilitate recognition and authentication even when no accessories are worn, and can also receive side images of the face to increase the recognition rate. , Enables user authentication to be performed more safely.

To this end, as shown in FIG. 2, the authentication server 300 includes a communication unit 310, an extraction unit 320, an authentication unit 330, a synthesis unit 340, a learning unit 350, and a database 360. Includes more.

The communication unit 310 provides a compatible communication protocol for transmitting and receiving data with the image acquisition means 100 or the access control device 400 through the communication network 200.

The extraction unit 320 performs a function of extracting a plurality of recognition elements necessary for face recognition and authentication from the personal image provided by the image acquisition means 100.

Here, the recognition element may be a feature point of a facial part, a unique characteristic element, or the distance between parts. For example, the distance between the eyes and nose, the size and shape of the mouth, the size of the ears, the unique characteristics of the ears, the unique characteristics of the forehead, etc. Multiple elements can be extracted for each element, and the more the number of extractions, the better the individual recognition rate. there is.

In addition, when wearing accessories, face recognition elements are limited and the exposed area is limited, but the ears and forehead are always exposed, so feature points and unique feature elements centered on the ears and forehead can be additionally extracted and used for recognition and authentication. For example, recognition elements allow additional extraction of the distance between the eyes and ears, the size of the ears, the unique features of the ears (ear lobes, ear shape, etc.), and the unique features of the forehead (state of forehead wrinkles, forehead size, forehead shape, etc.) .

Furthermore, the extraction unit 320 can receive images taken from multiple angles when necessary, in addition to the frontal face image and the face side image, and can extract additional recognition elements for a specific part that is well recognized at a specific angle from the image. As a result, the more recognition elements to extract, the easier individual recognition and identification becomes.

The authentication unit 330 may perform face authentication of an individual using data previously stored in the database 360 for individual recognition from the extracted recognition elements and face image data.

Specifically, for example, face data including pre-stored recognition elements and face data newly captured from the image acquisition means 100 upon entry may be provided, and personal authentication may be performed based on whether a preset face match value is exceeded through mutual matching.

Furthermore, the authentication unit 330 may perform additional authentication using biometric information to more safely perform facial authentication without or wearing an accessory.

For example, access authentication can be performed by additionally recognizing changes in facial expression recognition results and changes between facial and hand feature points.

Additionally, the health status of the visitor can be estimated from the face image.

Specifically, the image acquisition means 100 extracts and checks the heart rate from the image captured using a multi-modal sensor using an image sensor, a radar sensor, a heart rate sensor, a temperature sensor, etc., and can check the status of the person entering the room, The authentication server 300 may receive the extracted heart rate through wired or wireless communication and identify a series of emotions such as anger, contempt, disgust, fear, happiness, expressionless expression, sadness, and surprise in the face image obtained from the image sensor. .

As a specific example, using a continuous-wave based radar sensor located nearby, the breathing and heart rate of the visitor can be recognized, the heart rate can be estimated, and the health status of the visitor can be estimated using this.

Here, the radar sensor can acquire biometric information by recognizing differences in frequency, time, phase, and distance of transmitted and received signals generated by distant objects.

In other words, a continuous wave-based radar sensor transmits and receives continuous waves to recognize changes in human body movement due to biological signals, and collects transmitted and received signals to recognize breathing rate and heart rate, which are biometric information, using differences in frequency, time, phase, etc., and detects heart rate. It can be estimated.

At this time, a circularly polarized antenna was used to minimize the influence of the direction of movement of bio-signals, and a configuration that compares the signal generated by the voltage-controlled oscillator in the frequency mixer may be added to reduce power consumption and simplify system configuration.

As another method, the authentication unit 330 can estimate changes in blood flow on the visitor's face by analyzing time series data and images of the visitor's image acquired through an image sensor, which also makes it possible to estimate the heart rate. .

Furthermore, the skin area can be extracted by determining the skin area based on the Hue value in the pixel value of the captured image or the HSV (Hue Saturation value) model, and the change in G (Green) channel in the RGB model of the extracted skin area. The heart rate signal can be detected from.

Here, the HSV model means expressing the color of an image with three components: Hue, Saturation, and Value.

Additionally, if the captured skin image is an RGB model, it can be converted to an HSV model, and only the skin area can be extracted by forming a mask with an area other than the skin area in the captured skin image and masking the captured skin image. there is.

By taking the cumulative average for all frames of one skin area, the heart rate signal can be extracted as one value for a specific frame. At this time, if the number of frames of the video is set to 30 per second, the heart rate signal can be extracted as a 30Hz signal, for example.

In addition, to remove noise contained in the extracted heart rate signal, a band-pass filter (only signals in a specific frequency band pass through) can be used, and it can be configured as an RLC circuit, and can be composed of a combination of a low-pass filter and a high-pass filter. Alternatively, an adaptive filter (Kalman filter) can be used to make it less sensitive to lighting changes and motion noise.

In addition, stress and concentration can be measured using the heart rate previously extracted based on the health status of the visitor.

In addition, the image acquisition means 100 or the access control device 400 may further include a notification unit that notifies the occurrence of an abnormality so that the visitor can check the health status if the measured stress and concentration are higher than the set value.

In addition, methods such as already known deep learning neural network algorithms may be additionally used to estimate the health status of visitors.

Furthermore, additional authentication of the visitor may be performed by checking heart rate using a health management algorithm.

For example, additional authentication can be performed depending on whether the heart rate measurement matches the heart rate of the previously registered visitor.

Since authentication is performed in addition to authentication using facial recognition elements through this additional authentication method, it becomes less sensitive to facial recognition errors, etc., and security is further strengthened, allowing personal authentication to be performed safely upon entry.

The synthesis unit 340 may synthesize the side and front images of an individual's face image among the acquired individual face images and use them for authentication. In particular, when wearing accessories, recognition elements of the ear can be extracted better from a side image than from a frontal face image. In addition, recognition elements such as the curvature of the forehead or the shape of the ear can be more easily extracted through the side. Therefore, a composite image of the side and front images of an individual's face image can be generated through a synthesis algorithm, and the generated composite image is stored in the database 360 so that it can be used for matching during face authentication.

In addition, facial data when wearing an accessory and facial data when not wearing an accessory are combined, stored and managed as one data, and can be used for authentication, which has the advantage of making individual data management easier.

Furthermore, even if a plurality of face data is extracted, the synthesis unit 340 is used to restore images in cases where the face data image required for face authentication is not clear due to limitations in camera performance or camera errors, or to determine forgery of face data images. For this purpose, a neural network learning algorithm may be used.

In particular, by utilizing generative adversarial networks, new synthetic images or fake images can be created or regenerated, which can be used for face authentication using synthetic images, damaged image restoration, or fake image learning. .

Unlike existing deep learning networks, this generative adversarial network is composed of multiple deep neural networks. Although it requires tens of times more computation than existing deep neural network models to generate high-resolution images, it has excellent performance in image creation and restoration. can be provided. In addition, the generative adversarial network is an unsupervised learning-based generative model that adversarially trains two networks with a generator and a discriminator. Input data is input to the generator to create fake images similar to real images. It can be learned to do so. The input data may be a noise value. Noise values may follow any probability distribution. For example, it may be data generated by Zero-Mean Gaussian.

The discriminator can learn to distinguish between real images and fake images generated by the generator. More specifically, it can learn to have a high probability when inputting a real image, and to have a low probability when inputting a fake image. In other words, the discriminator can gradually learn to distinguish between real and fake images.

The learning unit 350 can use machine learning algorithms based on face data to better recognize the face and perform learning to improve recognition rate and authentication speed.

In other words, when using a machine learning algorithm, when applying machine learning techniques to recognition elements extracted from facial data, the number of data about people increases and as more data is accumulated, the number of machine learning training increases, and as a result, The accuracy of modeling obtained through training gradually increases.

This characteristic is that, while analysis techniques through specific mathematics or statistical modeling always have a certain level of prediction error rate, machine learning can gradually improve the prediction error rate in identifying the same person as the number of training increases. This is a big advantage of modeling through techniques.

As a result, it can be used to improve the accuracy of identifying the same person by accumulating face data over a long period of time using the above-described machine learning technique. For such machine learning algorithms, at least one well-known neural network learning algorithm such as SVM, CNN, or RNN can be used.

Furthermore, the learning unit 350 can be used to determine whether the image is forged through machine learning when a fake image is generated by a generative adversarial network algorithm.

The database 360 stores face data including recognition elements to be used for authentication of access users for each individual, and provides it to the authentication unit 330 upon request for authentication. Additionally, the database 360 can store and manage synthetic images or images generated by the synthesis unit 340, fake images for learning, etc.

Furthermore, for protection of data managed in the database 360, encryption/decryption technology can be applied to face data and data transmitted and received between terminals.

More specifically, identification information capable of identification is given to the image acquisition means 100, the authentication server 300, and the access control device 400 corresponding to each terminal, respectively, so that the identification information of each terminal is provided. A lightweight encryption algorithm is performed using as a private key.

The lightweight encryption algorithm is an encryption technology designed to be implemented in restricted environments such as smart devices, and includes HIGHT (HIGh security and light weigHT), a symmetric key encryption algorithm, Lightweight Encryption (LEA), and LSH (Lightweight Secure Hash), a hash function. You can utilize it.

By using this lightweight encryption algorithm to encrypt/decrypt face data, transmitted and received data, etc., external leakage of the data or illegal control due to external hacking can be prevented. It is desirable for the lightweight cryptographic algorithm to use a lightweight hash function suitable for this embedded computing environment.

A lightweight hash function is a hash function that consumes relatively low computing power and is designed to ensure the integrity of transmitted or received data, excluding some features that require high computing power in standard cryptographic hash algorithms such as SHA-3. (one-way function).

More specifically, among these lightweight hash functions, it is desirable to use the Sponge algorithm, which allows permutation of data without a key.

More specifically, the sponge creates the original message (here, the original data of the random key) to a certain size (padding) and then converts it into a specific standard size (e.g., the original message divided into a specific bit size) that only the creator of the key knows. After splitting into multiple pieces, random data is exchanged at the rear end of the split data (segmented original message) using several update functions, and on the other side, it is implemented to decrypt using an already known standard size.

In other words, by using this lightweight hash function, the security of the hash function is secured while relatively less computing power is required than when using a general hash function, resulting in less power consumption of the terminals and longer use. there is.

Figure 3 is a flowchart showing the extraction and authentication process of recognition elements from a face image in an access control authentication method capable of measuring heart rate using a multi-modal sensor according to an embodiment of the present invention.

First, the image acquisition means 100 captures a face image of each individual when wearing an accessory and when not wearing an accessory (S100).

Additionally, when wearing an accessory, a side image can be captured to extract additional recognition elements.

The image acquisition means 100 transmits facial images when wearing the accessory and when not wearing the accessory to the authentication server 300 through the communication network 200.

The authentication server 300 extracts recognition elements from facial images when the accessory is worn and when the accessory is not worn (S102, S104).

Each individual's face data image, including the extracted recognition elements, is stored and registered in the database 360 (S106).

Personal authentication is performed by matching the stored and registered facial data image with the newly acquired image (S108).

Figure 4 is a flowchart showing an image synthesis and additional authentication process for authentication of a face image in an access control authentication method capable of measuring heart rate using a multi-modal sensor according to an embodiment of the present invention.

In one embodiment of the present invention, as facial images are obtained from the front and the side when wearing an accessory, synthesis can be performed using the generative adversarial network algorithm described above for the front and side images (S114).

The synthesized image data is registered in the database 360 and used when authenticating the access of a user with the corresponding face (S116, S118).

Additionally, in order to strengthen security during access authentication, if the primary face authentication fails, or even if it succeeds, it can be robust against incorrect authentication due to recognition errors, or additional secondary authentication can be performed in duplicate for the purpose of strengthening security. There is (S120).

Secondary authentication can register an individual's biometric information (heart rate) in advance, perform authentication based on whether the biometric information matches, and check the individual's health status, so it can be used for various purposes.

In this specification, a ‘terminal’ may be a wireless communication device that guarantees portability and mobility, and may be any type of handheld-based wireless communication device such as a smartphone, tablet PC, or laptop, for example. Additionally, the ‘terminal’ may be a wired communication device such as a PC that can connect to another terminal or server through a network.

In addition, the communication network 200 refers to a connection structure that allows information exchange between nodes such as terminals and servers, including a local area network (LAN), a wide area network (WAN), Includes the Internet (WWW: World Wide Web), wired and wireless data communication networks, telephone networks, and wired and wireless television communication networks (200).

Examples of the wireless data communication network 200 include 3G, 4G, 5G, 3rd Generation Partnership Project (3GPP), Long Term Evolution (LTE), World Interoperability for Microwave Access (WIMAX), Wi-Fi, Bluetooth communication, and infrared. It includes, but is not limited to, communication, ultrasonic communication, Visible Light Communication (VLC), LiFi, etc.

100 ; Video acquisition means
200 ; communications network
300 ; Authentication server
310 ; Ministry of Communications
320 ; extraction part
330 ; Certification Department
340 ; synthesis part
350 ; learning department
360 ; database
400 ; access control device

Claims (8)

An image acquisition means capable of acquiring facial images for each user requiring face recognition in a state of wearing and not wearing an accessory, and transmitting the acquired images to an authentication server; and
Through a communication network, images of individuals requiring face recognition are provided from the image acquisition means in both the state of wearing and not wearing the accessory, the recognition elements necessary for face authentication are extracted from the facial area in the image, and based on the extracted recognition elements Including the authentication server that performs authentication,
The authentication server receives the face side image that is additionally acquired when acquiring the face image, extracts the recognition element, and uses it for face authentication.
Using a radar sensor located close to the visitor, the visitor's breathing is recognized and the heart rate is estimated.
The authentication server is,
an extraction unit that performs the function of extracting the recognition elements necessary for face recognition and authentication from the personal image provided from the image acquisition means;
An authentication unit that performs face authentication of an individual using data previously stored in a database for facial image recognition and recognition elements extracted from the extraction unit; and
A database that stores face data including recognition elements to be used for authentication of access users for each individual, and provides the authentication unit upon request for authentication,
The image acquisition means is,
To increase the recognition rate of camera shots, use one of AdaBoost, Support Vector Machine (SVM), Linear Disciminant Analysis (LDA), and Principal Component Analysis (PCA) algorithms. It is built-in to identify the recognition target area based on its appearance.
The authentication department,
Additional authentication is performed using biometric information to more safely perform the facial authentication when the accessory is not worn or worn, and changes in facial expression recognition results and changes between facial and hand feature points are additionally recognized. Performing access authentication
Access control authentication system capable of measuring heart rate using a multi-modal sensor. delete According to paragraph 1,
The recognition element is
It includes features about facial features, unique features, and distances between facial parts, and includes at least one of the distance between the eyes and nose, the size and shape of the mouth, the unique features of the ears, and the unique features of the forehead. An access control authentication system capable of measuring heart rate using a multi-modal sensor. According to paragraph 3,
The extraction unit
When extracting recognition elements from a face image while wearing an accessory, recognition elements are additionally extracted from the face side image and the front face image.
When extracting additional recognition elements, heart rate is measured using a multi-modal sensor, which is characterized by additionally extracting at least one of the distance between the eyes and ears, ear size, earlobe, ear shape, forehead wrinkle condition, forehead size, and forehead shape. Possible access control authentication system. According to paragraph 1,
The authentication server is,
Among the individual face images obtained from the above image acquisition means, the face side image and the front face image of the individual face image are synthesized and used for authentication, or the face image with the individual wearing the accessory and the face image without the accessory are used for authentication. A synthesis unit that synthesizes and uses it for authentication; and
An access control authentication system capable of measuring heart rate using a multi-modal sensor that includes a learning unit that uses machine learning algorithms based on facial data acquired for each individual to improve the recognition rate when authenticating an individual. According to paragraph 1,
The radar sensor is
It uses a continuous wave-based radar sensor,
The continuous wave-based radar sensor transmits and receives continuous waves to recognize changes in human body movement due to biological signals, collects transmitted and received signals, and recognizes breathing, which is biometric information, using frequency, time, and phase differences to estimate heart rate. An access control authentication system that can measure heart rate using a multi-modal sensor. In the access control authentication method capable of measuring heart rate using a multi-modal sensor using the authentication server of any one of claims 1, 3 to 6,
The image acquisition means includes the steps of capturing and acquiring facial images of each individual when wearing an accessory and when not wearing an accessory;
The authentication server extracts recognition elements from facial images received from the image acquisition means when wearing an accessory and when not wearing an accessory;
Storing and registering face data images for each individual, including extracted recognition elements, in a database; and
An access control authentication method capable of measuring heart rate using a multi-modal sensor, comprising the step of performing personal authentication by matching a facial data image stored and registered in the database with a newly acquired image. In clause 7,
The authentication server acquires face images from the front of the face and the side of the face when the accessory is worn from the image acquisition means, respectively, and performs synthesis of the front and face images using a synthesis algorithm;
registering the synthesized image data in a database to perform access authentication of a user with the corresponding face; and
An access control authentication method capable of measuring heart rate using a multi-modal sensor, further comprising performing additional secondary authentication after primary face authentication by the access authentication.