CN106503686A - The method and system of retrieval facial image - Google Patents
The method and system of retrieval facial image Download PDFInfo
- Publication number
- CN106503686A CN106503686A CN201610973398.6A CN201610973398A CN106503686A CN 106503686 A CN106503686 A CN 106503686A CN 201610973398 A CN201610973398 A CN 201610973398A CN 106503686 A CN106503686 A CN 106503686A
- Authority
- CN
- China
- Prior art keywords
- facial image
- characteristic point
- retrieved
- facial
- retrieval
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
Abstract
The present invention relates to a kind of method and system of retrieval facial image, methods described includes：According to the characteristic point position information that default face characteristic point model determines facial image to be retrieved；According to the facial characteristics that characteristic point position information is calculated facial image to be retrieved；Combination facial characteristics, obtain the characteristic vector of facial image to be retrieved；According to the facial image pre-saved in characteristic vector searching database, obtain and facial image similarity highest facial image to be retrieved.According to the characteristic point position information that default face characteristic point model determines facial image to be retrieved, the characteristic vector of facial image to be retrieved is then obtained according to the characteristic point position information for determining；As the characteristic point of the facial image to be retrieved for determining is limited, the characteristic point position information of the facial image to be retrieved of determination is the coordinate information of two dimension.So, rapidly and efficiently can be retrieved from the facial image of magnanimity and facial image similarity highest facial image to be retrieved according to characteristic vector.
Description
Technical field
The present invention relates to image identification technical field, more particularly to a kind of method and system of retrieval facial image.
Background technology
In recent years, with the swift and violent growth of the Internet, Internet picture explosive increase and safety monitoring equipment day
Benefit popularization, can all produce the face image data of magnanimity daily, and in so extensive face database, quick-searching is to oneself
Some people face image interested has become a urgent demand.
At present, human face detection and recognition technology is used widely in each field, becomes a current study hotspot.Phase
Search for like face, be a given face to be found, will look for from the image data base comprising hundreds thousand of even more plurality of human faces
The result similar to its appearance is arrived, and returns the sequence of pictures according to its similarity degree sequence.Facial image number in the face of magnanimity
According to needing to carry out human face data effectively tissue index and search analysis, so as to efficiently search for facial image.Traditional side
Method is to extract the complicated feature of the higher-dimensions such as the LBP features of facial image, ORB features and want the whole face database of linear sweep to find
Most like face, retrieval rate are slow.
Content of the invention
It is based on this, there is provided a kind of method and system of retrieval facial image, to overcome the slow-footed problem of face retrieval.
A kind of method of retrieval facial image, including：Face figure to be retrieved is determined according to default face characteristic point model
The characteristic point position information of picture；According to the facial characteristics that the characteristic point position information is calculated facial image to be retrieved；Group
The facial characteristics are closed, the characteristic vector of facial image to be retrieved is obtained；According to advance in the characteristic vector searching database
The facial image of preservation, obtains and facial image similarity highest facial image to be retrieved.
For the deficiency of conventional art, a kind of system of retrieval facial image is also provided.
A kind of system of retrieval facial image, including：Characteristic point position information determination module, facial characteristics computing module,
Characteristic vector acquisition module and retrieval module；The characteristic point position information determination module, for according to default face characteristic
Point model determines the characteristic point position information of facial image to be retrieved；The facial characteristics computing module, for according to the spy
Levy the facial characteristics that dot position information is calculated facial image to be retrieved；The characteristic vector acquisition module, for combining
Facial characteristics are stated, the characteristic vector of facial image to be retrieved is obtained；The retrieval module, for retrieving according to the characteristic vector
The facial image pre-saved in data base, obtains and facial image similarity highest facial image to be retrieved.
The beneficial effect of this programme：According to the characteristic point position that default face characteristic point model determines facial image to be retrieved
Confidence ceases, and then obtains the characteristic vector of facial image to be retrieved according to the characteristic point position information for determining；Due to treating for determining
Retrieval facial image characteristic point be limited, and determine facial image to be retrieved characteristic point position information be two dimension seat
Mark information.So, rapidly and efficiently from data base can be retrieved in the facial image of magnanimity and be treated according to the characteristic vector
Retrieval facial image similarity highest facial image.
Description of the drawings
Fig. 1 is the indicative flowchart of the method for the retrieval facial image of an embodiment；
Fig. 2 is the indicative flowchart of the method for the retrieval facial image of another embodiment；
Human face characteristic point schematic diagrams of the Fig. 3 for Fig. 2 embodiments；
Fig. 4 is the schematic diagram of the system of the retrieval facial image of an embodiment.
Specific embodiment
In order to further illustrate the effect of the technological means that is taken of the invention and acquirement, below in conjunction with the accompanying drawings and preferably
Embodiment, to technical scheme, carries out clear and complete description.
Fig. 1 is the indicative flowchart of the method for the retrieval facial image of an embodiment.As shown in figure 1, a kind of retrieval people
The method of face image, including：
S101, according to the characteristic point position information that default face characteristic point model determines facial image to be retrieved.
In the present embodiment, the determination method of default face characteristic point model is face contour, the eyebrow in facial image
Hair wheel exterior feature, nose profile, eye contour and face profile etc. determine multiple characteristic points.When Face datection is carried out, can adopt
ASM (Active Shape Model, active shape model), AAM (Active Appreance Model, active appearance models)
Or DLIB modes are according to the characteristic point of default human face characteristic point model extraction facial image to be retrieved.Wherein DLIB is a machine
The C++ storehouses of device study, contain the conventional algorithm of many machine learning.The present embodiment extracts facial image to be retrieved with DLIB
Human face characteristic point, then obtains the characteristic point position information of the facial image to be retrieved for having extracted；The face figure to be retrieved for extracting
The human face characteristic point of picture can be the characteristic point of face contour, the characteristic point of eyebrow outline, the characteristic point of nose profile, eyes wheel
At least 2 category feature points in the characteristic point of wide characteristic point and face profile, generally, by facial image to be retrieved
5 category feature points are all extracted.Except determining characteristic point position information, it may also be determined that going out face location information.
S102, according to the facial characteristics that the characteristic point position information is calculated facial image to be retrieved.
In the present embodiment, the characteristic point position information is two-dimensional coordinate information, by simple four arithmetic operation, can be with
The facial characteristics of to be retrieved facial image are quickly and easily obtained, and the facial characteristics include shape of face, eyebrow type, chin type, eyebrow
Tail sag, nose length-width ratio, looks away from, eyebrow peak, inside and outside palpebral fissure line and horizontal line angle, eyebrow within angle and eye wide.
S103, combines the facial characteristics, obtains the characteristic vector of facial image to be retrieved.
In the present embodiment, characteristic vector includes several elements, and each element represents each facial characteristics.Such as, feature to
Measure and beWherein first element a1Shape of face is represented, second element a2Eyebrow type is represented,
3rd element a3Represent eyebrow tail sag, the 4th element a4Represent eyebrow nose length-width ratio, the 5th element a5Represent looks away from,
By that analogy.
S104, according to the facial image pre-saved in the characteristic vector searching database, obtains and face to be retrieved
Image similarity highest facial image.
The facial image pre-saved in data base, usually thousands of magnanimity facial image, according to the feature
The facial image pre-saved in vector index data base, obtains and facial image similarity highest facial image to be retrieved,
As most like with facial image to be retrieved facial image.
According to the characteristic point position information that default face characteristic point model determines facial image to be retrieved, then according to really
Fixed characteristic point position information obtains the characteristic vector of facial image to be retrieved；Feature due to the facial image to be retrieved of determination
Point is limited, and the characteristic point position information of the facial image to be retrieved of determination is two-dimensional coordinate information.So, according to the spy
Levy vector fast and efficiently to retrieve in the facial image of magnanimity from data base with facial image similarity to be retrieved most
High facial image.
Fig. 2 is the indicative flowchart of the method for the retrieval facial image of another embodiment.As shown in Fig. 2 a kind of retrieval
The method of facial image, including：
S201, according to the characteristic point position information that default face characteristic point model determines facial image to be retrieved；
Fig. 3 is the human face characteristic point schematic diagram of this enforcement, as shown in figure 3, the method that human face characteristic point is extracted according to DLIB,
It is extracted 60 human face characteristic points.Wherein, the sequence number of face contour characteristic point is 0-16, totally 16, left eyebrow outline characteristic point
Sequence number be 17-21, totally 5, the sequence number of right eyebrow outline characteristic point be 22-26, totally 5, the sequence number of nose contour feature point
27-35, totally 9, the sequence number of left eye eyeball contour feature point be 36-41, totally 6, the sequence number of right eye eyeball contour feature point be 42-
47, totally 6,13 sequence numbers of face contour feature point be 48-60, totally 13.
The face of facial image to be retrieved is corrected by S202 according to the characteristic point position information.
In the present embodiment, according to the characteristic point position information by the face normalization of facial image to be retrieved to posture, institute
State face frontal pose be two eyes on a horizontal line, left and right be bold little basically identical, without pitching.By by face
Frontal pose is corrected to, the characteristic point position coordinate unification on facial image to vertical coordinate system can be obtained more accurately
Characteristic point position information.
S203, according to the facial characteristics that the characteristic point position information is calculated facial image to be retrieved.
Characteristic point position information according to face contour is calculated shape of face parameter.Shape of face is divided into round face, oval face, Ovum Anas domestica
Face, pears type face, state's word face and other.In the present embodiment, the shape of face ginseng of facial image to be retrieved can be obtained in the following manner
Number：
Obtain range data D of characteristic point 1 and characteristic point 151, as face width data；Obtain characteristic point 21 and characteristic point 22
Range data D of the midpoint of the line of determination to characteristic point 82, by 3*D2/ 2 used as the long data of face；By long for face data divided by face width
Data obtain the long face width ratio of face；According to the long face width ratio of face set in advance and the corresponding relation of shape of face, corresponding shape of face is obtained.
Characteristic point position information according to eyebrow outline is calculated eyebrow shape parameter, eyebrow shape parameter mainly with eyebrow peak,
Eyebrow tail sag and eyebrow within angle are represented；Eyebrow type can be straight eyebrow and curved eyebrow, according to brows, eyebrow tail, eyebrow peak characteristic point position letter
Breath determines eyebrow type.The ordinate of orthogonal axes data of eyebrow outline characteristic point are obtained, the maximum point of ordinate of orthogonal axes data is defined as eyebrow most
High point.Obtain the difference data D of characteristic point 18 and the ordinate of orthogonal axes of characteristic point 173, obtain characteristic point 17 and characteristic point 21 away from
From data D4, by D3/D4As eyebrow tail sag.Obtain the line l of characteristic point 20 and characteristic point 211, obtain characteristic point 22 and spy
Levy a little 23 line l2, obtain line l1And l2Angle data, as eyebrow within angle.
Characteristic point position information according to face contour is calculated chin type parameter；Chin type can be divided under point
Bar and circle chin；Determine chin type parameter in the following manner：Obtain the line l of characteristic point 5 and characteristic point 73, obtain feature
Point 9 and the line l of characteristic point 114, obtain line l3With line l4Angle data D5；Characteristic point 5 and characteristic point 11 are obtained simultaneously
Range data D6, as chin width parameter.Integrated clip angular data D5With chin width parameter, chin type parameter is obtained.Enter
One step, if chin width parameter is little and angle data D5Little as pointed chin parameter, is otherwise to justify chin parameter.
Characteristic point position information according to nose profile is calculated nose length-width ratio.Obtain characteristic point 27 and characteristic point 31 away from
From data D7With characteristic point 31 and 35 range data D of characteristic point8, by D7/D8As nose length-width ratio.
Positional information calculation according to the characteristic point position information and eyebrow outline characteristic point of eye contour obtains interior external eyes
Line is split with horizontal line angle, looks away from wide with eye.Obtain characteristic point 20 and 38 range data D of characteristic point9With characteristic point 23 with
Range data D of characteristic point 4310, by (D9+D10)/2 as looks away from.Characteristic point 36 is obtained to the line l of characteristic point 396, will
Line l6With horizontal angle data as inside and outside palpebral fissure line and horizontal line angle.Characteristic point 36 is obtained with characteristic point 39
Range data D11With characteristic point 42 and range data D of characteristic point 4512, by (D11+D12)/2 are wide as eye.
S204, the facial characteristics are classified by predetermined classification standard, obtain dividing for the facial characteristics
Level series.
As a preferred embodiment, before facial characteristics will be set by the classification of predetermined classification standard, face is first determined
The classification standard of portion's feature.For example, the method for determining the classification standard of eyebrow is that the eyebrow thickness to a number of face is entered
Row statistics, can be 10,000 people, calculate its average, variance, quantify to 5 grades.By all of setting facial characteristics by predetermined
Classification standard be classified, obtain the hierarchical level of all facial characteristics.
According to default semantic feature, S205, determines that rule, hierarchical level obtain the facial characteristics corresponding semantic special
Levy；According to the corresponding semantic feature of each facial characteristics, the semantic feature of facial image to be retrieved is obtained.
As a preferred embodiment, determine that rule, hierarchical level obtain the facial characteristics according to default semantic feature
Corresponding semantic feature；For example, eyebrow thickness grade series be 4, it is determined that the corresponding semantic feature of eyebrow be " heavy eyebrows ", eyes
Order of magnitude series is 4, it is determined that the corresponding semantic feature of eyes is " big eye "；Both combine the semantic feature for obtaining
" heavy features " can be used as the semantic feature of facial image to be retrieved.
S206, combines the corresponding hierarchical level of each facial characteristics, obtains the characteristic vector of facial image to be retrieved.
Such as, the corresponding hierarchical level of each facial characteristics is combined, and the characteristic vector for obtaining facial image to be retrieved isWherein first element 4 represents shape of face series, and second element 2 represents eyebrow type series, the 3rd unit
Element 5 represents eyebrow tail sag series, and the 4th element 1 represents nose length-width ratio, and the 5th element 3 represents looks away from series, with this
Analogize.
S207, the semantic feature according to the facial image to be retrieved enter line retrieval in whole data base, from data base
In determine range of search in the facial image that pre-saves.
As a preferred embodiment, the shape of face, eyebrow type, nose type, eyes and chin according to the facial image to be retrieved
Semantic feature enters line retrieval in whole data base, determines range of search in the facial image for pre-saving from data base, can
Greatly to reduce search space.
S208, by the feature of each facial image in the characteristic vector of the facial image to be retrieved and the range of search
Vector is compared, and obtains vector differentials；The minimum facial image of the vector differentials in range of search is found out, is defined as and is treated
Retrieval facial image similarity highest facial image, as most like with facial image to be retrieved facial image.
In the present embodiment, by each facial image in the characteristic vector of the facial image to be retrieved and the range of search
Characteristic vector be compared, obtain vector differentials, according to the vector differentials for obtaining, by data base in facial image to be retrieved
Range of search in facial image sequence, the facial image little with the vector differentials of facial image to be retrieved chosen before coming
Select the most front facial image of sequence to be and facial image similarity highest facial image to be retrieved.
The beneficial effect of the present embodiment is the feature for determining facial image to be retrieved according to default face characteristic point model
Dot position information, according to characteristic vector and semantic feature that the characteristic point position information for determining obtains facial image to be retrieved, first
Semantic feature according to the facial image to be retrieved enters line retrieval in whole data base, the people pre-saved from data base
Range of search is determined in face image；Then using the facial image to be retrieved characteristic vector can more rapidly, more efficient
Retrieve in the range of search and facial image similarity highest facial image to be retrieved.
Fig. 4 is the schematic diagram of the system of the retrieval facial image of an embodiment.As shown in figure 4, a kind of retrieval people
The system of face image, including：Characteristic point position information determination module 101, facial characteristics computing module 102, characteristic vector are obtained
Module 103 and retrieval module 104；The characteristic point position information determination module 101, for according to default human face characteristic point mould
Type determines the characteristic point position information of facial image to be retrieved；The facial characteristics computing module 102, for according to the feature
Dot position information is calculated the facial characteristics of facial image to be retrieved；The characteristic vector acquisition module 103, for combining
Facial characteristics are stated, the characteristic vector of facial image to be retrieved is obtained；The retrieval module 104, for according to the characteristic vector
The facial image pre-saved in searching database, obtains and facial image similarity highest facial image to be retrieved.
Used as a preferred embodiment, the system of the retrieval facial image also includes：Semantic feature acquisition module is (in figure not
Illustrate)；The semantic feature acquisition module, for obtaining the semantic feature of facial image to be retrieved according to the facial characteristics；
The retrieval module, is additionally operable to enter line retrieval according to the semantic feature of the facial image to be retrieved in whole data base, from
A range of search is determined in the facial image pre-saved in data base；Entered in the range of search according to the characteristic vector
Line retrieval, obtains and facial image similarity highest facial image to be retrieved.
Used as a preferred embodiment, the system of the retrieval facial image also includes：Diversity module (not shown)；Institute
Diversity module is stated, for being classified the facial characteristics by predetermined classification standard, the facial characteristics is obtained
Hierarchical level；Determine that rule, hierarchical level obtain the corresponding semantic feature of the facial characteristics according to default semantic feature；Group
The corresponding semantic feature of each facial characteristics is closed, the semantic feature of facial image to be retrieved is obtained.The characteristic vector acquisition module,
It is additionally operable to combine the corresponding hierarchical level of each facial characteristics, obtains the characteristic vector of facial image to be retrieved.
As a preferred embodiment, the retrieval module, be additionally operable to by the characteristic vector of the facial image to be retrieved with
The characteristic vector of each facial image in the range of search is compared, and obtains vector differentials；Find out described in range of search
The minimum facial image of vector differentials, is defined as and facial image similarity highest facial image to be retrieved.
Used as a preferred embodiment, the system of the retrieval facial image also includes：Human face posture correction module is (in figure not
Illustrate)；The human face posture correction module, for entering the face of facial image to be retrieved according to the characteristic point position information
Row correction, ordinary circumstance, by face normalization to frontal pose.The frontal pose of the face be two eyes in a horizontal line
Upper, left and right face is in the same size, without pitching.
Used as a preferred embodiment, described default face characteristic point model includes：The characteristic point of face contour, eyebrow
At least two classes in the characteristic point of the characteristic point of profile, the characteristic point of nose profile, the characteristic point of eye contour and face profile
Characteristic point.
The facial characteristics computing module 102, is additionally operable to obtain face according to the positional information calculation of face contour characteristic point
Type and chin type；Positional information calculation according to eyebrow outline characteristic point obtains eyebrow type；Characteristic point position according to nose profile
Confidence breath is calculated nose length-width ratio；Characteristic point position information and the characteristic point position letter of eye contour according to eyebrow outline
Breath, is calculated inside and outside palpebral fissure line with horizontal line angle, looks away from wide with eye.
Each technical characteristic of embodiment described above arbitrarily can be combined, for making description succinct, not to above-mentioned reality
Apply all possible combination of each technical characteristic in example to be all described, as long as however, the combination of these technical characteristics is not deposited
In contradiction, the scope of this specification record is all considered to be.
Embodiment described above only expresses the several embodiments of the present invention, and its description is more concrete and detailed, but simultaneously
Therefore can not be construed as limiting the scope of the patent.It should be pointed out that for one of ordinary skill in the art
Say, without departing from the inventive concept of the premise, some deformations and improvement can also be made, these belong to the protection of the present invention
Scope.Therefore, the protection domain of patent of the present invention should be defined by claims.
Claims (10)
1. a kind of retrieval facial image method, it is characterised in that include：
According to the characteristic point position information that default face characteristic point model determines facial image to be retrieved；
According to the facial characteristics that the characteristic point position information is calculated facial image to be retrieved；
The facial characteristics are combined, the characteristic vector of facial image to be retrieved is obtained；
According to the facial image pre-saved in the characteristic vector searching database, obtain and facial image similarity to be retrieved
Highest facial image.
2. according to claim 1 retrieval facial image method, it is characterised in that according to the characteristic point position information
Be calculated facial image to be retrieved facial characteristics the step of after also include：
According to the semantic feature that the facial characteristics obtain facial image to be retrieved；
According to the facial image pre-saved in the characteristic vector searching database, obtain and facial image similarity to be retrieved
The step of highest facial image, includes：
Semantic feature according to the facial image to be retrieved enters line retrieval in the data base, pre-saves from data base
Facial image in determine range of search corresponding with the semantic feature；
Enter line retrieval using the characteristic vector in the range of search, obtain and facial image similarity highest to be retrieved
Facial image.
3. according to claim 2 retrieval facial image method, it is characterised in that treated according to the facial characteristics
The step of semantic feature of retrieval facial image, includes：
The facial characteristics are classified by predetermined classification standard, the hierarchical level of the facial characteristics is obtained；
Determine that rule, hierarchical level obtain the corresponding semantic feature of the facial characteristics according to default semantic feature；
The corresponding semantic feature of each facial characteristics is combined, the semantic feature of facial image to be retrieved is obtained.
4. the method for retrieval facial image according to claim 3, it is characterised in that the combination facial characteristics, obtains
The step of characteristic vector of facial image to be retrieved, includes：
The corresponding hierarchical level of each facial characteristics is combined, the characteristic vector of facial image to be retrieved is obtained.
5. according to claim 2 retrieval facial image method, it is characterised in that with the characteristic vector in the inspection
Enter line retrieval in the range of rope, include the step of obtain with facial image similarity highest facial image to be retrieved：
The characteristic vector of each facial image in the characteristic vector of the facial image to be retrieved and the range of search is carried out
Relatively, vector differentials are obtained；
The minimum facial image of the vector differentials in range of search is found out, is defined as and the facial image similarity to be retrieved
Highest facial image.
6. according to claim 1 retrieval facial image method, it is characterised in that according to default human face characteristic point
The step of model determines the characteristic point position information of facial image to be retrieved and it is calculated according to the characteristic point position information
Include between the step of facial characteristics of facial image to be retrieved：
The face of facial image to be retrieved is corrected according to the characteristic point position information.
7. according to claim 1 to 6 any one retrieval facial image method, it is characterised in that described default
Face characteristic point model includes the characteristic point of face contour, the characteristic point of eyebrow outline, the characteristic point of nose profile, eye contour
Characteristic point and face profile characteristic point at least 2 category feature points.
8. according to claim 7 retrieval facial image method, it is characterised in that the default human face characteristic point mould
Type include the characteristic point of face contour, the characteristic point of eyebrow outline, the characteristic point of nose profile, the characteristic point of eye contour and
The characteristic point of face profile, be calculated according to the characteristic point position information facial image to be retrieved facial characteristics the step of
Including：
Characteristic point position information according to face contour is calculated shape of face parameter and chin type parameter；
Characteristic point position information according to eyebrow outline is calculated eyebrow shape parameter；
Characteristic point position information according to nose profile is calculated nose length-width ratio；
According to the characteristic point position information and the characteristic point position information of eye contour of eyebrow outline, it is calculated inside and outside palpebral fissure and connects
Line is with horizontal line angle, looks away from wide with eye.
9. a kind of retrieval facial image system, it is characterised in that include：Characteristic point position information determination module, facial characteristics
Computing module, characteristic vector acquisition module and retrieval module；
The characteristic point position information determination module, for determining facial image to be retrieved according to default face characteristic point model
Characteristic point position information；
The facial characteristics computing module, for being calculated the face of facial image to be retrieved according to the characteristic point position information
Portion's feature；
The characteristic vector acquisition module, for combining the facial characteristics, obtains the characteristic vector of facial image to be retrieved；
The retrieval module, for according to the facial image pre-saved in the characteristic vector searching database, obtaining and treating
Retrieval facial image similarity highest facial image.
10. according to claim 9 retrieval facial image system, it is characterised in that also include：Semantic feature obtains mould
Block；
The semantic feature acquisition module, for obtaining the semantic feature of facial image to be retrieved according to the facial characteristics；
The retrieval module, is additionally operable to be examined in whole data base according to the semantic feature of the facial image to be retrieved
Rope, determines range of search from data base in the facial image for pre-saving；According to the characteristic vector in the range of search
Inside enter line retrieval, obtain and facial image similarity highest facial image to be retrieved.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610973398.6A CN106503686A (en) | 2016-10-28 | 2016-10-28 | The method and system of retrieval facial image |
| PCT/CN2016/110134 WO2018076495A1 (en) | 2016-10-28 | 2016-12-15 | Method and system for retrieving face image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610973398.6A CN106503686A (en) | 2016-10-28 | 2016-10-28 | The method and system of retrieval facial image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106503686A true CN106503686A (en) | 2017-03-15 |
Family
ID=58323725
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610973398.6A Pending CN106503686A (en) | 2016-10-28 | 2016-10-28 | The method and system of retrieval facial image |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN106503686A (en) |
| WO (1) | WO2018076495A1 (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108038475A (en) * | 2017-12-29 | 2018-05-15 | 浪潮金融信息技术有限公司 | Facial image recognition method and device, computer-readable storage medium, terminal |
| CN108108711A (en) * | 2017-12-29 | 2018-06-01 | 深圳云天励飞技术有限公司 | Face supervision method, electronic equipment and storage medium |
| CN108229378A (en) * | 2017-12-29 | 2018-06-29 | 浪潮金融信息技术有限公司 | Face image data generation method and device, computer storage media, terminal |
| CN108564529A (en) * | 2018-04-23 | 2018-09-21 | 广东奥园奥买家电子商务有限公司 | A kind of implementation method of the real-time makeup of lip based on android system |
| CN108932321A (en) * | 2018-06-29 | 2018-12-04 | 金蝶软件（中国）有限公司 | Research on face image retrieval, device, computer equipment and storage medium |
| CN109376596A (en) * | 2018-09-14 | 2019-02-22 | 广州杰赛科技股份有限公司 | Face matching process, device, equipment and storage medium |
| CN111444374A (en) * | 2020-04-09 | 2020-07-24 | 上海依图网络科技有限公司 | Human body retrieval system and method |
| CN112307239A (en) * | 2020-10-29 | 2021-02-02 | 泰康保险集团股份有限公司 | Image retrieval method, device, medium and equipment |
| CN113243804A (en) * | 2021-06-03 | 2021-08-13 | 山东中新优境智能科技有限公司 | Automatic paper fetching method and device, readable storage medium and computer equipment |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108765551B (en) * | 2018-05-15 | 2022-02-01 | 福建省天奕网络科技有限公司 | Method and terminal for realizing face pinching of 3D model |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8300950B2 (en) * | 2008-02-29 | 2012-10-30 | Canon Kabushiki Kaisha | Image processing apparatus, image processing method, program, and storage medium |
| CN103207898A (en) * | 2013-03-19 | 2013-07-17 | 天格科技（杭州）有限公司 | Method for rapidly retrieving similar faces based on locality sensitive hashing |
| US20140270370A1 (en) * | 2013-03-18 | 2014-09-18 | Kabushiki Kaisha Toshiba | Person recognition apparatus and person recognition method |
| CN104090972A (en) * | 2014-07-18 | 2014-10-08 | 北京师范大学 | Image feature extraction and similarity measurement method used for three-dimensional city model retrieval |
| CN104715224A (en) * | 2013-12-11 | 2015-06-17 | 腾讯科技（深圳）有限公司 | Method and device for acquiring facial feature images of user group |
| CN105868716A (en) * | 2016-03-29 | 2016-08-17 | 中国科学院上海高等研究院 | Method for human face recognition based on face geometrical features |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1319013C (en) * | 2005-03-16 | 2007-05-30 | 沈阳工业大学 | Combined recognising method for man face and ear characteristics |
| CN102194131B (en) * | 2011-06-01 | 2013-04-10 | 华南理工大学 | Fast human face recognition method based on geometric proportion characteristic of five sense organs |
| CN103020607B (en) * | 2012-12-27 | 2017-05-03 | Tcl集团股份有限公司 | Face recognition method and face recognition device |
| CN104680121B (en) * | 2013-11-27 | 2022-06-03 | 腾讯科技（深圳）有限公司 | Method and device for processing face image |
| CN104951767A (en) * | 2015-06-23 | 2015-09-30 | 安阳师范学院 | Three-dimensional face recognition technology based on correlation degree |
-
2016
- 2016-10-28 CN CN201610973398.6A patent/CN106503686A/en active Pending
- 2016-12-15 WO PCT/CN2016/110134 patent/WO2018076495A1/en active Application Filing
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8300950B2 (en) * | 2008-02-29 | 2012-10-30 | Canon Kabushiki Kaisha | Image processing apparatus, image processing method, program, and storage medium |
| US20140270370A1 (en) * | 2013-03-18 | 2014-09-18 | Kabushiki Kaisha Toshiba | Person recognition apparatus and person recognition method |
| CN103207898A (en) * | 2013-03-19 | 2013-07-17 | 天格科技（杭州）有限公司 | Method for rapidly retrieving similar faces based on locality sensitive hashing |
| CN104715224A (en) * | 2013-12-11 | 2015-06-17 | 腾讯科技（深圳）有限公司 | Method and device for acquiring facial feature images of user group |
| CN104090972A (en) * | 2014-07-18 | 2014-10-08 | 北京师范大学 | Image feature extraction and similarity measurement method used for three-dimensional city model retrieval |
| CN105868716A (en) * | 2016-03-29 | 2016-08-17 | 中国科学院上海高等研究院 | Method for human face recognition based on face geometrical features |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108108711B (en) * | 2017-12-29 | 2019-12-17 | 深圳云天励飞技术有限公司 | Face control method, electronic device and storage medium |
| CN108108711A (en) * | 2017-12-29 | 2018-06-01 | 深圳云天励飞技术有限公司 | Face supervision method, electronic equipment and storage medium |
| CN108229378A (en) * | 2017-12-29 | 2018-06-29 | 浪潮金融信息技术有限公司 | Face image data generation method and device, computer storage media, terminal |
| CN108038475A (en) * | 2017-12-29 | 2018-05-15 | 浪潮金融信息技术有限公司 | Facial image recognition method and device, computer-readable storage medium, terminal |
| CN108564529A (en) * | 2018-04-23 | 2018-09-21 | 广东奥园奥买家电子商务有限公司 | A kind of implementation method of the real-time makeup of lip based on android system |
| CN108932321A (en) * | 2018-06-29 | 2018-12-04 | 金蝶软件（中国）有限公司 | Research on face image retrieval, device, computer equipment and storage medium |
| CN108932321B (en) * | 2018-06-29 | 2020-10-23 | 金蝶软件（中国）有限公司 | Face image retrieval method and device, computer equipment and storage medium |
| CN109376596A (en) * | 2018-09-14 | 2019-02-22 | 广州杰赛科技股份有限公司 | Face matching process, device, equipment and storage medium |
| CN111444374A (en) * | 2020-04-09 | 2020-07-24 | 上海依图网络科技有限公司 | Human body retrieval system and method |
| CN111444374B (en) * | 2020-04-09 | 2023-05-02 | 上海依图网络科技有限公司 | Human body retrieval system and method |
| CN112307239A (en) * | 2020-10-29 | 2021-02-02 | 泰康保险集团股份有限公司 | Image retrieval method, device, medium and equipment |
| CN112307239B (en) * | 2020-10-29 | 2024-02-02 | 泰康保险集团股份有限公司 | Image retrieval method, device, medium and equipment |
| CN113243804A (en) * | 2021-06-03 | 2021-08-13 | 山东中新优境智能科技有限公司 | Automatic paper fetching method and device, readable storage medium and computer equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2018076495A1 (en) | 2018-05-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106503686A (en) | The method and system of retrieval facial image | |
| CN110263774B (en) | A kind of method for detecting human face | |
| CN104866829B (en) | A kind of across age face verification method based on feature learning | |
| CN106682598B (en) | Multi-pose face feature point detection method based on cascade regression | |
| CN103577815B (en) | A kind of face alignment method and system | |
| CN105894047B (en) | A kind of face classification system based on three-dimensional data | |
| CN103810490B (en) | A kind of method and apparatus for the attribute for determining facial image | |
| CN103218609B (en) | A kind of Pose-varied face recognition method based on hidden least square regression and device thereof | |
| CN103632147A (en) | System and method for implementing standardized semantic description of facial features | |
| CN109508700A (en) | A kind of face identification method, system and storage medium | |
| CN105335726B (en) | Recognition of face confidence level acquisition methods and system | |
| CN112016464A (en) | Method and device for detecting face shielding, electronic equipment and storage medium | |
| CN101493887B (en) | Eyebrow image segmentation method based on semi-supervision learning and Hash index | |
| CN111126240B (en) | Three-channel feature fusion face recognition method | |
| CN105138968A (en) | Face authentication method and device | |
| CN105809113B (en) | Three-dimensional face identification method and the data processing equipment for applying it | |
| CN106529397B (en) | A kind of man face characteristic point positioning method in unconstrained condition and system | |
| CN106960181A (en) | A kind of pedestrian's attribute recognition approach based on RGBD data | |
| CN108108760A (en) | A kind of fast human face recognition | |
| CN106778489A (en) | The method for building up and equipment of face 3D characteristic identity information banks | |
| CN110796101A (en) | Face recognition method and system of embedded platform | |
| CN101833654A (en) | Sparse representation face identification method based on constrained sampling | |
| CN106599785A (en) | Method and device for building human body 3D feature identity information database | |
| CN110516533A (en) | A kind of pedestrian based on depth measure discrimination method again | |
| CN110232331A (en) | A kind of method and system of online face cluster |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170315 |