KR20170082078A - Apparatus and method for recognizing identity - Google Patents
Apparatus and method for recognizing identity Download PDFInfo
- Publication number
- KR20170082078A KR20170082078A KR1020160001249A KR20160001249A KR20170082078A KR 20170082078 A KR20170082078 A KR 20170082078A KR 1020160001249 A KR1020160001249 A KR 1020160001249A KR 20160001249 A KR20160001249 A KR 20160001249A KR 20170082078 A KR20170082078 A KR 20170082078A
- Authority
- KR
- South Korea
- Prior art keywords
- face
- recognition
- identity
- recognition result
- person
- Prior art date
Links
Images
Classifications
-
- G06K9/00369—
-
- G06F17/30259—
-
- G06K9/00228—
-
- G06K9/6201—
-
- G06K9/64—
Landscapes
- Collating Specific Patterns (AREA)
Abstract
An apparatus and method for identifying an identity are disclosed. The apparatus for recognizing an identity according to an embodiment of the present invention extracts a person's body from an image captured using a camera, and uses the result of comparing the body information and the clothing information existing in the database, A clothes recognizing unit for performing the clothes recognition; A face recognition unit for recognizing the face using a result of comparing the face of the person extracted from the image with face information existing in the database; And judging whether to use at least one of the recognition result of the clothes and the face recognition result based on the size of the value corresponding to the recognition result of the face, and recognizing the identity of the person based on the result of the determination And the like.
Description
TECHNICAL FIELD The present invention relates to a technology for recognizing a person's identity, and more particularly, to a technique for recognizing a person's identity by using both the recognition result of a garment worn by a person and the face recognition result.
Generally, the most promising user recognition technology used in indoor environments is a biometric method that utilizes the user's physical characteristics. Several biometric methods are used to recognize real users, for example, face recognition, speaker recognition, and gait recognition.
Of these biometric methods, facial recognition is realized as a method for recognizing the user because it has the merit of being capable of remote recognition without requiring direct contact with the user.
Face recognition has been studied for a very long time with great interest. Especially, since the introduction of the appearance based matching method in the early 1990's, the recognition performance has been remarkably improved, and many commercial systems related to face recognition have been released.
However, the conventional face recognition technology has a fatal weakness that is fundamentally vulnerable to changes in illumination, posture, facial expression, and face change over time. Existing research results that have been reported to exhibit superior performance have been recognized assuming a well-controlled environment, such as certain lighting conditions and frontal faces and no-faces. To satisfy these assumptions, the user must always look at the robot with no expression from the front of the device, the illumination environment similar to the state at the time of initial user registration template creation must be maintained at all times, and the face change over time In order to do this, the user must regularly regenerate the user template.
In addition, there is a disadvantage in that the reliability of the face recognition must be high only if the quality of the photographed image is higher than a specific quality for face recognition. That is, when the distance between the recognition device and the person is long, or when the quality of the camera is low, face recognition can not be performed.
Korean Patent Laid-Open Publication No. 2014-0001164 discloses a technique of recognizing a face. However, all the prior arts including Korean Patent Laid-Open Publication No. 2014-0001164 disclose technologies such as reducing the quality of an image or correcting an image to increase the recognition rate even when the distance is long, But does not disclose a technique for increasing the recognition rate.
Therefore, considering the recent security issues, there is an increasing need for techniques for recognizing person identities that can guarantee the recognition rate regardless of the quality and distance of the images.
It is an object of the present invention to recognize the identity of a person regardless of the quality of the image of a camera that photographs a person.
It is also an object of the present invention to recognize the identity of a person regardless of the distance between the camera and the person.
According to an aspect of the present invention, there is provided an apparatus for recognizing an identity of a person, which extracts a person's body from an image captured using a camera, Clothing recognition part performing; A face recognition unit for recognizing the face using a result of comparing the face of the person extracted from the image with face information existing in the database; And judging whether to use at least one of the recognition result of the clothes and the face recognition result based on the size of the value corresponding to the recognition result of the face, and recognizing the identity of the person based on the result of the determination And the like.
At this time, if the size of the value corresponding to the recognition result of the face is smaller than a predetermined value, the identity recognizing unit can recognize the identity by using both the face recognition result and the clothing recognition result.
In this case, if the size of the value corresponding to the recognition result of the face is equal to or larger than a predetermined value, the identity recognizing unit can recognize the identity using only the face recognition result.
At this time, the identity recognizing unit can control the predetermined value based on the environment of the place where the human identity recognizing apparatus is installed.
At this time, if the size of the value corresponding to the quality of the image is smaller than a predetermined value, the identity recognizing unit can recognize the identity using both the face recognition result and the clothes recognition result.
At this time, the clothing recognizing unit extracts clothing information corresponding to the season corresponding to the date, on which the clothing is recognized, from the database, and performs recognition of the clothes using the clothing information have.
In this case, the identity recognizing device may further include an apparel information registering unit for storing the apparel information extracted from the image photographed using the camera, in the database.
In this case, if the size of the value corresponding to the recognition result of the face is larger than the predetermined value in the image, the clothing information registration unit may store the clothing information in the database.
According to another aspect of the present invention, there is provided a method of recognizing an identity, comprising: extracting a human body from an image captured using a camera; Performing recognition of the garment based on a result of comparing clothing corresponding to the whole body and clothing information existing in a database; Performing face recognition based on a result of comparing a face corresponding to the whole body and face information existing in the database; And judging whether to use at least one of the recognition result of the clothes and the face recognition result based on the size of the value corresponding to the recognition result of the face, and recognizing the identity of the person based on the result of the determination .
At this time, the step of recognizing the identity of the person recognizes the identity by using both the face recognition result and the clothes recognition result when the size of the value corresponding to the face recognition result is smaller than a preset value can do.
In this case, when the size of the value corresponding to the face recognition result is equal to or larger than a preset value, the recognition of the person can recognize the identity using only the face recognition result.
At this time, the step of recognizing the identity of the person can control the preset value based on the environment of the environment where the person identity recognition apparatus is installed.
At this time, the step of recognizing the identity of the person recognizes the identity using both the face recognition result and the clothes recognition result when the size of the value corresponding to the image quality is smaller than a predetermined value .
At this time, the step of recognizing the clothes may include extracting, from the database, clothing information corresponding to the season corresponding to the date, a date on which the clothing is recognized, Can be performed.
In this case, the identity recognition method may further include storing the clothing information and the face information extracted from the image captured using the camera in the database.
In this case, the step of storing in the database may store the clothing information in the database if the size of the value corresponding to the recognition result of the face is larger than a predetermined value.
According to the present invention, when the quality of an image is low, the identity of a person can be recognized using both the result of recognition of a person's clothing and the result of recognition of a face, and the identity of a person can be recognized regardless of the quality of the image.
In addition, even if the distance between the camera and the person is long, the present invention recognizes the identity of a person using both the result of recognition of a person's clothing and the result of recognition of a face, so that a person's identity can be recognized regardless of the distance between the camera and a person .
1 is a block diagram illustrating a person identification apparatus according to an exemplary embodiment of the present invention.
2 is a block diagram of a person identity recognition apparatus according to an embodiment of the present invention.
3 to 5 are diagrams illustrating a process of recognizing a person's body and face by the person identification apparatus according to an embodiment of the present invention.
FIG. 6 is a diagram illustrating clothing recognition in a person identity recognition apparatus according to an exemplary embodiment of the present invention. Referring to FIG.
FIG. 7 is a diagram illustrating a process of storing clothing information in a database in a person identity recognition apparatus according to an embodiment of the present invention.
FIGS. 8 to 9 are pseudo codes that implement the process of recognizing the identity of the person identification apparatus according to an embodiment of the present invention.
10 is a flowchart illustrating a method of recognizing a person identity according to an exemplary embodiment of the present invention.
The present invention will now be described in detail with reference to the accompanying drawings. Hereinafter, a repeated description, a known function that may obscure the gist of the present invention, and a detailed description of the configuration will be omitted. Embodiments of the present invention are provided to more fully describe the present invention to those skilled in the art. Accordingly, the shapes and sizes of the elements in the drawings and the like can be exaggerated for clarity.
Hereinafter, preferred embodiments according to the present invention will be described in detail with reference to the accompanying drawings.
1 is a block diagram illustrating a person identification apparatus according to an exemplary embodiment of the present invention.
Referring to FIG. 1, an
At this time, the whole body of the person is extracted from the image taken by the camera, and the recognition of the clothes and the face of the person is performed.
At this time, based on the face recognition result, the
2, the
2 is a block diagram of a person identity recognition apparatus according to an embodiment of the present invention.
Referring to FIG. 2, the
The
At this time, the
At this time, the
At this time, the reason why the person is continuously tracked is that the
At this time, the
At this time, the
For example, if the current date is October 29, it usually corresponds to fall. Accordingly, it is possible to extract garment information corresponding to a date corresponding to fall in the database (for example, from September 15 to November 15).
At this time, the garment information means a garment worn by a person who performs identification.
The detailed process of garment recognition is described in FIG.
The
The
At this time, in the case of the face recognition result, the reliability of the recognition result is higher than the recognition result of the clothes. However, face recognition results are not as reliable as recognition methods such as RFID and fingerprint recognition. Therefore, the present invention discloses a configuration in which, when the value of the face recognition result is greater than or equal to a specific value, recognition of the person is performed using only the face recognition result. However, when the value of the face recognition result is less than the specific value, the recognition of the face is performed using both the face recognition result and the clothing recognition result.
At this time, the garment may be different depending on the space. However, in the case of a home or an office, since the kind of clothes worn by a person is limited, the recognition rate can be increased when the recognition results of the clothes are used in parallel .
Accordingly, the
Also, when the size of the value corresponding to the quality of the image is smaller than the predetermined value, the
Although not shown in FIG. 2, the apparatus for recognizing a human identity according to an exemplary embodiment of the present invention further includes an apparel information registration unit (not shown) for storing apparel information extracted from a video shot using a camera in a database You may.
At this time, the clothing information registration unit can store the clothing information in the database only when the size of the value corresponding to the face recognition result in the image is larger than the predetermined value. This will be described in detail in Fig.
The description of FIG. 2 will be summarized again, and a method of performing identity recognition according to time sequence will be described as follows.
First, the whole body detection of the person is performed while continuing to receive the camera image in the start state. Next, when a person appears in front of the camera and the whole body detection of the person is completed, the process goes to the next step to perform the clothing recognition.
At this time, in the garment recognition state, garment information is extracted from the detected image information of the person to perform garment based identity recognition.
At this time, when recognition of clothes-based identity is completed, tracking of a person is performed (as shown in FIG. 3 and FIG. 4, a person's body can be continuously tracked).
At this time, the user continuously tries to detect the face of the person being tracked, and when the detection of the face is completed, the face recognition is performed.
At this time, when the face recognition is completed, the final final recognition state is entered. Finally, the final recognition result is obtained by combining the results of the face recognition and the clothing recognition performed previously.
3 to 5 illustrate the tracking of the whole body and face of a person in the form of a quadrangle. However, it is needless to say that they can be tracked not only in the shape of a quadrangle, but also in various shapes.
FIG. 6 is a diagram illustrating clothing recognition in a person identity recognition apparatus according to an exemplary embodiment of the present invention. Referring to FIG.
First, the characteristics of the clothes are extracted from the input image (S610).
In this case, HSV and LAB color histograms are used to express the color information of clothes, and LBP features are used to express the texture pattern of clothes. After combining the color histogram and the LBP feature, the feature vector is finally extracted from the single image through the PCA process.
At this time, the features of the clothes can be extracted using the extracted feature vectors.
Further, the garment information is extracted from the database (S620).
At this time, the garment information means information of the garment worn by the person who performs the identity recognition. For example, information about a coat that a person wears may include colors, materials, patterns, and the like.
In addition, the clothing feature and the clothing information are compared (S630), and the recognition result of the clothes using the comparison result is extracted (S640).
At this time, the extracted feature vector is compared with the clothing information stored in the database, and the recognition result of the clothes is derived.
At this time, if the inputted whole body image x is represented by P (C i | x) as a probability that the input body image x is the image for a specific user C i , it is expressed as follows.
P clothes (C i | x) = max (re-id (imgs_in_db (C i ), x) / Z
The Imgs_in_db function is a function for extracting images of a user (C i ) from a database.
At this time, it is possible to extract only the images shot within two months on the present date, reflecting that the clothes to be worn are different according to the seasonal factors.
Also, even if the year is different, it can also include images taken at similar times. In particular, it may be included only if winter is similar, and may be excluded if summer is similar. This is derived from the results of reflecting the characteristics of wearing winter clothes for a long time. On the contrary, in the case of summer costume, it is derived from the result that reflects the characteristic that changes frequently (in the case of a t-shirt, it means that there are few cases to wear over several years).
Also, the re_id (a, b) function is a function for extracting and comparing clothing features for the input images (a, b).
Also, Z means a normalization parameter such that the P garment (C i | x) has a value between 0 and 1.
The above expression is briefly described again. The images of the users stored in the database are compared with the input images. However, only the images of similar dates on the date excluding the year are targeted, and the highest clothing recognition score Is used as P clothes (C i | x).
FIG. 7 is a diagram illustrating a process of storing clothing information in a database in a person identity recognition apparatus according to an embodiment of the present invention.
First, the whole body is detected in the image (S710), and the face is continuously detected by continuously tracking the user (S720).
In addition, the face recognition is performed based on the detected face (S730), and it is confirmed whether the confidence value of the face recognition result is greater than a predetermined value (S740).
At this time, only when the reliability value is larger than the predetermined value, the clothing information corresponding to the whole body is stored in the database (S750). That is, when the face recognition is performed, only when the confidence of the face recognition is very high, the clothing information of the person is registered in the database. .
FIGS. 8 to 9 are pseudo codes that implement the process of recognizing the identity of the person identification apparatus according to an embodiment of the present invention.
The pseudo code shown in Fig. 8 shows a pseudo code used for recognizing the identity in the
Referring to FIG. 8, it can be seen that the predetermined value is shown as 0.9.
At this time, when the size of the value corresponding to the recognition result of the face is larger than the preset value (0.9), it can be seen that identity recognition is performed using only the face recognition result.
Also, when the size of the value corresponding to the recognition result of the face is smaller than the preset value (0.9), it is understood that the recognition of the face is performed by using both the face recognition result and the clothing recognition result.
9 also shows a numeric code used for recognizing the identity in the
In the case of FIG. 9, when the size of the value corresponding to the recognition result of the face is larger than the preset value (0.9), it can be seen that identity recognition is performed using only the face recognition result.
In addition, although the value corresponding to the face recognition result is larger than the value (0.7) set differently, it is smaller than the predetermined value (0.9), the value corresponding to the recognition result of the clothes is larger than the preset value (0.9) If the recognition result of the identified subject and the recognition result of the face are the same, it can be seen that the recognition recognition is performed using the recognition result of the clothes.
In Fig. 9, c_hat indicates the ID of the object recognized as the final identity recognition result. In addition, 0.7 and 0.9, which are illustrated by preset values, can be controlled as much as possible according to the field environment. In summary, when the reliability of face recognition is high, the recognition result is derived using only face recognition. In the case where the reliability of face recognition is not high, but the reliability of clothing recognition is high, . ≪ / RTI >
10 is a flowchart illustrating a method of recognizing a person identity according to an exemplary embodiment of the present invention.
First, a whole body of a person is extracted from the inside of the image (S1010).
In addition, clothing information corresponding to the whole body is compared with clothing information of the database, and clothing recognition is performed using the comparison result (S1020).
At this time, in extracting the clothing information from the database, it is possible to extract the clothing information corresponding to the date corresponding to the date and the date corresponding to the date. For example, if the current date is October 29, it usually corresponds to fall. Accordingly, it is possible to extract garment information corresponding to a date corresponding to fall in the database (for example, from September 15 to November 15).
Further, face information corresponding to the whole body is compared with face information of the database, and face recognition is performed (S1030).
Also, the recognition of the face is performed using the face recognition result and the clothes recognition result (S1040).
At this time, in the case of the face recognition result, the reliability of the recognition result is higher than the recognition result of the clothes. However, face recognition results are not as reliable as recognition methods such as RFID and fingerprint recognition. Therefore, the present invention discloses a configuration in which, when the value of the face recognition result is greater than or equal to a specific value, recognition of the person is performed using only the face recognition result. However, when the value of the face recognition result is less than the specific value, the recognition of the face is performed using both the face recognition result and the clothing recognition result.
At this time, the garment may be different depending on the space. However, in the case of a home or an office, since the kind of clothes worn by a person is limited, the recognition rate can be increased when the recognition results of the clothes are used in parallel .
Accordingly, the specific value can be controlled based on the environment of the place where the
Further, when the value of the value corresponding to the quality of the image is smaller than the preset value, the identity can be recognized using both the face recognition result and the clothing recognition result. In other words, if the quality of the image is high, identification of the face can be performed using only the face recognition result. However, if the quality is low, it is difficult to recognize the face shape, . Therefore, when the quality of the image is lower than a specific value, the present invention can perform the identity recognition using both the recognition result of the clothes and the recognition result of the face.
The use of the apparatus and method for recognizing an identity according to an embodiment of the present invention may increase the reactivity of the personalization service.
In one embodiment, in the case of a robot that grasps a user and grants a user first, when only face recognition is used, the user can react only when the distance between the user and the robot is below a certain level. The recognition rate is lower than that of the face recognition, so it can not be used where the reliability of the recognition result is important. However, in areas where service responsiveness is more important than precision recognition rate such as entertainment and personal service, it is possible to use face recognition and clothing awareness sufficiently.
According to another embodiment, in the face recognition based access management field, when only face information is used, the quality of the image must be considerable. However, when the identity recognition is performed using both the clothing information and the face information disclosed in the present invention, it is possible to perform identity recognition that ensures reliability even in a variety of situations.
Also, in another embodiment, the identity of the user facing the robot or appliance present in the holder can be determined. In the case of a robot, it can identify the user and perform tasks such as mail and schedule notification. In the case of TV, audience identity information may be searched for and preferred viewer information may be collected for channel search or viewing information analysis. Especially, it is difficult to capture the face of the face in the situation where the camera frequently moves like a robot. Therefore, it is possible to more accurately recognize the identity through clothing recognition. In addition, in the case of a relatively small group of family members, since the types of clothes to be worn in the home do not vary, the credibility of clothes recognition can be further increased.
As described above, the apparatus and method for recognizing an identity according to the present invention are not limited to the configuration and method of the embodiments described above, but the embodiments can be applied to all of the embodiments Or some of them may be selectively combined.
Claims (16)
A face recognition unit for recognizing the face using a result of comparing the face of the person extracted from the image with face information existing in the database; And
Determining whether to use at least one of the recognition result of the clothes and the face recognition result based on the size of the value corresponding to the recognition result of the face, and recognizing the identity of the person based on the result of the determination Identity recognition unit
Wherein the identification information includes at least one of identification information and identification information.
The identity recognition unit
And recognizes the identity using both the recognition result of the face and the recognition result of the clothes when the size of the value corresponding to the face recognition result is smaller than a predetermined value.
The identity recognition unit
Wherein the recognition unit recognizes the identity using only the recognition result of the face when the size of the value corresponding to the recognition result of the face is equal to or larger than a predetermined value.
The identity recognition unit
And the predetermined value is controlled based on an environment of a place where the human person identification apparatus is installed.
The identity recognition unit
And recognizes the identity using both the face recognition result and the clothing recognition result when the value of the value of the image is smaller than a predetermined value.
The clothes recognizing unit
Extracting clothing information corresponding to the season corresponding to the date, performing a recognition of the clothes, and performing recognition of the clothes using the clothing information.
The identification device
Further comprising an apparel information registration unit for storing the apparel information extracted from the image captured using the camera in the database.
The clothing information registration unit
And stores the clothing information in the database when the size of the value corresponding to the recognition result of the face is larger than the predetermined value in the image.
Performing recognition of the garment based on a result of comparing clothing corresponding to the whole body and clothing information existing in a database;
Performing face recognition based on a result of comparing a face corresponding to the whole body and face information existing in the database; And
Determining whether to use at least one of the recognition result of the clothes and the face recognition result based on the size of the value corresponding to the recognition result of the face, and recognizing the identity of the person based on the result of the determination The method comprising the steps of:
The step of recognizing the identity of the person
And recognizing the identity using both the recognition result of the face and the recognition result of the clothes when the size of the value corresponding to the face recognition result is smaller than a predetermined value.
The step of recognizing the identity of the person
Wherein if the size of the value corresponding to the recognition result of the face is equal to or larger than a predetermined value, the identity is recognized using only the recognition result of the face.
The step of recognizing the identity of the person
Wherein the predetermined value is controlled based on an environment of a place of the environment in which the person identification apparatus is installed.
The step of recognizing the identity of the person
If the size of the value corresponding to the quality of the image is smaller than a preset value, recognizes the identity using both the face recognition result and the clothing recognition result.
The step of recognizing the garment
Extracting clothing information corresponding to a season corresponding to the date from the database, and performing recognition of the clothes using the clothing information.
The identity recognition method
And storing the clothing information and the face information extracted from the image captured by the camera in the database.
The step of storing in the database
Wherein if the size of the value corresponding to the recognition result of the face is larger than the size of the preset value in the image, the clothing information is stored in the database.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| KR1020160001249A KR20170082078A (en) | 2016-01-05 | 2016-01-05 | Apparatus and method for recognizing identity |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| KR1020160001249A KR20170082078A (en) | 2016-01-05 | 2016-01-05 | Apparatus and method for recognizing identity |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| KR20170082078A true KR20170082078A (en) | 2017-07-13 |
Family
ID=59352610
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| KR1020160001249A KR20170082078A (en) | 2016-01-05 | 2016-01-05 | Apparatus and method for recognizing identity |
Country Status (1)
| Country | Link |
|---|---|
| KR (1) | KR20170082078A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109465819A (en) * | 2017-09-08 | 2019-03-15 | 株式会社日立大厦系统 | Human body recognition system and human body recognition method |
| KR102012672B1 (en) * | 2019-01-17 | 2019-08-21 | 대주씨앤에스 주식회사 | Anti-crime system and method using face recognition based people feature recognition |
| CN110334564A (en) * | 2019-03-18 | 2019-10-15 | 特斯联（北京）科技有限公司 | A kind of permanent resident population's recognition methods and system based on target following |
| KR20200063705A (en) * | 2018-11-28 | 2020-06-05 | 초록소프트 주식회사 | Method and apparatus for implementing flow populent data collecting algorithm using real time load |
| US20230115071A1 (en) * | 2021-10-08 | 2023-04-13 | Openit Inc. | Method of verifying target person, and server and program |
-
2016
- 2016-01-05 KR KR1020160001249A patent/KR20170082078A/en unknown
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109465819A (en) * | 2017-09-08 | 2019-03-15 | 株式会社日立大厦系统 | Human body recognition system and human body recognition method |
| CN109465819B (en) * | 2017-09-08 | 2022-04-05 | 株式会社日立大厦系统 | Human body recognition system and human body recognition method |
| KR20200063705A (en) * | 2018-11-28 | 2020-06-05 | 초록소프트 주식회사 | Method and apparatus for implementing flow populent data collecting algorithm using real time load |
| KR102012672B1 (en) * | 2019-01-17 | 2019-08-21 | 대주씨앤에스 주식회사 | Anti-crime system and method using face recognition based people feature recognition |
| CN110334564A (en) * | 2019-03-18 | 2019-10-15 | 特斯联（北京）科技有限公司 | A kind of permanent resident population's recognition methods and system based on target following |
| CN110334564B (en) * | 2019-03-18 | 2020-04-24 | 特斯联（北京）科技有限公司 | Frequent population identification method and system based on target tracking |
| US20230115071A1 (en) * | 2021-10-08 | 2023-04-13 | Openit Inc. | Method of verifying target person, and server and program |
| KR20230050903A (en) * | 2021-10-08 | 2023-04-17 | 주식회사 오픈잇 | Method for verifying the target person, and server and program using the same |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9323979B2 (en) | Face recognition performance using additional image features | |
| KR20170082078A (en) | Apparatus and method for recognizing identity | |
| US10127439B2 (en) | Object recognition method and apparatus | |
| US8135220B2 (en) | Face recognition system and method based on adaptive learning | |
| US20080298643A1 (en) | Composite person model from image collection | |
| CN102799870B (en) | Based on the single training image per person method of the consistent LBP of piecemeal and sparse coding | |
| US8116534B2 (en) | Face recognition apparatus and face recognition method | |
| Wang et al. | InSight: recognizing humans without face recognition | |
| KR20160088224A (en) | Method for recognizing an object and apparatus thereof | |
| WO2009048513A2 (en) | Apparel as event marker | |
| JP5726596B2 (en) | Image monitoring device | |
| JP2014016968A (en) | Person retrieval device and data collection device | |
| Tsai et al. | Enabling identity-aware tracking via fusion of visual and inertial features | |
| Bagherian et al. | Facial feature extraction for face recognition: a review | |
| Nambiar et al. | Frontal gait recognition combining 2D and 3D data | |
| Denman et al. | Can you describe him for me? a technique for semantic person search in video | |
| KR20110046874A (en) | Method of scanning and recognizing an image | |
| Li et al. | Biometrics at a distance: issues, challenges, and prospects | |
| Denman et al. | Identifying customer behaviour and dwell time using soft biometrics | |
| Heo | Fusion of visual and thermal face recognition techniques: A comparative study | |
| Abbasnejad et al. | A hierarchical bayesian network for face recognition using 2d and 3d facial data | |
| JP2002208011A (en) | Image collation processing system and its method | |
| Alletto et al. | Egocentric object tracking: an odometry-based solution | |
| Ranganarayana et al. | A Study on Approaches For Identifying Humans In Low Resolution Videos | |
| Ling et al. | Analyzing human movements from silhouettes via Fourier descriptor |