KR20240003917A - Method and system for detecting noise feature using pair-wise similarity matrix between features stored in database - Google Patents

Abstract

A method and system for detecting noise features using a pairwise similarity matrix between features stored in a database are disclosed. A noise feature detection method according to an embodiment includes the steps of constructing a pair-wise similarity matrix including the similarity between each of the features registered in connection with identification information in a database, and using the pair-wise similarity matrix. It may include detecting features having a similarity greater than a threshold value among features of different identification information.

Description

Noise feature detection method and system using a pairwise similarity matrix between features stored in a database {METHOD AND SYSTEM FOR DETECTING NOISE FEATURE USING PAIR-WISE SIMILARITY MATRIX BETWEEN FEATURES STORED IN DATABASE}

The description below relates to a noise feature detection method and system using a pair-wise similarity matrix between features stored in a database.

Technology exists to recognize or authenticate faces using facial feature vectors. For example, facial recognition (or facial authentication) technology may generally include a registration step and a recognition (or authentication) step.

For example, in the registration stage, the facial recognition system finds the human face area in the video or image, aligns it using information such as eyes, nose, and mouth in the face area, and uses the aligned face image to find the area of the person's face. Extract N feature vectors. Afterwards, the facial recognition system can complete the registration step by storing the extracted features in the database. At this time, the extracted features may be stored in conjunction with the person's identification information, etc.

In addition, in the recognition stage, the facial recognition system finds the human face area in the video or image, aligns it using information such as eyes, nose, and mouth in the face area, and extracts N feature vectors from the aligned face image. can do. Afterwards, the facial recognition system can find and match features with high similarity to the extracted features in the database, and if the matched similarity is higher than the standard value, it can recognize (or authenticate) the face of the person included in the video or image.

However, in a facial recognition system, if features of duplicate images are registered in the database for different people's identification information or features of noisy images are registered, it may pose a threat to the overall performance of the facial recognition system. For example, this may be considered when the same person mistakenly registers features of his or her image in a database with different person's identification information, or when features of a photo that are too bright or too dark are registered.

In particular, in an environment where the registered original image cannot be confirmed, for example, in an environment where only identification information and features are registered in the database but the original image is not registered, there is a problem that it is difficult to even determine where the problem exists.

[Prior document number]

Korean Patent Publication No. 10-2021-0058157

A pair-wise similarity matrix is constructed for the features registered in the database of the facial recognition system, and in the constructed pair-wise similarity matrix, features whose similarity between features of different identification information is more than a threshold value are incorrectly registered. We provide a noise feature detection method and system that can refine the database of a facial recognition system by distinguishing between features.

Provided is a noise feature detection method and system that can refine the database of a facial recognition system by determining features registered by noisy images, such as light blur or images that are too dark, as noise features.

A method for detecting noise features in a computer system including at least one processor, comprising: a pair-wise similarity matrix including similarities between each of the features registered in connection with identification information in a database by the at least one processor; Constructing a similarity matrix; and detecting, by the at least one processor, features having a similarity greater than a threshold among features of different identification information through the pairwise similarity matrix.

According to one side, the database includes a database that stores features for a facial area extracted from an image or image in a facial recognition system in association with corresponding identification information, and the first threshold, which is the threshold, is set to the facial recognition system. may be determined based on a preset second threshold for recognizing or authenticating a person's face according to the degree of similarity between features.

According to another aspect, the noise feature detection method is performed by the at least one processor, based on n (where n is a natural number of 2 or more) features of different identification information through the pairwise similarity matrix. A step of detecting features having similarity may be further included.

According to another aspect, the third threshold may be less than the threshold.

According to another aspect, the method for detecting noise features may further include transmitting, by the at least one processor, a request to delete the detected features from the database.

According to another aspect, the method for detecting noise features includes providing, by the at least one processor, identification information associated with the detected features to a user or manager of the identification information associated with the detected features. It may further include.

A computer program stored on a computer-readable recording medium is provided in conjunction with a computer system to execute the method on the computer system.

A program recorded on a computer-readable recording medium that performs the above method is provided.

A pairwise similarity matrix (pair- A computer device is provided that configures a wise similarity matrix and detects features having a similarity greater than a threshold value among features of different identification information through the pairwise similarity matrix.

A pair-wise similarity matrix is constructed for the features registered in the database of the facial recognition system, and in the constructed pair-wise similarity matrix, features whose similarity between features of different identification information is more than a threshold value are incorrectly registered. By distinguishing between features, the database of the facial recognition system can be refined.

The database of the facial recognition system can be refined by identifying features registered by noisy images, such as light blur or images that are too dark, as noise features.

1 is a diagram illustrating an example of a network environment according to an embodiment of the present invention.
Figure 2 is a block diagram showing an example of a computer device according to an embodiment of the present invention.
Figure 3 is a diagram showing an example of the general appearance of a noise feature detection system according to an embodiment of the present invention.
Figure 4 is a flowchart showing an example of a noise feature detection method according to an embodiment of the present invention.
Figures 5 and 6 are diagrams showing an example of a pairwise similarity matrix according to an embodiment of the present invention.

Hereinafter, embodiments will be described in detail with reference to the accompanying drawings.

The noise feature detection system according to embodiments of the present invention may be implemented by at least one computer device, and the noise feature detection method according to embodiments of the present invention may be implemented by at least one computer device implementing the noise feature detection system. It can be performed through . The computer program according to an embodiment of the present invention may be installed and driven in the computer device, and the computer device may perform the noise feature detection method according to the embodiment of the present invention under the control of the driven computer program. The above-described computer program can be combined with a computer device and stored in a computer-readable recording medium to execute the noise feature detection method on the computer.

1 is a diagram illustrating an example of a network environment according to an embodiment of the present invention. The network environment in FIG. 1 shows an example including a plurality of electronic devices 110, 120, 130, and 140, a plurality of servers 150 and 160, and a network 170. Figure 1 is an example for explaining the invention, and the number of electronic devices or servers is not limited as in Figure 1. In addition, the network environment in FIG. 1 only explains one example of environments applicable to the present embodiments, and the environment applicable to the present embodiments is not limited to the network environment in FIG. 1.

The plurality of electronic devices 110, 120, 130, and 140 may be fixed terminals or mobile terminals implemented as computer devices. Examples of the plurality of electronic devices 110, 120, 130, and 140 include smart phones, mobile phones, navigation devices, computers, laptops, digital broadcasting terminals, PDAs (Personal Digital Assistants), and PMPs (Portable Multimedia Players). ), tablet PC, etc. For example, in FIG. 1, the shape of a smartphone is shown as an example of the electronic device 110. However, in embodiments of the present invention, the electronic device 110 actually communicates with other devices through the network 170 using a wireless or wired communication method. It may refer to one of various physical computer devices capable of communicating with electronic devices 120, 130, 140 and/or servers 150, 160.

The communication method is not limited, and may include not only a communication method utilizing a communication network that the network 170 may include (for example, a mobile communication network, wired Internet, wireless Internet, and a broadcast network), but also short-range wireless communication between devices. For example, the network 170 may include a personal area network (PAN), a local area network (LAN), a campus area network (CAN), a metropolitan area network (MAN), a wide area network (WAN), and a broadband network (BBN). , may include one or more arbitrary networks such as the Internet. Additionally, the network 170 may include any one or more of network topologies including a bus network, star network, ring network, mesh network, star-bus network, tree or hierarchical network, etc. Not limited.

Each of the servers 150 and 160 is a computer device or a plurality of computers that communicate with a plurality of electronic devices 110, 120, 130, 140 and a network 170 to provide commands, codes, files, content, services, etc. It can be implemented with devices. For example, the server 150 provides services (e.g., facial recognition service, cloud service, game service, content provision service) to a plurality of electronic devices 110, 120, 130, and 140 connected through the network 170. , group call service (or voice conference service), messaging service, mail service, social network service, map service, translation service, financial service, payment service, search service, etc.).

Figure 2 is a block diagram showing an example of a computer device according to an embodiment of the present invention. Each of the plurality of electronic devices 110, 120, 130, and 140 described above or each of the servers 150 and 160 may be implemented by the computer device 200 shown in FIG. 2.

As shown in FIG. 2, this computer device 200 may include a memory 210, a processor 220, a communication interface 230, and an input/output interface 240. The memory 210 is a computer-readable recording medium and may include a non-permanent mass storage device such as random access memory (RAM), read only memory (ROM), and a disk drive. Here, non-perishable large-capacity recording devices such as ROM and disk drives may be included in the computer device 200 as a separate permanent storage device that is distinct from the memory 210. Additionally, an operating system and at least one program code may be stored in the memory 210. These software components may be loaded into the memory 210 from a computer-readable recording medium separate from the memory 210. Such separate computer-readable recording media may include computer-readable recording media such as floppy drives, disks, tapes, DVD/CD-ROM drives, and memory cards. In another embodiment, software components may be loaded into the memory 210 through the communication interface 230 rather than a computer-readable recording medium. For example, software components may be loaded into memory 210 of computer device 200 based on computer programs installed by files received over network 170.

The processor 220 may be configured to process instructions of a computer program by performing basic arithmetic, logic, and input/output operations. Commands may be provided to the processor 220 by the memory 210 or the communication interface 230. For example, processor 220 may be configured to execute received instructions according to program code stored in a recording device such as memory 210.

The communication interface 230 may provide a function for the computer device 200 to communicate with other devices (eg, the storage devices described above) through the network 170. For example, a request, command, data, file, etc. generated by the processor 220 of the computer device 200 according to a program code stored in a recording device such as memory 210 is transmitted to the network ( 170) and can be transmitted to other devices. Conversely, signals, commands, data, files, etc. from other devices may be received by the computer device 200 through the communication interface 230 of the computer device 200 via the network 170. Signals, commands, data, etc. received through the communication interface 230 may be transmitted to the processor 220 or memory 210, and files, etc. may be stored in a storage medium (as described above) that the computer device 200 may further include. It can be stored as a permanent storage device).

The input/output interface 240 may be a means for interfacing with the input/output device 250. For example, input devices may include devices such as a microphone, keyboard, or mouse, and output devices may include devices such as displays and speakers. As another example, the input/output interface 240 may be a means for interfacing with a device that integrates input and output functions, such as a touch screen. The input/output device 250 may be configured as a single device with the computer device 200.

Additionally, in other embodiments, computer device 200 may include fewer or more components than those of FIG. 2 . However, there is no need to clearly show most prior art components. For example, the computer device 200 may be implemented to include at least some of the input/output devices 250 described above, or may further include other components such as a transceiver, a database, etc.

Figure 3 is a diagram showing an example of the general appearance of a noise feature detection system according to an embodiment of the present invention. In the embodiment of FIG. 3 , the facial recognition system 300 may include a feature extraction unit 310, a feature database 320, and a noise feature detection system 330. 3 shows an example in which the noise feature detection system 330 is implemented on the facial recognition system 300. However, depending on the embodiment, the noise feature detection system 330 implements the facial recognition system 300. It may also be implemented through a separate physical electronic device (eg, computer device 200). In this case, the noise feature detection system 330 is connected to the facial recognition system 300 through a network (e.g., network 170) to detect noise features from the feature database 320 included in the facial recognition system 300. It can be detected.

The feature extraction unit 310 finds the human face area in the input image 340, aligns it using information such as eyes, nose, and mouth in the face area, and selects N numbers from the aligned face image. Feature vectors can be extracted. 3 uses the input image 340 as an example, but it is also possible to find a face area in a video or the like and extract N feature vectors for each frame or for each frame at a certain period.

At this time, the facial recognition system 300 can process the registration step by registering the extracted N features (feature vectors) in the feature database 320 in connection with the identification information input together with the input image 340. .

In the case of the recognition (or authentication) step, the facial recognition system 300 compares the similarity with the features registered in the feature database 320 through the extracted N features to determine the similarity above the first threshold for face recognition. By obtaining the identification information of the features, the face of the person included in the input image 340 can be recognized (or authenticated). The method of calculating similarity between features may be performed using at least one of well-known methods such as Euclidean distance or cosine similarity.

Identification information is information to identify a person, and one or more of various items of information such as name, phone number, resident registration number, employee number, and student number may be used as needed. For example, name and phone number may be used together as a single identification information to distinguish users with the same name.

At this time, as already described, noise features may be registered in the feature database 320. A user may accidentally register his or her face (features) in another person's identification information, and during the image creation process, features in an image that are too bright or dark due to light may be registered. Noise feature detection system 320 may detect such noise features in feature database 320.

To this end, the noise feature detection system 320 may configure and utilize a pair-wise similarity matrix between features registered in the feature database 320. Here, the pairwise similarity matrix may be configured to include the similarity between each feature registered in connection with identification information in the database. In this case, the noise feature detection system 320 may detect features having a similarity greater than the second threshold among features of different identification information through a pairwise similarity matrix. At this time, the second threshold may be determined based on the first threshold. Basically, the second threshold may be set to be the same as the first threshold, but may also be set to a value smaller than the first threshold to increase the possibility of detecting noise features.

Meanwhile, in the case of an image that is too bright or dark due to light, it is difficult to uniquely derive features within the image, so most features of the face area included in the image may have a similar degree of similarity. In this case, the features of one identification information may have a similar degree of similarity to the features of a plurality of different identification information. In order to detect noise features in this case, the noise feature detection system 320 uses a pairwise similarity matrix to detect features of n (n is a natural number of 2 or more) or more different identification information with a similarity of more than a third threshold. Features can be detected. For example, the noise feature detection system 320 may detect feature A as a noise feature if, for a specific feature A, there are n or more features of other identification information whose similarity to feature A is greater than or equal to a third threshold. . At this time, since simply cases where two features of different identification information are similar are filtered out through the second threshold, the third threshold may be set to have a value less than the second threshold. In other words, a plurality of features (n features) having a similarity greater than the third threshold may be further detected by the noise feature detection system 320 due to blurring by light or an image that is too dark.

The noise feature detection system 320 may perform post-processing on the detected features. As an example, the noise feature detection system 320 may transmit a request to delete detected features from the feature database 320. This request may be sent directly to the feature database 320 or may be sent to a user or administrator. Upon request, a user or administrator may delete features detected in the feature database 320 and then re-register them. As another example, the noise feature detection system 320 may provide identification information associated with the detected features to a user or manager of the identification information associated with the detected features. As a more specific example, in the case of a system where an administrator exists, identification information can be transmitted to the administrator to register a new feature for the person corresponding to the identification information. Additionally, in the case of a system where there is no administrator and individual users directly register features, the identification information obtained through noise features can be directly delivered to the users, allowing the users to register new features. Here, delivering identification information may correspond to notifying users of the identification information to request registration of a new feature.

Figure 4 is a flowchart showing an example of a noise feature detection method according to an embodiment of the present invention. The noise feature detection method according to this embodiment may be performed by the computer device 200 implementing the noise feature detection system 330 described above. At this time, the processor 220 of the computer device 200 may be implemented to execute control instructions according to the code of an operating system included in the memory 210 or the code of at least one computer program. Here, the processor 220 causes the computer device 200 to perform steps 410 to 450 included in the method of FIG. 4 according to control instructions provided by code stored in the computer device 200. can be controlled.

In step 410, the computer device 200 may configure a pairwise similarity matrix including the similarity between each feature registered in connection with identification information in the database. Here, the database may correspond to the feature database 320 of the facial recognition system 300 described above with reference to FIG. 3. The computer device 200 may calculate the similarity of each feature registered in the database with each other feature and construct a pairwise similarity matrix including the similarity between each feature. This pairwise similarity matrix will be described in more detail later with reference to FIGS. 5 and 6.

In step 420, the computer device 200 may detect features having a similarity greater than a first threshold value among features of different identification information through a pairwise similarity matrix. Here, as already described, the database may include a database that stores features for a facial area extracted from an image or image in a facial recognition system (for example, the facial recognition system 300 of FIG. 3). In this case, the first threshold may be determined based on the second threshold preset for face authentication in the facial recognition system. At this time, the detected features may be determined as noise features stored in a database by linking features of the same face with different identification information.

In step 430, the computer device 200 may detect a feature having a similarity of more than a third threshold value with n (n is a natural number of 2 or more) features of other identification information through a pairwise similarity matrix. Here, the third threshold may have a value less than the first threshold. As described above, features with a similarity greater than the first threshold may be filtered out in step 420, and if there are many features that are not filtered out in step 420 but have a similarity greater than the third threshold, the light These may be noise features extracted from an image containing noise (an image that is too bright or too dark to normally extract features of the face area). This step 430 may be performed after step 420 as in the embodiment of FIG. 4, but depending on the embodiment, it may be performed in parallel with step 420 or before step 420.

When noise features are detected through the pairwise similarity matrix, the computer device 200 may perform at least one of steps 440 and 450 for post-processing.

In step 440, the computer device 200 may transmit a request to delete the detected features from the database. For example, the facial recognition system 300 may request to register the face area in response to a facial recognition (or facial authentication) request for the facial area of a feature that is not searched in the database, so facial recognition (or facial authentication) Upon request, the features for the corresponding user may be induced to be registered in the database again.

In step 450, the computer device 200 may provide the identification information associated with the detected features to the user or manager of the identification information associated with the detected features. As already explained, conveying identifying information may correspond to notifying users of that identifying information to request registration of new features.

In this way, the computer device 200 may perform post-processing on the detected noise features in a passive manner (deleting the detected features from the database) and/or an active manner (notifying a request for registration of a new feature).

Figures 5 and 6 are diagrams showing an example of a pairwise similarity matrix according to an embodiment of the present invention.

The table in FIG. 5 is an example of a pairwise similarity matrix showing the similarity between features registered for user A, user B, user C, user D, and user E. At this time, it is assumed that the first threshold described through step 420 of FIG. 4 is 0.9. In this case, the similarity between User A's feature "111000??" and User B's feature "111100??" is 0.95, and the similarity between User C's feature "000000??" and User D's feature "000000??" is 1. All may have a value greater than or equal to the first threshold value of 0.9. In this case, User A's feature "111000??", User B's feature "111100??", and User C's feature "000000??" and user D's feature “000000??” may be detected as noise features, respectively.

Meanwhile, the table in FIG. 6 is an example of a pairwise similarity matrix showing the similarity between features registered for user F, user G, user H, user I, and user J. At this time, it is assumed that the first threshold described through step 420 of FIG. 4 is 0.9. However, in the table of FIG. 6, there are no features with a similarity greater than the first threshold. However, it is assumed that the third threshold described through step 430 of FIG. 4 is 0.8 and n is 3. At this time, in the table of FIG. 6, the similarity between user F's feature "000000??" and user H's feature "000000??" is 0.83, and user F's feature "000000??" and user I's feature "000000??" The similarity between them is 0.82, and the similarity between user F's feature "000000??" and user J's feature "000000??" is 0.83. In other words, it can be seen that user F's features have a similarity of 0.8 or more, which is the third threshold, with three or more features of other identification information. In this case, user F's features can be detected at least as noise features. Depending on the embodiment, the features of user H, user I, and user J may also be detected as noise features.

As such, according to embodiments of the present invention, a pair-wise similarity matrix is constructed for the features registered in the database of the facial recognition system, and features of different identification information are selected from the constructed pair-wise similarity matrix. The database of the facial recognition system can be refined by determining features whose similarity between features is greater than a threshold as incorrectly registered features. Additionally, the database of the facial recognition system can be refined by determining features registered by noisy images, such as blurring by light or images that are too dark, as noise features.

The device described above may be implemented with hardware components or a combination of hardware components and software components. For example, the devices and components described in the embodiments include a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), and a programmable logic unit (PLU). It may be implemented using one or more general-purpose or special-purpose computers, such as a logic unit, microprocessor, or any other device capable of executing and responding to instructions. The processing device may execute an operating system (OS) and one or more software applications running on the operating system. Additionally, a processing device may access, store, manipulate, process, and generate data in response to the execution of software. For ease of understanding, a single processing device may be described as being used; however, those skilled in the art will understand that a processing device includes multiple processing elements and/or multiple types of processing elements. It can be seen that it may include. For example, a processing device may include a plurality of processors or one processor and one controller. Additionally, other processing configurations, such as parallel processors, are possible.

Software may include a computer program, code, instructions, or a combination of one or more of these, which may configure a processing unit to operate as desired, or may be processed independently or collectively. You can command the device. The software and/or data may be embodied in any type of machine, component, physical device, computer storage medium or device for the purpose of being interpreted by or providing instructions or data to the processing device. there is. Software may be distributed over networked computer systems and stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. At this time, the medium may continuously store a computer-executable program, or temporarily store it for execution or download. In addition, the medium may be a variety of recording or storage means in the form of a single or several pieces of hardware combined. It is not limited to a medium directly connected to a computer system and may be distributed over a network. Examples of media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical recording media such as CD-ROMs and DVDs, magneto-optical media such as floptical disks, And there may be something configured to store program instructions, including ROM, RAM, flash memory, etc. Additionally, examples of other media include recording or storage media managed by app stores that distribute applications, sites or servers that supply or distribute various other software, etc.

As described above, although the embodiments have been described with limited examples and drawings, various modifications and variations can be made by those skilled in the art from the above description. For example, the described techniques are performed in a different order than the described method, and/or components of the described system, structure, device, circuit, etc. are combined or combined in a different form than the described method, or other components are used. Alternatively, appropriate results may be achieved even if substituted or substituted by an equivalent.

Therefore, other implementations, other embodiments, and equivalents of the claims also fall within the scope of the claims described below.

Claims (10)

A method for detecting noise features in a computer system including at least one processor, comprising:
Constructing, by the at least one processor, a pair-wise similarity matrix including similarities between each of the features registered in connection with identification information in a database; and
Detecting, by the at least one processor, features having a similarity greater than a threshold value among features of different identification information through the pairwise similarity matrix.
A noise feature detection method comprising: According to paragraph 1,
The database includes a database that stores features of the facial area extracted from a video or image in a facial recognition system in association with corresponding identification information,
The first threshold, which is the threshold, is determined based on a preset second threshold to recognize or authenticate a person's face according to the similarity between features in the facial recognition system.
A noise feature detection method characterized by According to paragraph 1,
Detecting, by the at least one processor, a feature having a similarity greater than a third threshold based on n (where n is a natural number of 2 or more) features of different identification information through the pairwise similarity matrix.
A noise feature detection method further comprising: According to paragraph 3,
The method of detecting a noise feature, wherein the third threshold is less than the threshold. According to paragraph 1,
sending, by the at least one processor, a request to delete the detected features from the database.
A noise feature detection method further comprising: According to paragraph 1,
Providing, by the at least one processor, identification information associated with the detected features to a user or administrator of the identification information associated with the detected features.
A noise feature detection method further comprising: A computer program stored in a computer-readable recording medium in combination with a computer system to execute the method of any one of claims 1 to 6 on the computer system. A program recorded on a computer-readable recording medium that performs the method of any one of claims 1 to 6. At least one processor implemented to execute computer readable instructions
Including,
By the at least one processor,
Construct a pair-wise similarity matrix containing the similarity between each feature registered in the database in connection with identification information,
Detecting features with a similarity greater than a threshold among features of different identification information through the pairwise similarity matrix.
A computer device characterized by a. According to clause 9,
By the at least one processor,
Sending a request to delete the detected features from the database,
Providing identification information associated with the detected features to a user or administrator of the identification information associated with the detected features.
A computer device characterized by a.

Images