Facial recognition device and facial recognition method

Abstract

A face recognition device is provided. The device includes a camera configured to acquire a partial face image of a passenger, and a controller configured to learn a boarding condition of a user, and recognize a face based on the partial face image of the passenger when a current boarding condition of the passenger satisfies the learned boarding condition.

Description

[0001] Cross-references to related applications

[0002] This application claims priority to Korean Patent Application No. 10-2019-0135560, filed with the Korean Intellectual Property Office on October 29, 2019, which is incorporated herein by reference. Technical Field

[0003] This disclosure relates to a facial recognition device and a facial recognition method. Background Technology

[0004] Recently, a technology has been developed to increase the convenience of users by using facial recognition to determine whether a passenger is a specific user and to set up in-vehicle functions suitable for that user. Typically, feature points are extracted from the passenger's entire facial image, and facial recognition is determined based on whether the extracted feature points match a predetermined number of previously stored user feature points. However, when a passenger's entire facial image is not available due to partial occlusion, feature points are difficult to extract, and therefore, even if a user boards the vehicle, the passenger may not be recognized. Summary of the Invention

[0005] The embodiments disclosed herein can solve the above-mentioned problems in the prior art, while fully maintaining the advantages achieved by the prior art.

[0006] Embodiments of this disclosure provide a facial recognition device and a facial recognition method that can accurately recognize faces even when a passenger's face is partially obscured.

[0007] The technical problems to be solved by the embodiments of the present invention are not limited to those described above, and any other technical problems not mentioned herein will be clearly understood by those skilled in the art to which this disclosure pertains from the following description.

[0008] According to an embodiment of this disclosure, a facial recognition device includes: a camera for acquiring a partial facial image of a passenger; and a controller for learning the user's travel conditions and recognizing a face based on the partial facial image of the passenger when the passenger's current travel conditions meet the learned travel conditions.

[0009] In one embodiment, the travel conditions may include information about the travel location and travel time.

[0010] In one embodiment, the controller can acquire the first passenger's travel conditions and a first full facial image, extract feature points from the first full facial image and register the first passenger as a user, register the first passenger's travel conditions as the user's travel conditions, and store the number of times the user travels under the user's travel conditions as "once".

[0011] In one embodiment, when learning a user's travel conditions, the controller can acquire the second passenger's travel conditions for learning and a second full facial image, and extract feature points from the second full facial image to determine whether the second passenger is a registered user.

[0012] In one embodiment, when the second passenger is a registered user, the controller can determine whether the ride conditions used for learning meet the user's ride conditions.

[0013] In one embodiment, when the ride location used for learning is within a predetermined distance from the user's ride location, or when the ride time used for learning is within a predetermined time from the user's ride time, the controller can determine that the ride conditions used for learning meet the user's ride conditions.

[0014] In one embodiment, when the ride conditions used for learning meet the user's ride conditions, the controller can add one more ride to the user's ride history under the user's ride conditions.

[0015] In one embodiment, when a user takes a ride more than a predetermined number of times under the user's ride conditions, the controller can determine that it has completed learning the user's ride conditions.

[0016] In one embodiment, when the ride conditions used for learning do not meet the user's ride conditions, the controller can register the ride conditions used for learning as the user's new ride conditions and store the number of times the user rides under the user's new ride conditions as "once".

[0017] In one embodiment, when a passenger's current travel conditions meet the travel conditions learned by the user, the controller can reduce the number of feature points used to identify the face.

[0018] In one embodiment, the controller can determine that the passenger is a user when feature points extracted from a portion of a passenger's facial image match feature points extracted during user registration.

[0019] According to another embodiment of this disclosure, a facial recognition method includes: learning a user's travel conditions; acquiring a partial facial image of a passenger; and recognizing a face based on the partial facial image of the passenger when the passenger's current travel conditions meet the learned travel conditions of the user.

[0020] In one embodiment, the travel conditions may include information about the travel location and travel time.

[0021] In one embodiment, before learning the user's travel conditions, the method may further include: acquiring the travel conditions of a first passenger and a first full facial image; extracting multiple feature points from the first full facial image and registering the first passenger as a user; and registering the first passenger's travel conditions as the user's travel conditions, and storing the number of times the user travels under the user's travel conditions as "once".

[0022] In one embodiment, learning a user's travel conditions may include: acquiring a second passenger's travel conditions for learning and a second full facial image; extracting feature points from the second full facial image to determine whether the second passenger is a registered user; when the second passenger is a registered user, determining whether the travel conditions for learning meet the user's travel conditions; when the travel conditions for learning meet the user's travel conditions, adding one to the number of times the user has traveled under the user's travel conditions; determining whether the number of times the user has traveled under the user's travel conditions exceeds a predetermined number; and when the number of times the user has traveled under the user's travel conditions exceeds the predetermined number, determining that the learning of the user's travel conditions has been completed.

[0023] In one embodiment, determining whether the travel conditions for learning meet the user's travel conditions may include: determining that the travel conditions for learning meet the user's travel conditions when the travel location for learning is within a predetermined distance from the user's travel location, or when the travel time for learning is within a predetermined time from the user's travel time.

[0024] In one embodiment, the method may further include: registering the learned ride conditions as the user's new ride conditions when the ride conditions used for learning do not meet the user's ride conditions; and storing the number of times the user rides under the new ride conditions as "once".

[0025] In one embodiment, before recognizing faces based on partial facial images of passengers, the method may further include reducing the number of feature points used to recognize faces when the passenger's current travel conditions meet the learned user's travel conditions.

[0026] In one embodiment, identifying a face based on a partial facial image of a passenger may include: extracting feature points from the partial facial image of the passenger; determining whether the feature points extracted from the partial facial image of the passenger match the feature points extracted during user registration; and determining that the passenger is a user when the feature points extracted from the partial facial image of the passenger match the feature points extracted during user registration. Attached Figure Description

[0027] The above and other objects, features and advantages of this disclosure will become more apparent from the following detailed description taken in conjunction with the accompanying drawings, wherein:

[0028] Figure 1 This is a diagram illustrating the configuration of a facial recognition device according to an embodiment of the present disclosure;

[0029] Figure 2 This is a schematic view showing the feature points extracted from the entire facial image;

[0030] Figure 3 and Figure 4 This is an illustration showing an example where a full facial image could not be captured;

[0031] Figure 5 This is a view schematically illustrating feature points extracted according to an embodiment of the present disclosure;

[0032] Figure 6 This is a flowchart illustrating a method for registering a passenger as a user according to an embodiment of the present disclosure;

[0033] Figure 7 This is a flowchart illustrating a method for learning riding conditions according to an embodiment of the present disclosure;

[0034] Figure 8 This is a flowchart illustrating a method for determining whether a user's travel conditions are met according to an embodiment of the present disclosure;

[0035] Figure 9 This is a flowchart illustrating a facial recognition method according to an embodiment of the present disclosure; and

[0036] Figure 10 A computing system is shown that performs a method according to an embodiment of the present disclosure. Detailed Implementation

[0037] In the following, some embodiments of the present disclosure will be described in detail with reference to the exemplary accompanying drawings. When adding reference numerals to components in each drawing, it should be noted that the same or equivalent components are indicated by the same reference numerals even if shown in different drawings. Furthermore, in describing embodiments of the present disclosure, detailed descriptions of well-known features or functions will be omitted to avoid unnecessarily obscuring the gist of the disclosure.

[0038] In describing components according to embodiments of this disclosure, terms such as “first,” “second,” “A,” “B,” “(a),” and “(b)” may be used. These terms are intended only to distinguish one component from another, and they do not limit the nature, order, or sequence of the constituent components. Unless otherwise defined, all terms used herein, including technical or scientific terms, shall have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains. Terms such as those defined in a general dictionary shall be interpreted as having the same meaning as in the context of the relevant technical field, and unless a term is expressly defined herein as having an ideal or overly formal meaning, it shall not be interpreted as having an ideal or overly formal meaning.

[0039] Figure 1 This is a diagram illustrating the configuration of a facial recognition device according to an embodiment of the present disclosure.

[0040] like Figure 1 As shown, the facial recognition device 100 according to an embodiment of the present disclosure may include a camera 110, a sensor 120, a navigation device 130, a storage device 140, and a controller 150.

[0041] Camera 110 can acquire facial images of passengers. Camera 110 can be positioned around the instrument cluster so that the image is pointed towards the passenger's face. Camera 110 may include an infrared sensor, allowing for easy acquisition of facial images of passengers even in dimly lit vehicle interiors.

[0042] Sensor 120 can detect whether a passenger is in the vehicle. To this end, sensor 120 may include a weight sensor disposed in the seat and a sensor for detecting whether the seat belt is fastened.

[0043] Navigation 130 can obtain the travel conditions of passengers. These conditions may include travel time and location. To this end, navigation 130 may include a GPS receiver to receive the vehicle's current location and provide map image information for a specific area based on that location. Additionally, navigation 130 can provide time data output from the GPS receiver to controller 150.

[0044] Storage device 140 can store the results of operations determined by controller 150. For example, storage device 140 can store user information, user travel information, learning results of travel information, etc. Additionally, storage device 140 can store at least one algorithm for calculating or executing various commands and learning the operation of the facial recognition device according to embodiments of this disclosure. Storage device 140 may include at least one storage medium selected from flash memory, hard disk, memory card, read-only memory (ROM), random access memory (RAM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic storage, magnetic disk, and optical disk.

[0045] The controller 150 can be implemented by various processing devices, such as a microprocessor having a built-in semiconductor chip capable of performing various commands or operations, and can control the operation of the facial recognition device according to embodiments of the present disclosure. Specifically, the controller 150 can learn the passenger's travel conditions, and when the passenger's current travel conditions meet the previously learned travel conditions, the controller 150 can recognize the face based on a partial facial image of the passenger.

[0046] First, the controller 150 can register passengers as users and register their travel conditions. To do this, the controller 150 can identify passengers' travel and obtain their travel conditions. Additionally, the controller 150 can acquire a full facial image of the passenger from the camera 110 and extract feature points from the entire facial image. (Refer to...) Figure 2 Provide a detailed description.

[0047] Figure 2 This is a schematic view illustrating feature points extracted from the entire facial image. (Example) Figure 2 As shown, the controller 150 can extract multiple feature points from the entire facial image. Here, the entire facial image can include an image of a portion of the face that is not obscured. For example, the entire facial image can represent an image of a portion of the face that is not obscured by sunglasses, a hat, etc. Furthermore, the number of feature points extracted by the controller 150 can be sufficient to identify the face within the entire facial image. Thus, the controller 150 extracts feature points from the entire facial image to identify the face under normal conditions.

[0048] The controller 150 can extract feature points and use these extracted feature points to identify the passenger's face. When the passenger's face is identified based on the feature points extracted from the entire facial image, the controller 150 can determine that the face has been identified under normal conditions, and when the passenger's face is identified under normal conditions, the controller 150 can register the passenger as a user. When the passenger is registered as a user, the controller 150 can register the travel conditions obtained when the passenger boards the vehicle as the user's travel conditions. Furthermore, when user registration and travel condition registration are completed, the controller 150 can control the storage of user registration information and user travel conditions in the storage device 140. Additionally, when user registration and travel condition registration are completed, the controller 150 can control the storage of the number of times the user has boarded the vehicle under the user's travel conditions as "once".

[0049] Once user registration and user travel condition registration are completed, controller 150 learns the passenger's travel conditions. That is, controller 150 can continue to learn whether a passenger is a user after user registration and user travel condition registration.

[0050] More specifically, the controller 150 identifies the passenger's boarding and obtains the boarding conditions. Here, the boarding conditions may include the passenger's boarding time and boarding location. Furthermore, since the controller 150 obtains the boarding conditions for learning purposes, for convenience, the boarding conditions are referred to as boarding conditions for learning.

[0051] When acquiring ride conditions for learning, controller 150 acquires a full facial image of the passenger. Additionally, controller 150 extracts feature points from the entire facial image and identifies the passenger's face based on the extracted feature points. Controller 150 determines whether the identified passenger's face belongs to a previously registered user. If the passenger is determined to be a previously registered user, controller 150 determines whether the learned ride conditions meet the user's ride conditions.

[0052] Here, the controller 150 can determine the similarity between the ride location and ride time used for learning and the user's ride location and ride time, and based on the determination result, determine whether the ride conditions used for learning meet the user's ride conditions.

[0053] Specifically, the controller 150 can determine whether the learning location is within a predetermined distance from the user's location. Here, the predetermined distance can be a radius of 500 meters. Furthermore, when the learning location is within the predetermined distance from the user's location, the controller 150 can determine that the learning conditions meet the user's travel requirements.

[0054] Additionally, the controller 150 can determine whether the travel time for learning is within a predetermined timeframe from the user's travel time. Here, the predetermined time can be 60 minutes. Furthermore, when the travel time for learning is within the predetermined timeframe from the user's travel time, the controller 150 can determine that the travel conditions for learning meet the user's travel conditions.

[0055] When the controller 150 determines that the travel conditions used for learning meet the user's travel conditions, it can control the addition of the number of times the passenger travels under the user's travel conditions by one, and store the added number. When the controller 150 determines that the travel conditions used for learning do not meet the user's travel conditions, it can control the registration of the travel conditions used for learning as the user's new travel conditions, and store the number of times the passenger travels under the user's new travel conditions as "one".

[0056] The controller 150 can add a predetermined number of times the passenger takes the ride to the number of rides until the passenger has taken the ride more than N times (e.g., 10 times) under the user's ride conditions, and determine whether the passenger's ride conditions match the user's ride conditions each time the passenger takes the ride, thereby learning the user's ride conditions. According to an embodiment, the controller 150 can use a recurrent neural network (RNN) to learn the user's ride conditions, in which case the input values ​​can include location and time (date (day) and time). Furthermore, when the number of times the user takes the ride under the user's ride conditions exceeds the predetermined number (e.g., 10 times), the controller 150 can determine that it has learned the user's ride conditions.

[0057] The following describes the operation of controller 150 in recognizing the face of the passenger currently in the vehicle based on previously learned riding conditions.

[0058] The controller 150 identifies the passenger's boarding information and obtains the current boarding conditions. Here, the current boarding conditions may include the passenger's current boarding time and current boarding location.

[0059] The controller 150 determines whether the acquired current travel conditions meet the previously learned travel conditions. That is, the controller 150 can determine whether the current travel location and current travel time meet the previously learned travel conditions for the user. Here, the controller 150 can determine whether the passenger's travel location is within a predetermined distance from the user's travel location, and can determine whether the passenger's travel time is within a predetermined time from the user's travel time.

[0060] When the current travel conditions are determined to meet the previously learned travel conditions, the controller 150 reduces the number of feature points used to identify the passenger's face. Although the controller 150 extracts enough feature points from the entire facial image to identify the face during user registration and travel condition learning operations, it can reduce and reset the number of feature points when the current travel conditions are determined to meet the previously learned travel conditions.

[0061] For example, when the number of extracted feature points is 70, the controller 150 can determine that at least 70 feature points are needed to identify the face in the entire facial image. On the other hand, according to embodiments of this disclosure, when it is determined that the current riding conditions meet the previously learned riding conditions, the controller 150 can reduce the number of feature points used to identify the face and reduce the number of feature points used to identify the face to 35.

[0062] The controller 150 acquires a partial facial image of the passenger. Here, "partial facial image" may refer to an image where a portion of the face is obscured, rather than a full facial image, which will be referenced... Figure 3 and Figure 4 Describe it.

[0063] Figure 3 and Figure 4 This is an illustration showing an example of why it's impossible to capture the entire face image.

[0064] like Figure 3 As shown, the controller 150 can acquire an image of a portion of the passenger's face being obscured by the steering wheel 30, and as... Figure 4 As shown, the controller 150 can acquire images that do not include the face due to the passenger's sitting posture or body structure.

[0065] The controller 150 can extract feature points based on an image where a portion of the face is occluded. This will refer to... Figure 5 Describe it.

[0066] Figure 5 This is a view schematically illustrating feature points extracted according to an embodiment of the present disclosure.

[0067] like Figure 5 As shown, the controller 150 can extract feature points 50 based on an image where a portion of the face is occluded (obscured by the steering wheel 51). Here, because a portion of the face is occluded, the number of feature points 50 that the controller 150 can extract is less than the number of feature points that can be extracted from the entire face image. Figure 2 The number of “20” in the text.

[0068] The controller 150 can determine whether feature points extracted from an image in which a portion of the face is occluded match feature points extracted during user registration. Here, the controller 150 can determine whether the number of feature points extracted from the image in which a portion of the face is occluded matches the number of feature points extracted during user registration. For example, when the controller 150 extracts 40 feature points from the image in which a portion of the face is occluded and the number of feature points reduced is 35, the controller 150 can determine whether at least 35 of the 40 extracted feature points match the feature points extracted during user registration.

[0069] When the controller 150 determines that the feature points extracted from the image where part of the face is obscured match the feature points extracted during user registration, the controller 150 can identify the passenger as the user. That is, since the controller 150 determines that the passenger's current travel conditions meet the user's travel conditions, even when a small number of feature points are extracted from the image where part of the face is obscured, the controller 150 can easily identify the face and determine whether the face is the user's face.

[0070] When the controller 150 determines that the passenger is a user, it can set the vehicle settings information to the user's information. This vehicle settings information may include the steering wheel position, seat position, air conditioning temperature setting, rearview mirror position, radio receiving frequency, navigation settings, etc.

[0071] According to embodiments of this disclosure, even when a passenger is determined to be a user, the controller 150 can continue to learn the user's riding information during vehicle operation (after vehicle start is enabled until vehicle start is disabled), and can use an RNN for learning.

[0072] Figure 6 This is a flowchart illustrating a method for registering a passenger as a user according to an embodiment of the present disclosure.

[0073] like Figure 6 As shown, controller 150 acquires information sensed by sensor 120 to identify passenger boarding (S110). Additionally, it acquires boarding conditions from navigation 130 (S120). Here, boarding conditions may include the passenger's boarding location and boarding time (date (day)). Controller 150 acquires a full facial image of the passenger from camera 110 (S130). In S130, controller 150 can extract feature points from the acquired full facial image of the passenger.

[0074] The controller 150 uses extracted feature points to identify the passenger's face (S140). In S140, when the passenger's face is identified based on feature points extracted from the entire facial image, the controller 150 can determine that the face has been identified under normal conditions. In this case, the controller 150 can register the passenger as a user. When the passenger is registered as a user, the controller 150 registers the travel conditions obtained when the passenger boards the vehicle as the user's travel conditions (S150). In S150, when user registration and travel condition registration are completed, the controller 150 can control the storage of user registration information and user travel conditions in the storage device 140. Additionally, in S150, when user registration and travel condition registration are completed, the controller 150 can control the storage of the number of times the user has boarded the vehicle under the user's travel conditions as "once".

[0075] Figure 7 This is a flowchart illustrating a method for learning riding conditions according to an embodiment of the present disclosure.

[0076] like Figure 7 As shown, the controller 150 identifies the passenger's boarding based on information acquired from the sensor 120 (S210). Additionally, the controller 150 acquires boarding conditions for learning (S220). In S220, the controller 150 can acquire the passenger's boarding time and boarding location to learn the passenger's boarding conditions. Therefore, the boarding conditions used for learning can also be referred to as the boarding time and boarding location used for learning.

[0077] When the controller 150 acquires the riding conditions for learning, the controller 150 acquires a full facial image of the passenger (S230). In S230, the controller 150 can extract feature points from the acquired full facial image of the passenger.

[0078] Controller 150 uses extracted feature points to identify the passenger's face (S240). Then, controller 150 determines whether the passenger is a previously registered user based on the identified face (S250). When it is determined in S250 that the passenger is a previously registered user ("yes"), controller 150 determines whether the learned boarding conditions meet the user's boarding conditions (S260). In one example, when the passenger is not a previously registered user ("no") in S250, controller 150 can control the execution of S230.

[0079] In S260, controller 150 can determine the similarity between the learned boarding location and boarding time and the user's boarding location and boarding time, and based on the determination result, determine whether the learned boarding conditions meet the user's boarding conditions. (Refer to...) Figure 8 The operation of S260 is described in more detail.

[0080] When it is determined in S260 that the ride conditions used for learning meet the user's ride conditions ("Yes"), the controller 150 can control the addition of the number of times the passenger rides under the user's ride conditions by one, and store the added number (S270). On the other hand, when it is determined in S260 that the ride conditions used for learning do not meet the user's ride conditions ("No"), the controller 150 can control the registration of the ride conditions used for learning as the user's new ride conditions, and store the number of times the passenger rides under the user's new ride conditions as "one" (S280).

[0081] Controller 150 determines whether the number of times the passenger has ridden under the user's riding conditions exceeds a predetermined number N (e.g., 10 times) (S290). When it is determined in S290 that the number of times the passenger has ridden under the user's riding conditions exceeds the predetermined number ("yes"), controller 150 can determine that it has completed learning the user's riding conditions. On the other hand, when it is determined in S290 that the number of times the passenger has ridden under the user's riding conditions does not exceed the predetermined number N, controller 150 can control the execution of S220. According to an embodiment, a recurrent neural network (RNN) can be used to perform the operations from S220 to S280.

[0082] Figure 8 This is a flowchart illustrating a method for determining whether a user's travel conditions are met according to an embodiment of the present disclosure.

[0083] The controller 150 determines whether the learning location is within a predetermined distance from the user's location (S261). Here, the predetermined distance can be a radius of 500 meters. When it is determined in S261 that the learning location is within the predetermined distance from the user's location ("Yes"), the controller 150 can determine that the learning location meets the user's travel conditions (S260Y). On the other hand, when it is determined in S261 that the learning location is not within the predetermined distance from the user's location ("No"), the controller 150 determines that the learning location does not meet the user's travel conditions (S260N).

[0084] Additionally, controller 150 determines whether the travel time for learning is within a predetermined timeframe from the user's travel time (S262). Here, the predetermined time can be 60 minutes. Furthermore, when it is determined in S262 that the travel time for learning is within the predetermined timeframe from the user's travel time ("Yes"), controller 150 can determine that the travel conditions for learning meet the user's travel conditions (S260Y). On the other hand, when it is determined in S262 that the travel time for learning is not within the predetermined timeframe from the user's travel time ("No"), controller 150 determines that the travel conditions for learning do not meet the user's travel conditions (S260N).

[0085] Figure 9 This is a flowchart illustrating a facial recognition method according to an embodiment of the present disclosure.

[0086] like Figure 9 As shown, the controller 150 identifies the passenger's boarding based on information acquired from the sensor 120 (S310). Additionally, the controller 150 acquires the current boarding conditions from the navigation system 130 (S320). Here, the current boarding conditions may include the passenger's current boarding time and current boarding location.

[0087] Controller 150 determines whether the acquired current travel conditions meet the previously learned travel conditions (S330). In S330, controller 150 can determine whether the current travel location and current travel time meet the previously learned travel conditions for the user. Here, controller 150 can determine whether the passenger's current travel location is within a predetermined distance from the user's travel location, and can determine whether the passenger's current travel time is within a predetermined time from the user's travel time.

[0088] If, in S330, it is determined that the current travel conditions do not meet the previously learned travel conditions ("No"), the controller 150 allows relearning of the current travel conditions (S340). Conversely, if, in S330, it is determined that the current travel conditions meet the previously learned travel conditions ("Yes"), the controller 150 reduces the number of feature points used to identify the passenger's face (S350).

[0089] In S350, although the controller 150 can extract enough feature points from the entire face image to identify the face during user registration and ride condition learning operations, the controller 150 can reduce and reset the number of feature points when it is determined that the current ride condition meets the previously learned ride condition.

[0090] For example, when the number of feature points extracted from the entire facial image to identify a face is 70, the controller 150 can determine that at least 70 feature points are needed to identify a face. However, according to embodiments of this disclosure, when it is determined that the current riding conditions meet the previously learned riding conditions, the controller 150 can reduce the number of feature points used to identify the face and reset the number of feature points used to identify the face to 35.

[0091] Controller 150 acquires a partial facial image of the passenger (S360). Here, a partial facial image may refer to an image where a portion of the face is obscured, rather than an image of the entire face (see, for example, see...). Figure 3 and Figure 4 ).

[0092] The controller 150 extracts feature points based on an image where a portion of the face is occluded (S370). See reference... Figure 5 A detailed description is provided.

[0093] Controller 150 determines whether the feature points extracted from the image where a portion of the face is occluded match the feature points extracted during user registration (S380). In S380, controller 150 may determine whether the feature points extracted from the image where a portion of the face is occluded match the feature points extracted during user registration by reducing the number of feature points. For example, when controller 150 extracts 40 feature points from the image where a portion of the face is occluded and the number of feature points reduced is 35, controller 150 may determine whether at least 35 of the 40 extracted feature points match the feature points extracted during user registration.

[0094] When it is determined in S380 that the feature points extracted from the image where part of the face is occluded match the feature points extracted during user registration (“Yes”), the controller 150 can determine that the passenger is the user (S390). That is, since the controller 150 determines that the passenger's current travel conditions meet the learned travel conditions of the user, even when a small number of feature points are extracted from the image where part of the face is occluded, the controller 150 can easily identify the face and determine whether the face is the user's face.

[0095] When the passenger is identified as a user, the controller 150 can set the vehicle settings information to the user's information (S400). Here, the vehicle settings information may include the position of the steering wheel, the position of the seat, the air conditioning temperature setting, the position of the rearview mirror, the radio receiving frequency, navigation settings, etc.

[0096] According to embodiments of this disclosure, even when a passenger is determined to be a user, the controller 150 can continue to learn the user's riding information during vehicle operation (after vehicle start is enabled until vehicle start is disabled), and can use an RNN for learning.

[0097] Figure 10 A computing system is shown that performs a method according to an embodiment of the present disclosure.

[0098] Reference Figure 10 The computing system 1000 may include at least one processor 1100, a memory 1300, a user interface input device 1400, a user interface output device 1500, a storage device 1600, and a network interface 1700 connected to each other via a bus 1200.

[0099] Processor 1100 may be a central processing unit (CPU) or a semiconductor device that processes instructions stored in memory 1300 and / or storage device 1600. Memory 1300 and storage device 1600 may include various types of volatile or non-volatile storage media. For example, memory 1300 may include ROM (Read-Only Memory) 1310 and RAM (Random Access Memory) 1320.

[0100] Therefore, the operation of the methods or algorithms described in conjunction with the embodiments disclosed herein can be directly implemented in a hardware or software module run by processor 1100, or in a combination of hardware and software modules. The software module can reside in a storage medium such as RAM, flash memory, ROM, EPROM, EEPROM, registers, hard disk, removable disk, and CD-ROM (i.e., memory 1300 and / or storage device 1600). An exemplary storage medium can be coupled to processor 1100, which can read information from and record information in the storage medium. Optionally, the storage medium can be integrated with processor 1100. Processor 1100 and storage medium can reside in an application-specific integrated circuit (ASIC). The ASIC can reside within a user terminal. In another case, processor 1100 and storage medium can reside as separate components in the user terminal.

[0101] According to the embodiments of the present disclosure, the facial recognition device and facial recognition method can accurately identify users even when part of a passenger's face is obscured, by improving the facial recognition rate.

[0102] Therefore, exemplary embodiments of this disclosure are provided to explain the ideas and scope of this disclosure, and not to limit them. Consequently, the ideas and scope of this disclosure are not limited to the embodiments. The scope of this disclosure should be interpreted based on the appended claims, and all technical ideas equivalent to the scope of the claims should be included within the scope of this disclosure.

[0103] Although the present disclosure has been described above with reference to exemplary embodiments and accompanying drawings, the present disclosure is not limited thereto. Various modifications and changes can be made to the present disclosure by those skilled in the art without departing from the spirit and scope of the present disclosure as claimed in the appended claims.

Claims

1.A face recognition apparatus comprising: a camera that acquires a partial face image of a passenger; and a controller that learns a boarding condition of a user and recognizes a face based on the partial face image of the passenger when a current boarding condition of the passenger satisfies the learned boarding condition. 2.The apparatus according to claim 1, wherein the boarding condition includes information about a boarding location and a boarding time. 3.The apparatus according to claim 1, wherein the controller acquires a boarding condition of a first passenger and a first entire face image, extracts feature points from the first entire face image, and registers the first passenger as the user, registers the boarding condition of the first passenger as the boarding condition of the user, and stores a number of times that the user boards under the boarding condition of the user as "one". 4.The apparatus according to claim 1, wherein the controller acquires a boarding condition for learning of a second passenger and a second entire face image, and extracts feature points from the second entire face image to determine whether the second passenger is the user. 5.The apparatus according to claim 4, wherein when the second passenger is the user, the controller determines whether the boarding condition for learning satisfies the boarding condition of the user. 6.The apparatus according to claim 4, wherein when a boarding location for learning is within a predetermined distance from a boarding location of the user, or when a boarding time for learning is within a predetermined time from a boarding time of the user, the controller determines that the boarding condition for learning satisfies the boarding condition of the user. 7.The apparatus according to claim 4, wherein when the boarding condition for learning satisfies the boarding condition of the user, the controller adds one to a number of times that the user boards under the boarding condition of the user. 8.The apparatus according to claim 4, wherein when the boarding condition for learning does not satisfy the boarding condition of the user, the controller registers the boarding condition for learning as a new boarding condition of the user, and stores a number of times that the user boards under the new boarding condition as "one". 9.The apparatus according to claim 1, wherein when the number of times that the user boards under the boarding condition of the user exceeds a predetermined number of times, the controller determines that learning of the boarding condition of the user is completed. 10.The apparatus according to claim 1, wherein when the current boarding condition of the passenger satisfies the learned boarding condition of the user, the controller reduces a number of feature points used to recognize a face. 11.The apparatus according to claim 1, wherein when feature points extracted from the partial face image of the passenger match feature points extracted from the user, the controller determines that the passenger is the user. 12.A face recognition method of a user, comprising: learning a boarding condition of the user; acquiring a partial face image of a passenger; and recognizing a face based on the partial face image of the passenger when a current boarding condition of the passenger satisfies the learned boarding condition of the user. ​ ​ 13.The method of claim 12, wherein, the boarding condition includes information about a boarding location and a boarding time. 14.The method of claim 12, wherein, the boarding condition of the user is developed by: obtaining a boarding condition of a first passenger and a first full face image; extracting a plurality of feature points from the first full face image and registering the first passenger as the user; and registering the boarding condition of the first passenger as the boarding condition of the user and storing a number of times the user boards under the boarding condition of the user as “one”. 15.The method of claim 12, wherein, the boarding condition of the user is developed by: obtaining a boarding condition for learning of a second passenger and a second full face image; extracting feature points from the second full face image to determine whether the second passenger is the user; when the second passenger is the user, determining whether the boarding condition for learning satisfies the boarding condition of the user; when the boarding condition for learning satisfies the boarding condition of the user, adding one to a number of times the user boards under the boarding condition of the user; determining whether the number of times the user boards under the boarding condition of the user exceeds a predetermined number of times; and when the number of times the user boards under the boarding condition of the user exceeds the predetermined number of times, determining that the boarding condition of the user is developed. 16.The method of claim 15, wherein, determining whether the boarding condition for learning satisfies the boarding condition of the user includes: when a boarding location for learning is within a predetermined distance from a boarding location of the user, or when a boarding time for learning is within a predetermined time from a boarding time of the user, determining that the boarding condition for learning satisfies the boarding condition of the user. 17.The method of claim 15, further comprising: when the boarding condition for learning does not satisfy the boarding condition of the user, registering the boarding condition for learning as a new boarding condition of the user; and storing a number of times the user boards under the new boarding condition as “one”. 18.The method of claim 12, wherein, before recognizing a face based on a partial face image of the passenger, the method further comprises: when a current boarding condition of the passenger satisfies the boarding condition of the user for learning, reducing a number of feature points used to recognize the face. 19.The method of claim 12, wherein, recognizing a face based on a partial face image of the passenger includes: extracting feature points from the partial face image of the passenger; determining whether the feature points extracted from the partial face image of the passenger match feature points extracted when the user is registered; and when the feature points extracted from the partial face image of the passenger match the feature points extracted when the user is registered, determining that the passenger is the user. 20.A face recognition method, comprising: obtaining a face image of a passenger who boards, wherein the face image is a partial face image; ​ ​ acquiring current ride condition information of the passenger; comparing the current ride condition information of the passenger with learned ride conditions of a registered user; comparing a face image of the passenger with face images of the registered user; and judging whether the passenger is the registered user.

Images