CN105138954A - Image automatic screening, query and identification system - Google Patents
Image automatic screening, query and identification system Download PDFInfo
- Publication number
- CN105138954A CN105138954A CN201510406384.1A CN201510406384A CN105138954A CN 105138954 A CN105138954 A CN 105138954A CN 201510406384 A CN201510406384 A CN 201510406384A CN 105138954 A CN105138954 A CN 105138954A
- Authority
- CN
- China
- Prior art keywords
- face
- module
- eyes
- image
- pixel
- 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.)
- Granted
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
- G06V40/168—Feature extraction; Face representation
- G06V40/171—Local features and components; Facial parts ; Occluding parts, e.g. glasses; Geometrical relationships
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
- G06F16/5838—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content using colour
Abstract
The invention discloses an image automatic screening, query and identification system. The image automatic screening, query and identification system comprises an image input device, a database system, a storage system, a face detection system, a face registration system and a face comparison system. The image input device carries out the processes of real-time video inputting, video playback inputting and photograph batch importing. The database system stores personal associated portrait information including personal certificate information, a certificate photo, a life photo, a group photo and corresponding characteristic values. The storage system stores 3D-modeled portrait information namely the characteristic values and source indication. The face detection system reads an image input by the image input device, carries out face detection and collects a portrait according with modeling conditions. The face registration system carries out face modeling, the modeled face image is from the face detection system, and face modeling information is registered in the storage system. A face identification system carries out characteristic constant comparison retrieval on the portrait information, which is stored in the storage system, in the database system, thereby realizing person querying by persons and finding out the real identity of the person.
Description
Technical field
The present invention relates to a kind of area of computer aided living creature characteristic recognition system, relate to a kind of image automatic screening inquiry recognition system in particular.
Background technology
Along with rapid development of economy, the dense degree of population is concentrated further, and particularly along with the sternness of anti-terrorism situation, we are faced with all kinds of public safety situation.Meanwhile, modern commerce, logistics and network develop rapidly also also extremely urgent to the requirement of the examination & verification of the other side's identity, confirmation.
Science and technology is development constantly, and cloud storage, cloud computing, the large database concept recognition of face of excavating under a series of technical support are undoubtedly the best mode that remote identity confirms.
Remote human face identification, takes dynamic human face identification best, the most reliable beyond doubt.But be limited to the reasons such as current hardware, horizontal network and high cost, dynamic online recognition of face is also difficult to large area and popularizes, and therefore adopts the selection that the recognition of face mode of dynamic and static combination is best beyond doubt.
Summary of the invention
This patent mainly combines recognition of face retrieval and large data store two large technology.
The maximum feature of this face recognition technology is dynamic acquisition, static identification, combining dynamic and static research.Video camera uploads real-time dynamic video by network, face identification system gathers containing can people's picture frame of modeling automatically, be converted to photo, then human face photo modeling is transferred to backstage storage system, again by large data query, the static concurrent comparison of 1:N, retrieves with the portrait comparison stored in Database Systems, look into people with people, inquire suspect's true identity information.
According to this technical field of recognition of face, face is formed by forehead, eyebrow, eyes, nose, face, cheeks six areas combine, the identity characteristic information contained in each face of the size of these six interregional relative positions, region organ, relative scale and structural feature.Extract the identity characteristic information contained in each face, thus identifiable design goes out everyone identity.
The main eigenwert of face obtains, and below the eyebrow being limited to people, between lower jaw, separate according to the mankind achievement that plane learns, learn that face is in specific trigonum, the size of some organ, position, relative scale, have uniqueness and unchangeability; And the ratio of other organs, though change, relative stability can be kept.Around this principle, as long as find in face characteristic region, the size of relevant organ, position, relative scale just can be crossed and extract its eigenwert.
The facial zone feature analysis al that the face recognition technology that this patent adopts adopts, it has merged computer image processing technology and biostatistics principle in one, utilize computer image processing technology from video image, extract portrait unique point, utilize the principle of biostatistics to carry out analysis and set up 3D mathematical model, i.e. skin detection.The eigenwert comparison utilizing the face characteristic value generated in photo library and the face collected to generate, is provided a similarity, can be determined whether as same people by this value.
Specifically, the present invention is a kind of image automatic screening inquiry recognition system, comprises image input device, Database Systems, storage system, face detection system, face registration system and face alignment system.
Described image input device comprises real-time video input (dynamically), playing back videos input (dynamically), photo batch importing (static state);
Described Database Systems store individual's association figure information, comprise personal document information (I.D., driver's license, household register, social security, passport etc.), certificate photo certificate photos such as () I.D., driver's license, social security card, passports, living photo, group picture and characteristic of correspondence value etc.;
Described storage system preserves figure information (i.e. eigenwert) and the source sign of 3D modeling;
The image of described face detection system reading images input equipment input, carries out Face datection, gathers the portrait meeting modeling conditions;
Described face registration system carries out face modeling, and the human face photo of modeling derives from face detection system, and to storage system registration face modeling information;
Described face identification system, by carrying out eigenwert comparison retrieval to the figure information in storage system in Database Systems, is looked into people with people, is inquired the true identity of people.
Described face identification system is combining dynamic and static research, and namely front end is dynamic figure acquisition system, and rear end is the concurrent Compare System in still photo 4000 tunnel;
Described face motion capture engine comprises screening module, spells hardwood composograph module, background process module, track following module;
Described interference reduction engine comprises light interference recovery module, ethnic group identification module, age recovery module, expression recovery module, attitude recovery module, blocks recovery module.
Described face Modeling engine is to collecting 2D portrait, by the fixed attribute such as size, ratio, relative position, distance of image surface face profile, launch by 3D image surface organ template, corresponding geometric relationship forms identification parameter and data, calculate mutual association geometric vector (eigenwert), namely generate 3D eigenwert.
Described data register engine carries out source associated designation (information such as time, camera number) to modeling data, and presses the storage of database standard login mode, so that data query.
Described face alignment system is the eigenwert comparison engine of " three-in-one ", comprises three kinds of eigenwert comparing module: 1.: 12 pixel-24 pixels (400 point) comparing module between eyes; 2.: 24 pixel-40 pixels (1500 point) comparing module between eyes; 3.: 40 pixel-60 pixels (4000 point) comparing module between eyes.System calculates the quantity of pixel between face eyes automatically, according to the quantity of pixel between face eyes, automatically chooses the corresponding one in above-mentioned three kinds of comparing module, three kinds of comparing module assemblings together, and synthesis " three-in-one " comparison engine.
Described screening module comprises the steps:
Image in step one, described video input apparatus and the degree of conformity inspection of face basic templates, i.e. face basic templates filtration method, the trigonum that two eyes and nose are formed is the most essential characteristic of face, the qualified step 2 that enters;
In step 2, described video input apparatus, face angle compares with standard portrait, differential seat angle scope in left and right ± 25 °/up and down ± 15 °/rotate ± 10 ° eligiblely enter step 3;
Step 3, eyes are visible, pixel inspection between eyes, between the pixel value total according to video input apparatus and eyes, region area accounts for the ratio of whole camera picture, calculates the pixel point value in region between eyes, need pixel point value between eyes to be greater than 12, what meet above condition does face collection;
Described track following module: system is to video flow point hardwood, can the face frame of comparison from collecting the first frame, in in 2 subsequently second, system can carry out verifying mutually between hardwood automatically, choose in 50 hardwoods (2 seconds * 25 hardwoods/seconds) the most clearly two width portrait hardwoods as comparison hardwood, spell hardwood, synthesis, as comparison source in comparing module; Identify the portrait collected, based on the algorithm combined with motion model, tracking of comparing in front end, if be confirmed to be same people, will not do the collection of second time face simultaneously.Like this, the hardware resource of backstage CPU, transmission bandwidth, storage is greatly saved.
Accompanying drawing explanation
Fig. 1 is for identifying part face structural representation.
Fig. 2, Fig. 3 are background process functions of modules schematic diagram.
Fig. 4 is light interference recovery module functional schematic.
Fig. 5 is different ethnic group template schematic diagram.
Fig. 6 is expression reduction schematic diagram.
Fig. 7, Fig. 8 are attitude reduction schematic diagram.
Fig. 9 blocks reduction schematic diagram.
Figure 10 is face modeling schematic diagram.
Figure 11 is that twins identify schematic diagram.
Figure 12 is system architecture schematic diagram.
Figure 13 is system flowchart
illustrate:
1-pouch
2-tear ditch, apple flesh are sagging
3-decree line
4-puppet line
5-outline line
Embodiment
Below in conjunction with Figure of description, embodiment is described:
One, unique point summary
As described in summary of the invention part, unique point is the core of this patent, and face is in specific trigonum, and the size of some organ, position, relative scale, have uniqueness and unchangeability; And the ratio of other organs, though change, can keep relative stability, according to the intensity of variation of the size of these organs, position, relative scale, this patent is divided into three major types:
A) uniqueness invariant relation:
Binocular interval;
The position proportional relation of pupil of both eyes and face bridge of the nose tip
Bridge of the nose radian
The width ratio relation of bridge of the nose arc length and nose
The geometry of cheekbone;
Place between the eyebrows is to upper lip spacing;
The spacing at eyes angle;
B) with the feature of age regular change, the trend that these parts of face change can be drawn, to reduce according to the deduction of these features:
Eyes angle is sagging;
Pouch;
Lips angle is sagging;
Decree line changes; (decree line be position wing of nose limit extend and under twice lines, be typical skin tissue aging, the phenomenon that causes skin surface to cave in; And often make up, laugh and do not note that maintenance all can make female friend produce decree line.)
C) extremely labile part
From the point in portion between ear-lobe and lower jaw, be commonly called as outline line.
In sum, not only have a unique point at various typical parts, but the function curve of a stack features point composition, the set of final formation facial feature points; Fig. 1 identifies above-mentioned part human face structure.
Two, each module describes in detail
Face identification system comprises face and catches engine, interference reduction engine, face Modeling engine, face alignment engine, photo eigen storehouse, template base.
1, face catches engine
Face catches engine and first screens the image obtained, screening module specifically comprises following three steps: step one, the degree of conformity inspection of the image in video camera and face basic templates (trigonum that two eyes and a nose are formed), i.e. face basic templates filtration method, qualifiedly enters step 2; Step 2, in video, face angle compares with portrait, differential seat angle scope in left and right ± 25 °/up and down ± 15 °/rotate ± 10 ° eligiblely enter step 3; Step 3, eyes are visible, pixel inspection between eyes, between the pixel value total according to video input apparatus and eyes, region area accounts for the ratio of whole camera picture, calculate the pixel point value in region between eyes, need pixel point value between eyes to be greater than 12, what meet above condition does face collection;
Above-mentioned screening is by cascade classifier screening method: detected image by each sorter, can pass through, can be judged to be qualifying object, enter next sorter successively.Meanwhile, in order to consider efficiency, the strictest sorter can be placed on the top of whole cascade classifier, can matching times be reduced like that.
Spell hardwood composograph module comprise framing and spell frame, judge that the video flowing of first two seconds of the video flowing being face resolves into hardwood in screening module, the frame that carries out per second compares, be exactly specifically the every hardwood image obtained is done available pixel point compare, pick out the maximum frame of available pixel point and be used for the frame of spelling frame, in two seconds, obtain two picture frames the most clearly.
Spell frame technique, exactly by above-mentioned two the most clearly picture frame carry out spellings frame, with prevent transmit in frame losing, two frame a preparation and a use.
In practical operation, framing, frame are compared, spelling frame can be repeatedly mutual with screening module.
Background process module refers to distinguishes the background of complexity and face the process come, and therefore first it want the border can judging face, could distinguish background.As Fig. 2, Fig. 3.
Track following module: system is to video flow point hardwood, can the face frame of comparison from collecting the first frame, in in 2 subsequently second, system can carry out verifying mutually between hardwood automatically, choose in 50 hardwoods (2 seconds * 25 hardwoods/seconds) the most clearly two width portrait hardwoods as comparison hardwood, spell hardwood, synthesis, as comparison source in comparing module; Identify the portrait collected, based on the algorithm combined with model that moves, tracking of comparing in front end, if be confirmed to be same people, will not do the collection of second time face simultaneously.Like this, the hardware resource of backstage CPU, transmission bandwidth, storage is greatly saved.
2. interference reduction engine
Second largest module is interference reduction engine, carries out correction reduction to the human face photo captured, and specifically, face reduction engine comprises:
2.1 light interference reduction: light interference mainly two kinds: negative and positive face and backlight.
Negative and positive face is revised by the principle of facial symmetry.
Backlight is by revising (light intensity) the brightness contrast of background and portrait.As Fig. 4, wherein X-axis represents grey black whiteness, and Y represents proportion.
2.2 ethnic group identifications
Ethnic group is divided into yellow/white people/black race/brown people
Identified, as Fig. 5 by face's elementary contour characteristic sum colour of skin of four large ethnic groups.
The reduction of 2.3 ages
Age reduction refers to, to category-B unique point in ", unique point summary ", produce one group of changing value in a certain scope, as additional feature value; Female thyroid cartilage, determines the size (A ± B%) of positive and negative modified value B.Photo age difference in practical photograph and storehouse is larger, the suitable amplification of this B value.
In the comparison process of module afterwards, these additional feature value are the same with the former characteristic parameter at this place, have same weight power, if for example the former eigenwert of this point is variant with the value being compared photo, but in this group additional feature value, but there is the value met, the accordance of comparison can be improved equally.
2.4 expression reduction
By dissect physiology principle, the point of simulating the variation of each epidermis is kept in the center and is put.Modified value revised law, specific algorithm.
Expression reduction, refers to, within the specific limits, by expression little for distortion, can revert to normal expression, as Fig. 6.
2.5 attitude reduction
This patent can to left and right ± 25 °/up and down ± 15 °/rotate ± 10 °, and in eyes visible range, photo carries out attitude reduction normotopia, and reaching eyes is horizontal coordinate, symmetrical, adjusts to normotopia.As Fig. 7, Fig. 8.
2.6 block reduction
Defect symmetry correction is carried out to shelters such as glasses/bang/scarf/polo-neck/thes brim of a hat, or mean value compensates.As Fig. 9. such as: left part is covered, and by the principle of right face and left face symmetry, revises.And for example: chin has been blocked by polo-neck, according to ethnic group chin mean value, as the eigenwert of this part.
3. face Modeling engine
The third-largest engine is the essential characteristic meeting the 2D portrait face of modeling conditions to collecting, the fixed attribute such as size, position, distance of face profile, launch by 3D image surface organ template, corresponding geometric relationship forms identification parameter and data, calculate mutual association geometric vector (eigenwert), namely generate 3D eigenwert.As Figure 10.
3D modeling can resist the change of light, skin color, facial hair, hair style, glasses, expression and attitude, has powerful reliability.
4. face alignment engine
What face recognition technology was conventional has the methods such as Gabor wavelet, Adaboost learning algorithm and support vector machine.
This patent have employed the comparison to human face characteristic point, and first by a kind of comparing module corresponding in selected " three-in-one " comparison engine of the quantity of pixel between eyes, the neural network algorithm learnt by the degree of depth is compared.Degree of depth study is by a kind of structural information algorithm of simulating human neural circuit " neuroid " on computers.It can form more abstract high-level characteristic by multi-level combination low-level feature, thus realizes automatic learning characteristic, and does not need people to participate in choosing of feature.Degree of depth study neural network algorithm improves accuracy and the analysis speed of analysis just by the analysis mode that simulation human brain is multi-level.
As can be seen from " one, unique point summary ", the ratio change of each several part, some changes are little, and some changes are large, and therefore algorithm is when determining the weight of unique point, is different in fact.
Point is fewer, and the proportion that category-A point accounts for is larger.For category-A unique point, because they are in the whole body of people, the specific proportionate relationship of basic maintenance, and there is uniqueness, so when trying to achieve face essential characteristic, occupy larger weight, this kind of point, on about 4000 points altogether, occupy the ratio of 56%, and weight allocation is more than 60%; For category-B unique point, although along with all one's life of people, can change, but this change is foreseeable, therefore, when comparison, can carry out redundancy deduction, this kind of point accounts for always count 32%, and weight allocation is about 30%, and last class point, such as facial contour line, along with the change of age and environment, frequent change can be there is, account for always count 12%, because change greatly, weight allocation is minimum only has 10%.According to above principle, we have drawn the essential characteristic point cohort of face, and obtain a part of eigenwert, but sometimes these deviations still can not react the uniqueness of face accurately.
Therefore this patent also takes compensatory algorithm and face surface area algorithm, its reason is, by research, if the surface area around these 4000 points is done one to calculate discovery, the registration probability of everyone surface area will be far smaller than the probability of eigenwert repetition, therefore above-mentioned unique point is connected with each other by we, (summit is upper to make each 3 adjacent some composition equilateral triangles, non-" equilateral triangle "), when getting these some positions, consider the positional factor of its geometry equilateral triangle, to guarantee to obtain these equilateral triangles, certainly these triangles also have weight allocation in fact, principle is consistent with unique point, adopt Gauss theorem, by calculating the area that A B C tri-groups of equilateral triangles obtain, form a redundancy value, be appended to eigenwert to obtain in parameter end.Through face surface area algorithm, the accuracy of recognition of face promotes further, in actual motion, can differentiate twins, as Figure 11.
The 3D eigenwert that in the 3D eigenwert that the people's picture frame modeling intercepted in video flowing generates by face alignment engine and photo library, photo generates is compared, and draws comparing result.
Three, recognition of face is used for the embodiment of large data retrieval
As shown in Figure 12 and Figure 13, Database Systems store individual's association figure information, millions and even hundreds of millions grades, and large data store; Storage system preserves figure information and the source sign of the built good mould that each aspect transmits; Face detection system reads front-end image, carries out Face datection, gathers the portrait meeting modeling conditions; Described face registration system carries out face modeling, and by network, the figure information of built good mould is uploaded the storage system being registered to Surveillance center; Described face identification system, by the figure information in storage system and figure information in Database Systems, by the 4000 concurrent comparison retrievals in tunnel, is looked into people with people, is inquired the true identity of people.
beneficial effect
Look into people with people, by the portrait comparison in the portrait that collects and database, retrieve the identity information in database, accurately inquire target person.The model database of hundreds of millions grades, the high-accuracy of identification and inquiry velocity are three large beneficial effects of this patent soon.This patent on the basis of existing technology, combines multinomial new model, algorithm, comprises the recognition of face of dynamic and static combination, 4000 tunnel concurrent comparison retrievals, large data store query, can complete the large data query looking into people with people fast and accurately.
Ministry of Public Security's measured data:
Unit static test:
Static database: 1,000 ten thousand standards are shone;
The static comparison (1:N) of one-to-many, discrimination >98%, recognition speed <2 second/people.
Claims (7)
1. an image automatic screening inquiry recognition system, comprises image input device, Database Systems, storage system, face detection system, face registration system and face alignment system,
Described image input device comprises real-time video input, playing back videos input, the importing of photo batch;
Described Database Systems store individual's association figure information, comprise personal document information, certificate photo, living photo, group picture and characteristic of correspondence value;
Described storage system preserves figure information and the eigenwert of 3D modeling and sign of originating;
The image of described face detection system reading images input equipment input, carries out Face datection, gathers the portrait meeting modeling conditions;
Described face registration system carries out face modeling, and the human face photo of modeling derives from face detection system, and to storage system registration face modeling information;
Described face identification system, by carrying out eigenwert comparison retrieval to the figure information in storage system in Database Systems, is looked into people with people, is inquired the true identity of people.
2. an image automatic screening inquiry recognition system, is characterized in that: face identification system is the concurrent Compare System in still photo 4000 tunnel.
3. image automatic screening inquiry recognition system according to claim 1, is characterized in that: described face detection system comprises face motion capture engine and interference reduction engine;
Face motion capture engine comprises screening module, spells hardwood composograph module, background process module, track following module;
Interference reduction engine comprises light interference recovery module, ethnic group identification module, age recovery module, expression recovery module, attitude recovery module, blocks recovery module.
4. image automatic screening inquiry recognition system according to claim 1, is characterized in that: described face registration system comprises face Modeling engine and data register engine;
Described face Modeling engine is to the 2D portrait collected, by the fixed attribute of image surface face profile, comprise: size, ratio, relative position, distance, launch by 3D image surface organ template, corresponding geometric relationship forms identification parameter and data, calculate mutual association geometric vector, namely generate 3D eigenwert;
Data register engine carries out source associated designation to modeling data, and presses the storage of database standard login mode, so that data query.
5. image automatic screening inquiry recognition system according to claim 1, it is characterized in that: described face alignment system is the eigenwert comparison engine of " three-in-one ", comprise three kinds of eigenwert comparing module: 1.: i.e. 400 comparing module of 12 pixel-24 pixel comparing module between eyes; 2.: i.e. 1500 comparing module of 24 pixel-40 pixel comparing module between eyes; 3.: i.e. 4000 comparing module of 40 pixel-60 pixel comparing module between eyes; System calculates the quantity of pixel between face eyes automatically, according to the quantity of pixel between face eyes, automatically chooses the corresponding one in above-mentioned three kinds of comparing module, three kinds of comparing module assemblings together, and synthesis " three-in-one " comparison engine.
6. image automatic screening inquiry recognition system according to claim 3, is characterized in that: described screening module comprises the steps:
Image in step one, described video input apparatus and the degree of conformity inspection of face basic templates, i.e. face basic templates filtration method, the trigonum that two eyes and nose are formed is the most essential characteristic of face, the qualified step 2 that enters;
In step 2, described video input apparatus, face angle compares with standard portrait, differential seat angle scope in left and right ± 25 °/up and down ± 15 °/rotate ± 10 ° eligiblely enter step 3;
Step 3, eyes are visible, pixel inspection between eyes, between the pixel value total according to video input apparatus and eyes, region area accounts for the ratio of whole camera picture, calculates the pixel point value in region between eyes, need pixel point value between eyes to be greater than 12, what meet above condition does face collection.
7. image automatic screening inquiry recognition system according to claim 3, it is characterized in that: described track following module is to video flow point hardwood, from collecting the first frame and meeting the face frame of face collection standard, in in 2 subsequently second, system can carry out verifying mutually between hardwood automatically, chooses two width portrait hardwoods the most clearly and, as comparison hardwood, spell hardwood in 50 hardwoods, synthesis, as comparison source in comparing module; Identify the portrait collected, based on the algorithm combined with motion model, tracking of comparing in front end, if be confirmed to be same people, will not do the collection of second time face simultaneously.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510406384.1A CN105138954B (en) | 2015-07-12 | 2015-07-12 | A kind of image automatic screening inquiry identifying system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510406384.1A CN105138954B (en) | 2015-07-12 | 2015-07-12 | A kind of image automatic screening inquiry identifying system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105138954A true CN105138954A (en) | 2015-12-09 |
| CN105138954B CN105138954B (en) | 2019-06-04 |
Family
ID=54724298
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510406384.1A Active CN105138954B (en) | 2015-07-12 | 2015-07-12 | A kind of image automatic screening inquiry identifying system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105138954B (en) |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105912997A (en) * | 2016-04-05 | 2016-08-31 | 福建兴宇信息科技有限公司 | Face recognition method and system |
| CN106408658A (en) * | 2016-10-31 | 2017-02-15 | 北京旷视科技有限公司 | Face modeling method and apparatus |
| CN106778645A (en) * | 2016-12-24 | 2017-05-31 | 深圳云天励飞技术有限公司 | A kind of image processing method and device |
| CN106778925A (en) * | 2016-11-03 | 2017-05-31 | 五邑大学 | A kind of super complete face automatic registration method of the attitude of recognition of face and its device |
| CN108171207A (en) * | 2018-01-17 | 2018-06-15 | 百度在线网络技术（北京）有限公司 | Face identification method and device based on video sequence |
| CN108206929A (en) * | 2016-12-16 | 2018-06-26 | 北京华泰科捷信息技术股份有限公司 | A kind of contactless personnel information acquisition device and its acquisition method |
| CN108256479A (en) * | 2018-01-17 | 2018-07-06 | 百度在线网络技术（北京）有限公司 | Face tracking method and device |
| CN108345847A (en) * | 2018-01-30 | 2018-07-31 | 石数字技术成都有限公司 | A kind of facial image label data generation system and method |
| CN108399247A (en) * | 2018-03-01 | 2018-08-14 | 深圳羚羊极速科技有限公司 | A kind of generation method of virtual identity mark |
| CN108764350A (en) * | 2018-05-30 | 2018-11-06 | 苏州科达科技股份有限公司 | Target identification method, device and electronic equipment |
| CN108877001A (en) * | 2018-06-19 | 2018-11-23 | 北京金山安全软件有限公司 | Visitor information processing method and device and electronic equipment |
| CN108897777A (en) * | 2018-06-01 | 2018-11-27 | 深圳市商汤科技有限公司 | Target object method for tracing and device, electronic equipment and storage medium |
| CN109087157A (en) * | 2018-06-08 | 2018-12-25 | 成都第二记忆科技有限公司 | A kind of video-photographic works sale service system and method and business model |
| CN109214335A (en) * | 2018-09-04 | 2019-01-15 | 高新兴科技集团股份有限公司 | A kind of alarm method and equipment |
| CN109255319A (en) * | 2018-09-02 | 2019-01-22 | 珠海横琴现联盛科技发展有限公司 | For the recognition of face payment information method for anti-counterfeit of still photo |
| CN109886360A (en) * | 2019-03-25 | 2019-06-14 | 山东浪潮云信息技术有限公司 | A kind of certificate photo Classification and Identification based on deep learning and detection method without a hat on and system |
| CN110059576A (en) * | 2019-03-26 | 2019-07-26 | 北京字节跳动网络技术有限公司 | Screening technique, device and the electronic equipment of picture |
| WO2019178711A1 (en) * | 2018-03-18 | 2019-09-26 | 广东欧珀移动通信有限公司 | Image processing method and apparatus, storage medium, and electronic device |
| CN111127639A (en) * | 2019-12-30 | 2020-05-08 | 深圳小佳科技有限公司 | Cloud-based face 3D model construction method, storage medium and system |
| CN111241328A (en) * | 2018-11-28 | 2020-06-05 | 航天信息股份有限公司 | Identity authentication and identification service method and device, readable storage medium and electronic equipment |
| CN111680280A (en) * | 2020-05-20 | 2020-09-18 | 青岛黄海学院 | Computer portrait recognition system |
| CN111738742A (en) * | 2020-05-07 | 2020-10-02 | 广东电网有限责任公司 | Portrait data processing system for power customer service |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102201061A (en) * | 2011-06-24 | 2011-09-28 | 常州锐驰电子科技有限公司 | Intelligent safety monitoring system and method based on multilevel filtering face recognition |
| US20140270374A1 (en) * | 2013-03-15 | 2014-09-18 | Nito, Inc. | Systems, Methods, and Software for Detecting an Object in an Image |
| CN104077804A (en) * | 2014-06-09 | 2014-10-01 | 广州嘉崎智能科技有限公司 | Method for constructing three-dimensional human face model based on multi-frame video image |
| CN104091176A (en) * | 2014-07-18 | 2014-10-08 | 吴建忠 | Technology for applying figure and head portrait comparison to videos |
-
2015
- 2015-07-12 CN CN201510406384.1A patent/CN105138954B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102201061A (en) * | 2011-06-24 | 2011-09-28 | 常州锐驰电子科技有限公司 | Intelligent safety monitoring system and method based on multilevel filtering face recognition |
| US20140270374A1 (en) * | 2013-03-15 | 2014-09-18 | Nito, Inc. | Systems, Methods, and Software for Detecting an Object in an Image |
| CN104077804A (en) * | 2014-06-09 | 2014-10-01 | 广州嘉崎智能科技有限公司 | Method for constructing three-dimensional human face model based on multi-frame video image |
| CN104091176A (en) * | 2014-07-18 | 2014-10-08 | 吴建忠 | Technology for applying figure and head portrait comparison to videos |
Non-Patent Citations (1)
| Title |
|---|
| 田强 等: "3D人脸识别技术及应用", 《警察技术》 * |
Cited By (29)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105912997A (en) * | 2016-04-05 | 2016-08-31 | 福建兴宇信息科技有限公司 | Face recognition method and system |
| CN105912997B (en) * | 2016-04-05 | 2019-05-28 | 福建兴宇信息科技有限公司 | Face recognition method and system |
| CN106408658B (en) * | 2016-10-31 | 2021-06-15 | 北京旷视科技有限公司 | Face modeling method and device |
| CN106408658A (en) * | 2016-10-31 | 2017-02-15 | 北京旷视科技有限公司 | Face modeling method and apparatus |
| CN106778925A (en) * | 2016-11-03 | 2017-05-31 | 五邑大学 | A kind of super complete face automatic registration method of the attitude of recognition of face and its device |
| CN106778925B (en) * | 2016-11-03 | 2021-10-08 | 五邑大学 | Face recognition pose over-complete face automatic registration method and device |
| CN108206929A (en) * | 2016-12-16 | 2018-06-26 | 北京华泰科捷信息技术股份有限公司 | A kind of contactless personnel information acquisition device and its acquisition method |
| WO2018113523A1 (en) * | 2016-12-24 | 2018-06-28 | 深圳云天励飞技术有限公司 | Image processing method and device, and storage medium |
| CN106778645A (en) * | 2016-12-24 | 2017-05-31 | 深圳云天励飞技术有限公司 | A kind of image processing method and device |
| CN108256479A (en) * | 2018-01-17 | 2018-07-06 | 百度在线网络技术（北京）有限公司 | Face tracking method and device |
| CN108171207A (en) * | 2018-01-17 | 2018-06-15 | 百度在线网络技术（北京）有限公司 | Face identification method and device based on video sequence |
| CN108256479B (en) * | 2018-01-17 | 2023-08-01 | 百度在线网络技术（北京）有限公司 | Face tracking method and device |
| CN108345847A (en) * | 2018-01-30 | 2018-07-31 | 石数字技术成都有限公司 | A kind of facial image label data generation system and method |
| CN108399247A (en) * | 2018-03-01 | 2018-08-14 | 深圳羚羊极速科技有限公司 | A kind of generation method of virtual identity mark |
| WO2019178711A1 (en) * | 2018-03-18 | 2019-09-26 | 广东欧珀移动通信有限公司 | Image processing method and apparatus, storage medium, and electronic device |
| CN108764350A (en) * | 2018-05-30 | 2018-11-06 | 苏州科达科技股份有限公司 | Target identification method, device and electronic equipment |
| CN108897777B (en) * | 2018-06-01 | 2022-06-17 | 深圳市商汤科技有限公司 | Target object tracking method and device, electronic equipment and storage medium |
| CN108897777A (en) * | 2018-06-01 | 2018-11-27 | 深圳市商汤科技有限公司 | Target object method for tracing and device, electronic equipment and storage medium |
| US11195284B2 (en) | 2018-06-01 | 2021-12-07 | Shenzhen Sensetime Technology Co., Ltd. | Target object tracking method and apparatus, and storage medium |
| CN109087157A (en) * | 2018-06-08 | 2018-12-25 | 成都第二记忆科技有限公司 | A kind of video-photographic works sale service system and method and business model |
| CN108877001A (en) * | 2018-06-19 | 2018-11-23 | 北京金山安全软件有限公司 | Visitor information processing method and device and electronic equipment |
| CN109255319A (en) * | 2018-09-02 | 2019-01-22 | 珠海横琴现联盛科技发展有限公司 | For the recognition of face payment information method for anti-counterfeit of still photo |
| CN109214335A (en) * | 2018-09-04 | 2019-01-15 | 高新兴科技集团股份有限公司 | A kind of alarm method and equipment |
| CN111241328A (en) * | 2018-11-28 | 2020-06-05 | 航天信息股份有限公司 | Identity authentication and identification service method and device, readable storage medium and electronic equipment |
| CN109886360A (en) * | 2019-03-25 | 2019-06-14 | 山东浪潮云信息技术有限公司 | A kind of certificate photo Classification and Identification based on deep learning and detection method without a hat on and system |
| CN110059576A (en) * | 2019-03-26 | 2019-07-26 | 北京字节跳动网络技术有限公司 | Screening technique, device and the electronic equipment of picture |
| CN111127639A (en) * | 2019-12-30 | 2020-05-08 | 深圳小佳科技有限公司 | Cloud-based face 3D model construction method, storage medium and system |
| CN111738742A (en) * | 2020-05-07 | 2020-10-02 | 广东电网有限责任公司 | Portrait data processing system for power customer service |
| CN111680280A (en) * | 2020-05-20 | 2020-09-18 | 青岛黄海学院 | Computer portrait recognition system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105138954B (en) | 2019-06-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105138954A (en) | Image automatic screening, query and identification system | |
| CN105046219B (en) | A kind of face identification system | |
| CN104951773B (en) | A kind of real-time face recognition monitoring system | |
| CN104866829B (en) | A kind of across age face verification method based on feature learning | |
| Polikovsky et al. | Facial micro-expression detection in hi-speed video based on facial action coding system (FACS) | |
| CN105740779B (en) | Method and device for detecting living human face | |
| Kähm et al. | 2d face liveness detection: An overview | |
| Medioni et al. | Identifying noncooperative subjects at a distance using face images and inferred three-dimensional face models | |
| Aurangzeb et al. | Human behavior analysis based on multi-types features fusion and Von Nauman entropy based features reduction | |
| Rouhi et al. | A review on feature extraction techniques in face recognition | |
| CN106650619A (en) | Human action recognition method | |
| CN103136516A (en) | Face recognition method and system fusing visible light and near-infrared information | |
| CN103544478A (en) | All-dimensional face detection method and system | |
| Jiang et al. | Application of a fast RCNN based on upper and lower layers in face recognition | |
| Strueva et al. | Student attendance control system with face recognition based on neural network | |
| CN111694980A (en) | Robust family child learning state visual supervision method and device | |
| Shen et al. | Image based hair segmentation algorithm for the application of automatic facial caricature synthesis | |
| Wu et al. | Ggvit: Multistream vision transformer network in face2face facial reenactment detection | |
| Anantharajah et al. | Quality based frame selection for video face recognition | |
| Abusham | Face verification using local graph stucture (LGS) | |
| Chen et al. | Illumination processing in face recognition | |
| Huang et al. | Research on learning state based on students’ attitude and emotion in class learning | |
| Dong | 3D face recognition neural network for digital human resource management | |
| Hu et al. | Behavioral Feature Description Method Based on the Vector Module Ratio and Vector Angle of Human Body Structure | |
| Jingxin et al. | 3D multi-poses face expression recognition based on action units |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |