KR102577276B1 - Palm information-based multi dimensional biometric authentification apparatus, and control, method thereof - Google Patents

Abstract

The present invention relates to a multidimensional biometric authentication device for effectively classifying (recognizing) a user's identity from a photographed image of a palm and an operation method thereof. The multidimensional biometric authentication device includes: a generating unit which processes each region of interest image predicted in different ranges from a photographed image of a palm to generate a multi-channel image composed of an index; and a classification unit which classifies the identity of a user matching the captured image based on feature points extracted from the multi-channel image.

Description

Multidimensional biometric authentication device based on palm information and its operation method {PALM INFORMATION-BASED MULTI DIMENSIONAL BIOMETRIC AUTHENTIFICATION APPARATUS, AND CONTROL, METHOD THEREOF}

The present invention relates to a method for effectively classifying (recognizing) a user's identity from a photographed image of the palm.

In modern society, when it is necessary to apply security systems such as mobile devices, desktops, security facilities, and security management systems, various user recognition methods are used to grant appropriate permissions to users.

However, passwords and resident registration numbers are at high risk of being leaked and security is not guaranteed, so two-factor authentication is often used.

In the case of double recognition, various alternatives emerged, such as entering a secondary password or recognizing a specially designed USB.

However, this also carries the risk of loss and duplication, and has the disadvantage of being leakable.

As an alternative to this, a user recognition method using biometric information is used.

Biometrics, or biometrics, is an information security technology that identifies or authenticates individuals using the unique biometric information each person has. It has a low risk of loss and forgery/falsification, and provides high reliability and convenience. It's attracting attention.

There is physical biometric recognition in biometrics, and a representative example of such physical biometric recognition is an identity recognition method based on palm information.

In the case of identity recognition based on palm information, unlike the fingerprint recognition method, which is sensitive to wounds or whose accuracy varies greatly depending on the degree of contamination, biometric recognition is performed using the entire palm area, so wounds or contamination do not affect performance. It has advantages.

However, the vein recognition method is mainly applied to identity recognition based on palm information. This vein recognition method is not only expensive and complicated in the system installation process, but also uses only one vein as the basis information reflected in identity recognition. There is a limitation in that it lacks key information.

Accordingly, the present invention would like to propose a new method that can recognize the user's identity using only an inexpensive camera sensor or a mobile phone as an alternative to the gastric vein recognition method.

The present invention was created in consideration of the above-mentioned circumstances, and the goal to be achieved by the present invention is to effectively classify (recognize) user identity from a photographed image of the palm.

In order to achieve the above object, a multi-dimensional biometric authentication device according to an embodiment of the present invention generates a multi-channel image composed of an index by processing each region of interest image predicted in different ranges from a photographed image of the palm. generation unit; and a classification unit that classifies the identity of the user matching the captured image based on feature points extracted from the multi-channel image.

Specifically, the multi-channel image includes a first index consisting of an image in which the palm area including the fingers is set as the area of interest, a second index consisting of an image in which the palm area excluding the fingers is set as the area of interest, and the palm. It may include a third index composed of an image with the region set as the region of interest.

Specifically, the generator predicts the image of the first index from the captured image, and predicts the image of the second index from an image obtained by bitwise ANDing the image of the first index and the captured image. And, the image of the third index can be predicted from the image that emphasizes the curved part in the image of the second index.

Specifically, the generator may sequentially predict the image of each index in the multi-channel image through a previously learned segmentation model for each index.

Specifically, the index-specific segmentation model emphasizes the region of interest by lowering the weight of the remaining regions excluding the region of interest, and the weight is determined through an attention gate (at least one of skip connections and upsampling). Depending on the adoption of the attention gate structure, it may be determined based on the first feature point transmitted to a specific layer through skip connections and the second feature point transmitted from the previous layer neighboring the specific layer.

Specifically, the classification unit receives the multi-channel image through a pre-learned classification model and determines the identity of the user matching the captured image through an ensemble classifier that ensembles the feature point extraction results from the classification model. Can be classified.

A method of operating a multidimensional biometric authentication device according to an embodiment of the present invention to achieve the above object is a multi-channel image composed of an index by processing each region of interest image predicted in different ranges from a photographed image of the palm. A generation step of generating; And a classification step of classifying the identity of the user matching the captured image based on the feature points extracted from the multi-channel image.

Specifically, the multi-channel image includes a first index consisting of an image in which the palm area including the fingers is set as the area of interest, a second index consisting of an image in which the palm area excluding the fingers is set as the area of interest, and the palm. It may include a third index composed of an image with the region set as the region of interest.

Specifically, the generation step predicts the image of the first index from the captured image, and predicts the image of the second index from the image obtained by bitwise ANDing the image of the first index and the captured image. In addition, the image of the third index can be predicted from an image that emphasizes the curved portion in the image of the second index.

Specifically, in the generation step, the image of each index in the multi-channel image can be sequentially predicted through a pre-learned segmentation model for each index.

Specifically, the index-specific segmentation model emphasizes the region of interest by lowering the weight of the remaining regions excluding the region of interest, and the weight is determined through an attention gate (at least one of skip connections and upsampling). Depending on the adoption of the attention gate structure, it may be determined based on the first feature point transmitted to a specific layer through skip connections and the second feature point transmitted from the previous layer neighboring the specific layer.

Specifically, in the classification step, the multi-channel image is input through a pre-learned classification model, and the user's identity is matched with the captured image through an ensemble classifier that ensembles the feature point extraction results from the classification model. can be classified.

Accordingly, in the palm information-based multidimensional biometric authentication device and its operating method of the present invention, a multi-channel image is generated in which unnecessary pixels such as the background are removed in the data augmentation process of a photographed image of the palm, and a multi-channel image is generated within the multi-channel image. By classifying the user's identity based on the ensemble result of feature points for each index extracted from each dimension image, the accuracy of the identity classification (recognition) result can be greatly improved.

1 is an exemplary diagram illustrating an identity classification (recognition) environment based on palm information according to an embodiment of the present invention.
Figure 2 is a schematic configuration diagram of a multidimensional biometric authentication device according to an embodiment of the present invention.
Figure 3 is an example diagram for explaining a multi-channel image generation structure according to an embodiment of the present invention.
Figure 4 is an example diagram illustrating a user identity classification structure according to an embodiment of the present invention.
Figures 5 and 6 are exemplary diagrams for explaining a segmentation model according to an embodiment of the present invention.
Figure 7 is a flowchart illustrating a method of operating a multidimensional biometric authentication device according to an embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described with reference to the attached drawings.

In one embodiment of the present invention, technology related to biometric recognition based on palm information is addressed.

In modern society, when it is necessary to apply security systems such as mobile devices, desktops, security facilities, and security management systems, various user recognition methods are used to grant appropriate permissions to users.

However, passwords and resident registration numbers are at high risk of being leaked and security is not guaranteed, so two-factor authentication is often used.

In the case of double recognition, various alternatives emerged, such as entering a secondary password or recognizing a specially designed USB.

However, this also carries the risk of loss and duplication, and has the disadvantage of being leakable.

As an alternative to this, a user recognition method using biometric information is used.

Biometrics, or biometrics, is an information security technology that identifies or authenticates individuals using the unique biometric information each person has. It has a low risk of loss and forgery/falsification, and can provide high reliability and convenience. It is attracting attention in this regard.

There is physical biometric recognition in biometrics, and a representative example of such physical biometric recognition is an identity recognition method based on palm information.

In the case of identity recognition based on palm information, unlike the fingerprint recognition method, which is sensitive to wounds or whose accuracy varies greatly depending on the degree of contamination, biometric recognition is performed using the entire palm area, so wounds or contamination do not affect performance. It has advantages.

However, the vein recognition method is mainly applied to identity recognition based on palm information. In the case of this vein recognition method, not only is the price expensive and the system installation process complicated, but only one vein is used as the basis information reflected in identity recognition. Due to this, there is a limitation that key information is lacking.

Accordingly, in one embodiment of the present invention, as an alternative to the gastric vein recognition method, we would like to propose a new method that can recognize the user's identity using only an inexpensive camera sensor or a mobile phone.

In this regard, Figure 1 exemplarily shows an identity classification (recognition) environment based on palm information according to an embodiment of the present invention.

As shown in FIG. 1, the palm information-based identity classification (recognition) environment according to an embodiment of the present invention includes a biometric recognition device 100 that classifies (recognizes) the user's identity from a photographed image of the palm. It can have a configuration that does.

The biometric recognition device 100 generates a multi-channel image from which unnecessary pixels such as the background are removed during the data augmentation process of the captured image of the palm, and ensembles feature points for each index extracted from images for each dimension in the multi-channel image. The user's identity is classified based on the ensemble results.

The biometric recognition device 100 may be implemented, for example, in the form of a computing device (e.g., a mobile phone) equipped with software (e.g., an application), or a server that can be accessed through a wired or wireless communication network.

For reference, when the multidimensional biometric authentication device 100 is implemented as a computing device (e.g., a mobile phone), both capturing images of the palm and classifying the user's identity from these captured images are performed as one multidimensional biometric authentication device ( 100) can be done on its own.

On the other hand, when the multidimensional biometric authentication device 100 is implemented in the form of a server, the operation of classifying the user's identity from the captured image is performed in the multidimensional biometric authentication device 100, and the image capture of the palm is performed by the multidimensional biometric authentication device 100. This can be done using a camera sensor separate from the authentication device 100 or a mobile phone.

Meanwhile, when the multidimensional biometric authentication device 100 is implemented in the form of a server, for example, it may be implemented in the form of a web server, database server, proxy server, etc., and a network load balancing mechanism or service device may be implemented through the Internet or other One or more of a variety of software that allows operation on a network may be installed, and thus may be implemented as a computerized system.

As above, in the palm information-based identity classification (recognition) environment according to an embodiment of the present invention, user identity can be effectively classified (recognized) from a photographed image of the palm through the above-described configuration. Hereinafter, this will be realized. The configuration of the multidimensional biometric authentication device 100 for this purpose will be described in more detail.

Figure 2 shows a schematic configuration of a multidimensional biometric authentication device 100 according to an embodiment of the present invention.

As shown in FIG. 2, the multidimensional biometric authentication device 100 according to an embodiment of the present invention includes a generator 110 that generates a multi-channel image, and feature points extracted from the multi-channel image. It may have a configuration including a classification unit 120 that classifies (recognizes) the user's identity based on.

The entire structure or at least part of the multidimensional biometric authentication device 100, including the above generation unit 110 and classification unit 120, is implemented in the form of a hardware module or a software module, or a combination of a hardware module and a software module. It can be implemented as:

Here, the software module can be understood as, for example, an instruction executed by a processor that controls operations within the multidimensional biometric authentication device 100, and these instructions are stored in the memory of the multidimensional biometric authentication device 100. You can have it.

The multidimensional biometric authentication device 100 according to an embodiment of the present invention can effectively classify (recognize) the user's identity from a photographed image of the palm through the above-described configuration. Hereinafter, multidimensional biometric authentication to realize this will be described. We will continue with a more detailed description of each component within the device 100.

The generator 110 performs a function of generating a multi-channel image from a captured image.

Specifically, the generator 110 processes each region of interest image predicted in different ranges from a photographed image of the palm to generate a multi-channel image composed of an index.

At this time, the generator 110 predicts the image of the first index from the captured image through a pre-learned segmentation model for each index, predicts the image of the second index from the image of the first index, and predicts the image of the second index from the image of the first index. The image of the third index can be predicted from an image that emphasizes the curved portion in the index image.

Here, emphasis on the curved part within the two-dimensional image can be achieved, for example, through CLAHE (Contrast Limited Adaptive Histogram Equalization), which redistributes pixel values above a certain height by limiting the histogram height before calculation.

In other words, the generator 110 generates an image in which the palm area including the finger is set as the region of interest through a segmentation model for each index (Attention U-Net #1 to #3), for example, as shown in FIG. 3, and the finger. Predict an image in which the palm area excluded is set as the area of interest, and an image in which the palm area is set as the area of interest, and an image in which the palm area including the fingers is set as the area of interest is set as the first index, and the palm area excluding the fingers is predicted. A multi-channel image can be generated by configuring the image with the area of interest set as the second index, and the image with the palm line area set as the area of interest as the third index.

Based on this, in one embodiment of the present invention, attention is sequentially emphasized starting from the area of the palm containing the fingers through a multi-channel image consisting of the palm area including the fingers, the palm area excluding the fingers, and the palm area. can do.

That is, in the image of the first index (palm area containing fingers) in the multi-channel image, the overall outline or length characteristics such as the length of the fingers and the size of the palm that are most prominent are emphasized, and the image of the second index (the palm area containing the fingers) is emphasized. Excluding the palm area), the features of the palm and the palm area are emphasized, and the last component, the image of the third index of the palm area (palm area), can play a role in emphasizing the palm area.

Through this, in one embodiment of the present invention, when the artificial intelligence model simply extracts only the palm lines, even if the extraction of the palm lines is not successful, the information that the individual has in the area of the palm excluding the fingers and the area of the palm including the fingers It can be expected that the unique characteristics of the palm may have the advantage of helping with identity recognition.

Meanwhile, in relation to this, in order to support the above advantages, the previously learned index-specific segmentation model uses attention in at least one of Upsampling and Skip connections, which convert low-dimensional feature points into high-dimensional feature points, as shown in FIG. 4. By adopting an attention gate structure, the area of interest can be emphasized by lowering the weight of the remaining areas excluding the area of interest.

In this regard, the weight is based on the first feature point (x) transmitted to a specific layer through skip connections, and the second feature point (g) transmitted from the previous layer neighboring the specific layer, as shown in FIG. 5, for example. This can be decided.

Here, the first feature point (x) transmitted through skip connections has a lower feature point depiction ability than the second feature point (g) transmitted from the previous neighboring layer, but may include more spatial information.

The classification unit 120 performs the function of classifying (recognizing) the user's identity from multi-channel images.

More specifically, when a multi-channel image is generated from a captured image of the palm, the classification unit 120 classifies the identity of the user matching the captured image based on feature points for each index extracted from the multi-channel image.

At this time, the classification unit 120 receives the multi-channel image through a previously learned classification model, and determines the identity of the user matching the captured image through an ensemble classifier that ensembles the feature point extraction results from the classification model. Can be classified.

Here, classifying a user's identity can be understood as a process of predicting the user according to class and deriving the probability of belonging to the actual class.

In other words, for example, as shown in FIG. 6, the classification unit 120 inputs images for each dimension in the multi-channel image into pre-trained classification models (Pre-trained CNN Model #1 to #3), and According to the input of the star image, the feature points derived from the classification model (Pre-trained CNN Model #1~#3) are ensembled, and the identity of the user matching the captured image is classified from the ensembled feature points through a classifier. It is possible.

Here, at least one of voting, bagging, and boosting may be applied as the ensemble method.

As discussed above, according to the configuration of the multidimensional biometric authentication device 100 according to an embodiment of the present invention, a region of interest (ROI) is obtained through segmentation in relation to a photographed image of the palm. ) setting, extracting feature points using the region of interest, creating a multi-channel image for complex comparison, and using computer vision algorithms to accurately set the region of interest for only the palm and identify feature points such as palm lines even in a complex background. It can be seen that security and accuracy can also be improved and system installation costs can be minimized by extracting and reconstructing/generating data for identity recognition.

Hereinafter, a method of operating the multidimensional biometric authentication device 100 according to an embodiment of the present invention will be described with reference to FIG. 7.

First, the generator 110 processes each region of interest image predicted in different ranges from a photographed image of the palm to generate a multi-channel image composed of an index (S110-S120).

At this time, the generator 110 predicts the image of the first index from the captured image through a pre-learned segmentation model for each index, predicts the image of the second index from the image of the first index, and predicts the image of the second index from the image of the first index. The image of the third index can be predicted from an image that emphasizes the curved portion in the index image.

Here, emphasis on the curved part within the two-dimensional image can be achieved, for example, through CLAHE (Contrast Limited Adaptive Histogram Equalization), which redistributes pixel values above a certain height by limiting the histogram height before calculation.

In other words, the generator 110 generates an image in which the palm area including the finger is set as the region of interest through a segmentation model for each index (Attention U-Net #1 to #3), for example, as shown in FIG. 3, and the finger. Predict an image in which the palm area excluded is set as the area of interest, and an image in which the palm area is set as the area of interest, and an image in which the palm area including the fingers is set as the area of interest is set as the first index, and the palm area excluding the fingers is predicted. A multi-channel image can be generated by configuring the image with the area of interest set as the second index, and the image with the palm line area set as the area of interest as the third index.

Based on this, in one embodiment of the present invention, attention is sequentially emphasized starting from the area of the palm containing the fingers through a multi-channel image consisting of the palm area including the fingers, the palm area excluding the fingers, and the palm area. can do.

That is, in the image of the first index (palm area containing fingers) in the multi-channel image, the overall outline or length characteristics such as the length of the fingers and the size of the palm that are most prominent are emphasized, and the image of the second index (the palm area containing the fingers) is emphasized. Excluding the palm area), the features of the palm and the palm area are emphasized, and the last component, the image of the third index of the palm area (palm area), can play a role in emphasizing the palm area.

Through this, in one embodiment of the present invention, when the artificial intelligence model simply extracts only the palm lines, even if the extraction of the palm lines is not successful, the information that the individual has in the area of the palm excluding the fingers and the area of the palm including the fingers It can be expected that the unique characteristics of the palm may have the advantage of helping with identity recognition.

Meanwhile, in relation to this, in order to support the above advantages, the previously learned segmentation model for each index includes at least one process of Upsampling and Skip connections for converting low-dimensional feature points into high-dimensional feature points as shown in FIG. 4. By adopting the attention gate structure, the area of interest can be emphasized by lowering the weight of the remaining areas excluding the area of interest.

In this regard, the weight is a first feature point (x) transmitted to a specific layer through skip connections, as shown in FIG. 5, and a second feature point (g) transmitted from a previous layer neighboring the specific layer. It can be decided based on.

Here, the first feature point (x) transmitted through skip connections has a lower feature point depiction ability than the second feature point (g) transmitted from the previous neighboring layer, but may include more spatial information.

Afterwards, when a multi-channel image is generated from a photographed image of the palm, the classification unit 120 classifies the identity of the user matching the photographed image based on feature points for each index extracted from the multi-channel image (S130-S150). .

At this time, the classification unit 120 receives the multi-channel image through a previously learned classification model, and determines the identity of the user matching the captured image through an ensemble classifier that ensembles the feature point extraction results from the classification model. Can be classified.

Here, classifying a user's identity can be understood as a process of predicting the user according to class and deriving the probability of belonging to the actual class.

In other words, for example, as shown in FIG. 6, the classification unit 120 inputs images for each dimension in the multi-channel image into pre-trained classification models (Pre-trained CNN Model #1 to #3), and According to the input of the star image, the feature points derived from the classification model (Pre-trained CNN Model #1~#3) are ensembled, and the identity of the user matching the captured image is classified from the ensembled feature points through a classifier. It is possible.

Here, at least one of voting, bagging, and boosting may be applied as the ensemble method.

As discussed above, according to the operating method of the multidimensional biometric authentication device 100 according to an embodiment of the present invention, a region of interest (ROI: Region Of Interest) is generated through segmentation in relation to a photographed image of the palm. Interest) setting, extracting feature points using the region of interest, creating a multi-channel image for complex comparison, and using a computer vision algorithm to accurately set the region of interest for only the palm and feature points such as palm lines even in a complex background. It can be seen that security and accuracy can also be improved and system installation costs can be minimized by extracting and reconstructing/generating data for identity recognition.

Meanwhile, implementations of the functional operations and subjects described in this specification are implemented as digital electronic circuits, computer software, firmware, or hardware including the structure disclosed in this specification and its structural equivalents, or one or more of these. It can be implemented by combining. Implementations of the subject matter described herein may comprise one or more computer program products, that is, one or more modules of computer program instructions encoded on a tangible program storage medium for processing or execution by the operation of a processing system. It can be implemented.

The computer-readable medium may be a machine-readable storage device, a machine-readable storage substrate, a memory device, or a combination of one or more of these.

In this specification, “system” or “device” includes all instruments, devices, and machines for processing data, including, for example, programmable processors, computers, or multiple processors or computers. In addition to the hardware, the processing system may include code that forms an execution environment for computer programs on demand, such as code making up processor firmware, a protocol stack, a database management system, an operating system, or a combination of one or more of these. .

A computer program (also known as a program, software, software application, script, or code) may be written in any form of a programming language, including compiled, interpreted, a priori, or procedural languages; as a stand-alone program or module; It can be deployed in any form, including components, subroutines, or other units suitable for use in a computer environment. Computer programs do not necessarily correspond to files in a file system. A program may be stored within a single file that serves the requested program, or within multiple interacting files (e.g., files storing one or more modules, subprograms, or portions of code), or as part of a file that holds other programs or data. (e.g., one or more scripts stored within a markup language document). The computer program may be deployed to run on a single computer or multiple computers located at one site or distributed across multiple sites and interconnected by a communications network.

Meanwhile, computer-readable media suitable for storing computer program instructions and data include semiconductor memory devices such as EPROM, EEPROM, and flash memory devices, magnetic disks such as internal hard disks and external disks, magneto-optical disks, and CDs. -Can include all forms of non-volatile memory, media, and memory devices, including ROM and DVD-ROM disks. The processor and memory may be supplemented by, or integrated into, special-purpose logic circuitry.

Implementations of the subject matter described herein may include backend components, such as a data server, middleware components, such as an application server, or, such as a web browser or graphical user, through which a user may interact with an implementation of the subject matter described herein. It may be implemented in a front-end component, such as a client computer with an interface, or in a computing system that includes any combination of one or more of such back-end, middleware, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication, such as a telecommunications network.

Although this specification contains details of numerous specific implementations, these should not be construed as limitations on the scope of any invention or what may be claimed, but rather as descriptions of features that may be unique to particular embodiments of particular inventions. It must be understood. Likewise, certain features described herein in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination. Furthermore, although features may be described as operating in a particular combination and initially claimed as such, one or more features from a claimed combination may in some cases be excluded from that combination, and the claimed combination may be a sub-combination. It can be changed to a variant of a sub-combination.

Additionally, although operations are depicted in the drawings in a specific order herein, this should not be understood to mean that such operations must be performed in the specific order or sequential order shown or that all illustrated operations must be performed to obtain desirable results. Can not be done. In certain cases, multitasking and parallel processing may be advantageous. Additionally, the separation of various system components in the above-described embodiments should not be construed as requiring such separation in all embodiments, and the described program components and systems may generally be integrated together into a single software product or packaged into multiple software products. You must understand that you can

As such, this specification is not intended to limit the invention to the specific terms presented. Accordingly, although the present invention has been described in detail with reference to the above-described examples, those skilled in the art may make modifications, changes, and variations to the examples without departing from the scope of the present invention. The scope of the present invention is indicated by the claims described below rather than the detailed description above, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts should be interpreted as being included in the scope of the present invention. do.

According to the palm information-based multidimensional biometric authentication device and its operating method according to the present invention, user identity can be effectively classified (recognized) from a photographed image of the palm, thereby overcoming the limitations of existing technologies and related technologies. It is an invention that has industrial applicability because it not only has sufficient potential for commercialization or sales of the applied device, but also can be clearly implemented in reality.

100: Multidimensional biometric authentication device
110: generation unit 120: classification unit

Claims (12)

a generating unit that processes each region of interest image predicted in different ranges from a photographed image of the palm to generate a multi-channel image composed of an index; and
A classification unit that classifies the identity of the user matching the captured image based on feature points extracted from the multi-channel image,
The generating unit,
A multidimensional biometric authentication device characterized by sequentially predicting the image of each index in the multi-channel image through a previously learned segmentation model for each index. According to claim 1,
The multi-channel image is,
A first index consisting of an image in which the palm area including the fingers is set as a region of interest,
A second index consisting of an image in which the palm area excluding the fingers is set as the area of interest, and
A multidimensional biometric authentication device comprising a third index consisting of an image with the palm area set as the area of interest. According to claim 2,
The generating unit,
Predicting the image of the first index from the captured image,
Predicting the image of the second index from an image obtained by bitwise ANDing the image of the first index and the captured image,
A multi-dimensional biometric authentication device characterized in that the image of the third index is predicted from an image that emphasizes curved portions in the image of the second index. delete According to claim 1,
The segmentation model for each index is,
The area of interest is emphasized by lowering the weight of the remaining areas excluding the area of interest,
The weight is,
According to the adoption of an attention gate structure in at least one of skip connections and upsampling, a first feature point is transmitted to a specific layer through skip connections, and a second feature point is transmitted from a previous layer neighboring the specific layer. A multidimensional biometric authentication device characterized in that it is determined based on characteristic points. According to claim 1,
The classification department,
A multi-dimensional method characterized by receiving the multi-channel image through a pre-learned classification model and classifying the identity of the user matching the captured image through an ensemble classifier that ensembles the feature point extraction results from the classification model. Biometric authentication device. A generation step of generating a multi-channel image composed of an index by processing each region of interest image predicted in different ranges from a photographed image of the palm; and
It includes a classification step of classifying the identity of the user matching the captured image based on the feature points extracted from the multi-channel image,
The creation step is,
A method of operating a multidimensional biometric authentication device, characterized in that sequentially predicting the image of each index in the multi-channel image through a previously learned segmentation model for each index. According to claim 7,
The multi-channel image is,
A first index consisting of an image in which the palm area including the fingers is set as a region of interest,
A second index consisting of an image in which the palm area excluding the fingers is set as the area of interest, and
A method of operating a multidimensional biometric authentication device, comprising a third index consisting of an image with the palm area set as the area of interest. According to claim 8,
The creation step is,
Predicting the image of the first index from the captured image,
Predicting the image of the second index from an image obtained by bitwise ANDing the image of the first index and the captured image,
A method of operating a multidimensional biometric authentication device, characterized in that predicting the image of the third index from an image that emphasizes a curved part in the image of the second index. delete According to claim 7,
The segmentation model for each index is,
The area of interest is emphasized by lowering the weight of the remaining areas excluding the area of interest,
The weight is,
According to the adoption of an attention gate structure in at least one of skip connections and upsampling, a first feature point is transmitted to a specific layer through skip connections, and a second feature point is transmitted from a previous layer neighboring the specific layer. A method of operating a multidimensional biometric authentication device, characterized in that it is determined based on characteristic points. According to claim 7,
The classification step is,
A multi-dimensional method characterized by receiving the multi-channel image through a pre-learned classification model and classifying the identity of the user matching the captured image through an ensemble classifier that ensembles the feature point extraction results from the classification model. How a biometric authentication device operates.