CN108229335A

CN108229335A - It is associated with face identification method and device, electronic equipment, storage medium, program

Info

Publication number: CN108229335A
Application number: CN201711327139.7A
Authority: CN
Inventors: 李蔚琳; 张广程; 王晶; 张帅; 刘祖希; 朱斌
Original assignee: Shenzhen Sensetime Technology Co Ltd
Current assignee: Shenzhen Sensetime Technology Co Ltd
Priority date: 2017-12-12
Filing date: 2017-12-12
Publication date: 2018-06-29

Abstract

The embodiment of the invention discloses a kind of association face identification method and device, electronic equipment, storage medium, program, wherein, method includes：It obtains corresponding each capture of target person to record, each record of capturing includes location information and temporal information；It is recorded for each described capture, doubtful image set is obtained based on the corresponding location information and the temporal information；The doubtful image set includes at least one doubtful facial image, and the doubtful facial image includes corresponding doubtful face；Based on the number that each doubtful face occurs in each doubtful image set, the association personage of the target person is determined.Based on the above embodiment of the present invention provide method, realize, when association probability sets probability significantly, illustrate the doubtful face and target face be same pedestrian possibility it is very high, can using the doubtful facial image as be associated with facial image.

Description

It is associated with face identification method and device, electronic equipment, storage medium, program

Technical field

The present invention relates to computer vision technique, especially a kind of association face identification method and device, are deposited electronic equipment Storage media, program.

Background technology

When some criminal gang is obstructed or is required to look up to the clue of suspect, by searching for same administrative staff's energy of suspect Promising public security solve a case provide effective clue, can by investigate suspect lodging information, search setting time setting regions with The same administrative staff of MAC positioning suspects that suspect occurs simultaneously.

But offender may carry out check-in by false identity card, meanwhile, the mobile phone and number that when crime uses are not It is fixed, and possible less booting, therefore searched using the above method and can be potentially encountered many obstructions with administrative staff.

Invention content

A kind of association face recognition technology provided in an embodiment of the present invention.

One side according to embodiments of the present invention, a kind of association face identification method provided, including：

It obtains corresponding each capture of target person to record, each record of capturing includes location information and temporal information；

It is recorded for each described capture, doubtful image is obtained based on the corresponding location information and the temporal information Collection；The doubtful image set includes at least one doubtful facial image, and the doubtful facial image includes corresponding doubtful face；

Based on the number that each doubtful face occurs in each doubtful image set, the affiliated person of the target person is determined Object.

It is described to obtain the corresponding each candid photograph note of target person in another embodiment based on the above method of the present invention Record, including：

Corresponding each capture of target person is obtained based on picture pick-up device to record, the location information captured in record includes The position of the picture pick-up device, the temporal information captured in record include the picture pick-up device and capture the target person Period.

In another embodiment based on the above method of the present invention, the record of capturing further includes target facial image；

It is described that the corresponding each candid photograph record of target person is obtained based on picture pick-up device, including：

Sequence of frames of video is acquired based on picture pick-up device；The sequence of frames of video includes at least one video frame, the video Frame includes video human face image；

The target facial image is obtained based on the video human face image in the video frame.

In another embodiment based on the above method of the present invention, the video human face figure based in the video frame As obtaining the target facial image, including：

Using recognition of face network, the video human face figure in the recognition of face acquisition video frame is carried out to the video frame Picture；

Face in the video human face image is matched with the facial image of target person described in mesh, in response to institute The facial image similarity for stating face and the target person in video human face image is more than setting similarity, by the video Facial image is as the target facial image.

According to the relevant information of the correspondence target person from the video human face in the video frame, corresponded to described in The target facial image of relevant information；The relevant information is used to distinguish each target person.

In another embodiment based on the above method of the present invention, the doubtful face is with the target person identical Or similar time appears in same position；

It is described to be recorded for each described capture, it is obtained based on the corresponding location information and the temporal information doubtful Image set, including：

Obtain the location information and temporal information captured in record；

The time interval of the corresponding temporal information is obtained according to the temporal information；The time interval be with it is described when Between extend setting time forward and backward centered on information；

The corresponding associated video frame sequence of the time interval is acquired based on the corresponding picture pick-up device of the location information；Institute It states associated video frame sequence and includes at least one associated video frame；

Doubtful facial image is obtained based on the associated video frame, the doubtful facial image based on the acquisition obtains doubtful Image set.

In another embodiment based on the above method of the present invention, further include：

Exist in response to corresponding two time intervals of the location information and overlap；

Using the time that two time intervals overlap as coincidence section；

A correspondence being stored in based on the doubtful facial image that the coincidence section obtains in described two time intervals Doubtful image set in.

It is described to include first in the presence of two time intervals of coincidence in another embodiment based on the above method of the present invention Time interval and the second time interval, each doubtful facial image correspond to an acquisition time；

A correspondence being stored in based on the doubtful facial image that the coincidence section obtains in described two time intervals Doubtful image set in, including：

Obtain the corresponding acquisition time of doubtful facial image obtained based on the coincidence section；

It is less than or equal to the acquisition in response to the difference of the temporal information in the acquisition time and first time section The doubtful facial image is stored in the first time section pair by time and the difference of the temporal information in the second time interval In the doubtful image set answered；

In response to the temporal information in the acquisition time and first time section difference be more than the acquisition time with It is corresponding doubtful to be stored in second time interval by the difference of the temporal information in the second time interval for the doubtful facial image Like in image set.

In another embodiment based on the above method of the present invention, doubtful face figure is obtained based on the associated video frame Picture, including：

Using recognition of face network, the people in the recognition of face acquisition associated video frame is carried out to the associated video frame Face image；

Re-scheduling is carried out to the facial image of the acquisition, obtains at least one facial image as doubtful facial image.

In another embodiment based on the above method of the present invention, re-scheduling is carried out to the facial image of the acquisition, is obtained At least one facial image as doubtful facial image, including：

Facial image based on the acquisition establishes face track；The face track includes at least the one of corresponding same people A facial image；

The doubtful facial image of the corresponding face track is obtained based on the face track；Each doubtful face figure As a corresponding face track.

In another embodiment based on the above method of the present invention, the corresponding face is obtained based on the face track The doubtful facial image of track, including：

Face images in the face track based on quality from getting well to going bad are ranked up, sequentially obtain quality most The good facial image is as doubtful facial image.

Similarity-rough set is carried out to the facial image of the acquisition, is greater than or equal at least two of preset value from similarity A facial image is obtained in the facial image as doubtful facial image；The doubtful facial image is described similar Degree is more than or equal to top-quality facial image at least two facial images of preset value.

Cluster operation is performed to the acquisition facial image, obtains at least one cluster image set；Each dendrogram Image set includes at least one facial image；

Top-quality facial image is obtained as doubtful facial image from the cluster image set.

It is described to be based on each doubtful face in each doubtful figure in another embodiment based on the above method of the present invention The number occurred in image set determines the association personage of the target person, including：

Time that the number and the doubtful face occurred based on the target face is occurred in each doubtful image set Number calculates the association probability of the doubtful face and the target face；

It is greater than or equal to setting probability, output in response to the association probability of the doubtful face and the target facial image Association personage of the corresponding personage of the doubtful facial image as the target person.

In another embodiment based on the above method of the present invention, further include：In response to the doubtful face with it is described The association probability of target facial image is less than setting probability；

The association personage of the target person is not exported；

Or the doubtful facial image of the setting quantity that occurrence number is most in each doubtful image set is exported, it will Association personage of the corresponding personage of doubtful facial image of the output as the target person.

In another embodiment based on the above method of the present invention, the number and described that is occurred based on the target face The number that doubtful face occurs in each doubtful image set, calculates the doubtful face and being associated with for the target face is general Rate, including：

The number of the target face appearance is obtained based on the corresponding quantity for capturing record of the target person；Often A record correspondence of capturing accumulates once；

The number that doubtful facial image based on the correspondence target person occurs in each doubtful image set obtains The number that the doubtful face occurs；The doubtful facial image that same people is corresponded in a doubtful image set only preserves one It is a；

The number that the number divided by the target face occurred by the doubtful face occurs, obtains the doubtful face With the association probability of the target face.

In another embodiment based on the above method of the present invention, the doubtful people based on the correspondence target person The number that face image occurs in each doubtful image set obtains the number that the doubtful face occurs, including：

Judge in each doubtful image set with the presence or absence of the doubtful facial image of the corresponding target person；

In response to including the doubtful facial image of at least two corresponding target persons, by the corresponding target person The number that the quantity of the corresponding doubtful image set of doubtful facial image occurs as the doubtful face；

In response to including the doubtful facial image of a same people of correspondence, the number that the doubtful facial image occurs It is 1.

In another embodiment based on the above method of the present invention, judge in each doubtful image set with the presence or absence of pair The doubtful facial image of the target person is answered, including：

All doubtful facial images in each doubtful image set are obtained, the doubtful facial image are carried out two-by-two similar Degree compares；

It is greater than or equal to setting value, described two doubtful people in response to the similarity between two doubtful facial images Face image corresponds to the target person；

It is less than setting value, described two doubtful facial images in response to the similarity between two doubtful facial images The target person is not corresponded to.

In another embodiment based on the above method of the present invention, judge in each doubtful image set with the presence or absence of pair The doubtful facial image of same people is answered, including：

Obtain all doubtful facial images in each doubtful image set；

Cluster operation is performed to all doubtful facial images, obtains at least one cluster set；

Each doubtful facial image in each cluster set corresponds to the target person.

Other side according to embodiments of the present invention, a kind of association face identification device provided, including：

Acquiring unit is recorded, is recorded for obtaining corresponding each capture of target person, each record of capturing includes position Confidence ceases and temporal information；

Doubtful acquiring unit, for be directed to it is each it is described capture record, based on the corresponding location information and it is described when Between the doubtful image set of acquisition of information；The doubtful image set includes at least one doubtful facial image, the doubtful facial image Including corresponding doubtful face；

Determination unit is associated with, for the number occurred in each doubtful image set based on each doubtful face, determines institute State the association personage of target person.

In another embodiment based on above device of the present invention, the record acquiring unit, specifically for being based on taking the photograph It is recorded as equipment obtains corresponding each capture of target person, the location information captured in record includes the picture pick-up device Position, the temporal information captured in record include the period that the picture pick-up device captures the target person.

In another embodiment based on above device of the present invention, the record of capturing further includes target facial image；

The record acquiring unit, including：

Photographing module, for being based on picture pick-up device acquisition sequence of frames of video；The sequence of frames of video includes at least one regard Frequency frame, the video frame include video human face image；

Face acquisition module, for obtaining the target facial image based on the video human face image in the video frame.

In another embodiment based on above device of the present invention, the face acquisition module, specifically for utilizing people Face identifies network, and the video human face image in the recognition of face acquisition video frame is carried out to the video frame；By the video Face in facial image is matched with the facial image of target person described in mesh, in response in the video human face image The facial image similarity of face and the target person is more than setting similarity, using the video human face image as the mesh Mark facial image.

In another embodiment based on above device of the present invention, the face acquisition module, specifically for according to right The relevant information of the target person is answered from the video human face in the video frame, obtains the target of the corresponding relevant information Facial image；The relevant information is used to distinguish each target person.

In another embodiment based on above device of the present invention, the doubtful face is with the target person identical Or similar time appears in same position；

The doubtful acquiring unit, including：

Information module, for obtaining the location information and temporal information captured in record；

Interval determination module, for obtaining the time interval of the corresponding temporal information according to the temporal information；It is described Time interval is extends setting time forward and backward centered on the temporal information；

Related acquisition module, it is corresponding for being based on the location information corresponding picture pick-up device acquisition time interval Associated video frame sequence；The associated video frame sequence includes at least one associated video frame；

Doubtful determining module obtains doubtful facial image, doubting based on the acquisition for being based on the associated video frame Doubtful image set is obtained like facial image.

In another embodiment based on above device of the present invention, the doubtful acquiring unit further includes：

Section overlaps module, for there is weight in response to corresponding two time intervals of the location information It closes；Using the time that two time intervals overlap as coincidence section；

Doubtful memory module, the doubtful facial image for being obtained based on the coincidence section are stored in described two times In a corresponding doubtful image set in section.

It is described to include first in the presence of two time intervals of coincidence in another embodiment based on above device of the present invention Time interval and the second time interval, each doubtful facial image correspond to an acquisition time；

The doubtful memory module, it is corresponding specifically for obtaining the doubtful facial image based on coincidence section acquisition Acquisition time；

In another embodiment based on above device of the present invention, the doubtful determining module, including：

Face recognition module for utilizing recognition of face network, carries out recognition of face to the associated video frame and obtains institute State the facial image in associated video frame；

Re-scheduling module for carrying out re-scheduling to the facial image of the acquisition, obtains at least one facial image as doubtful Like facial image.

In another embodiment based on above device of the present invention, the re-scheduling module, including：

Module is established in track, and face track is established for the facial image based on the acquisition；The face track includes At least one facial image of corresponding same people；

Track re-scheduling module, for obtaining the doubtful facial image of the corresponding face track based on the face track； Each doubtful facial image corresponds to a face track.

In another embodiment based on above device of the present invention, the track re-scheduling module, specifically for described Face images in face track are ranked up based on quality from getting well to going bad, and sequentially obtain the top-quality face figure As being used as doubtful facial image.

In another embodiment based on above device of the present invention, the re-scheduling module, specifically for the acquisition Facial image carry out similarity-rough set, from similarity be greater than or equal to preset value at least two facial images in obtain One facial image is as doubtful facial image；The doubtful facial image is greater than or equal to preset value for the similarity At least two facial images in top-quality facial image.

In another embodiment based on above device of the present invention, the re-scheduling module, specifically for the acquisition Facial image performs cluster operation, obtains at least one cluster image set；Each cluster image set includes at least one Facial image；Top-quality facial image is obtained as doubtful facial image from the cluster image set.

In another embodiment based on above device of the present invention, the association determination unit, including：

Number acquisition module, for the number and the doubtful face that are occurred based on the target face each described doubtful The number occurred in image set calculates the association probability of the doubtful face and the target face；

Be associated with output module, for being more than in response to the association probability of the doubtful face and the target facial image or Equal to setting probability, association personage of the corresponding personage of the doubtful facial image as the target person is exported.

In another embodiment based on above device of the present invention, the association output module is additionally operable in response to institute The association probability for stating doubtful face and the target facial image is less than setting probability；The affiliated person of the target person is not exported Object；

In another embodiment based on above device of the present invention, the number acquisition module, including：

Targeted number module, for obtaining the target based on the corresponding quantity for capturing record of the target person The number that face occurs；Each record correspondence of capturing accumulates once；

Doubtful number module, for the doubtful facial image based on the correspondence target person in each doubtful image set The number of middle appearance obtains the number that the doubtful face occurs；Doubting for same people is corresponded in a doubtful image set One is only preserved like facial image；

Probability evaluation entity, time that number divided by the target face for occurring by the doubtful face occur Number obtains the association probability of the doubtful face and the target face.

In another embodiment based on above device of the present invention, the doubtful number module, including：

Judgment module, for judging in each doubtful image set with the presence or absence of the doubtful face of the corresponding target person Image；

Number statistical module, will for the doubtful facial image in response to including at least two corresponding target persons The quantity of the corresponding doubtful image set of doubtful facial image of the corresponding target person occurs as the doubtful face Number；In response to including the doubtful facial image of a same people of correspondence, the number that the doubtful facial image occurs It is 1.

In another embodiment based on above device of the present invention, the judgment module is each described specifically for obtaining The doubtful facial image is carried out similarity comparison by all doubtful facial images in doubtful image set two-by-two；In response to two Similarity between the doubtful facial image is greater than or equal to setting value, and described two doubtful facial images correspond to the target Personage；It is less than setting value in response to the similarity between two doubtful facial images, described two doubtful facial images are not The corresponding target person.

In another embodiment based on above device of the present invention, the judgment module is each described specifically for obtaining All doubtful facial images in doubtful image set；Cluster operation is performed to all doubtful facial images, is obtained at least one Cluster set；Each doubtful facial image in each cluster set corresponds to the target person.

Other side according to embodiments of the present invention, a kind of electronic equipment provided, including processor, the processor Including being associated with face identification device as described above.

Other side according to embodiments of the present invention, a kind of electronic equipment provided, including：Memory, for storing Executable instruction；

And processor, it completes to close as described above to perform the executable instruction for communicating with the memory Join the operation of face identification method.

Other side according to embodiments of the present invention, a kind of computer storage media provided, for storing computer The instruction that can be read, described instruction are performed the operation for performing association face identification method as described above.

Other side according to embodiments of the present invention, a kind of computer program provided, including computer-readable code, When the computer-readable code in equipment when running, the processor execution in the equipment is used to implement association as described above The instruction of face identification method.

A kind of association face identification method and device based on the above embodiment of the present invention offer, electronic equipment, storage are situated between Matter, program obtain corresponding each capture of target person and record, and each record of capturing includes location information and temporal information；For It is each to capture record, doubtful image set is obtained based on corresponding location information and temporal information；Pass through the corresponding meaning for capturing record Think image set and will be seen which personage puts appearance in the same time and place with target person；Based on each doubtful face in each doubtful image set The number of middle appearance, determines the association personage of target person, and number of pass times will be seen that suspicious figure went together with target person Probability passes through the association personage for determine the probability target person of going together.

Below by drawings and examples, technical scheme of the present invention is described in further detail.

Description of the drawings

The attached drawing of a part for constitution instruction describes the embodiment of the present invention, and is used to explain together with description The principle of the present invention.

With reference to attached drawing, according to following detailed description, the present invention can be more clearly understood, wherein：

Fig. 1 is the flow chart of present invention association face identification method one embodiment.

Fig. 2 is the structure diagram of present invention association face identification device one embodiment.

Fig. 3 is the structure diagram for realizing the terminal device of the embodiment of the present application or the electronic equipment of server.

Specific embodiment

Carry out the various exemplary embodiments of detailed description of the present invention now with reference to attached drawing.It should be noted that：Unless in addition have Body illustrates that the unlimited system of component and the positioned opposite of step, numerical expression and the numerical value otherwise illustrated in these embodiments is originally The range of invention.

Simultaneously, it should be appreciated that for ease of description, the size of the various pieces shown in attached drawing is not according to reality Proportionate relationship draw.

It is illustrative to the description only actually of at least one exemplary embodiment below, is never used as to the present invention And its application or any restrictions that use.

Technology, method and apparatus known to person of ordinary skill in the relevant may be not discussed in detail, but suitable In the case of, the technology, method and apparatus should be considered as part of specification.

It should be noted that：Similar label and letter represents similar terms in following attached drawing, therefore, once a certain Xiang Yi It is defined in a attached drawing, then in subsequent attached drawing does not need to that it is further discussed.

The embodiment of the present invention can be applied to computer system/server, can be with numerous other general or specialized calculating System environments or configuration operate together.Suitable for be used together with computer system/server well-known computing system, ring The example of border and/or configuration includes but not limited to：Personal computer system, server computer system, thin client, thick client Machine, hand-held or laptop devices, the system based on microprocessor, set-top box, programmable consumer electronics, NetPC Network PC, Minicomputer system, large computer system and distributed cloud computing technology environment including any of the above described system, etc..

Computer system/server can be in computer system executable instruction (such as journey performed by computer system Sequence module) general linguistic context under describe.In general, program module can include routine, program, target program, component, logic, number According to structure etc., they perform specific task or realize specific abstract data type.Computer system/server can be with Implement in distributed cloud computing environment, in distributed cloud computing environment, task is long-range by what is be linked through a communication network Manage what equipment performed.In distributed cloud computing environment, program module can be located at the Local or Remote meter for including storage device It calculates in system storage medium.

As government and the public are for the pay attention to day by day of anti-terrorism and public safety, there are more than hundreds thousand of grades in many cities Monitoring device come into operation；With reference to face recognition technology, rapidly a suspect can be positioned in the video of magnanimity.

In the implementation of the present invention, inventor has found, has at least the following problems：Be obstructed when the clue of suspect or When requiring to look up some criminal gang, even if applying face recognition technology, useful information can not be also obtained.

Fig. 1 is the flow chart of present invention association face identification method one embodiment.As shown in Figure 1, the embodiment method Including：

Step 101, it obtains corresponding each capture of target person to record, each record of capturing includes location information and time letter Breath.

Specifically, the candid photograph record of acquisition is to correspond to different target persons respectively, based on each target person at certain One position and sometime appearance are primary, that is, obtain one and capture record, location information and time are included at least in candid photograph record Information can also include the facial image captured of corresponding target person.

Step 102, it is recorded for each capture, doubtful image set is obtained based on corresponding location information and temporal information.

Wherein, each doubtful image set corresponds to one and captures record, and doubtful image set includes at least one doubtful face Image, doubtful facial image include corresponding doubtful face, and doubtful face refers in the same or similar time go out with target face Present same position.

Step 103, the number occurred in each doubtful image set based on each doubtful face determines the affiliated person of target person Object.

The association personage of target person is determined based on the number of doubtful face appearance, can be occurred based on doubtful face The number that the ratio of number and target person occurrence number is more than predetermined threshold value or is occurred based on doubtful face is more than default time Number can also be the highest doubtful face of the one or more occurrence numbers of acquisition.

Based on a kind of association face identification method that the above embodiment of the present invention provides, obtain that target person is corresponding respectively grabs Record is clapped, each record of capturing includes location information and temporal information；Record is captured for each, based on corresponding location information Doubtful image set is obtained with temporal information；Which personage and target person will be seen that by the corresponding meaning image set for capturing record Object puts appearance in the same time and place；Based on the number that each doubtful face occurs in each doubtful image set, the association of target person is determined Personage, number of pass times will be seen that the probability that suspicious figure goes together with target person, pass through determine the probability target person of going together It is associated with personage.

In a specific example of present invention association face identification method above-described embodiment, operation 101 includes：

Corresponding each capture of target person is obtained based on picture pick-up device to record, the location information captured in record includes camera shooting The position of equipment captures the temporal information in record and includes the period that picture pick-up device captures target person.

Specifically, generally use picture pick-up device acquisition video frame is obtained based on video frame and captures record, is typically based on acquisition The location information of the location determination of picture pick-up device of record candid photograph record is captured, the position of picture pick-up device can be complete by GPS The location information that the positioning device that ball alignment system or picture pick-up device carry captures obtains, and is obtained by picture pick-up device and captures record Time determine the candid photograph record temporal information, picture pick-up device obtain capture record time can be carried by picture pick-up device Clock obtain or according to other timing devices obtain.

Optionally, in a specific example of present invention association face identification method the various embodiments described above, record is captured Further include target facial image；

Operation 101 includes：

Sequence of frames of video is acquired based on picture pick-up device；Sequence of frames of video includes at least one video frame, and video frame includes Video human face image；

Specifically, sequence of frames of video is acquired by picture pick-up device, is wrapped respectively in each frame video frame in sequence of frames of video A video human face image is included, determining for target person could be realized based on video human face image.

Target facial image is obtained based on the video human face image in video frame.

Specifically, target facial image is obtained, recognition of face network can be utilized, recognition of face acquisition is carried out to video frame Video human face image in video frame；

Face in video human face image is matched with the facial image of target person, in response to video human face image In face and target person facial image similarity be more than setting similarity, using video human face image as target face figure Picture.

By the embodiment, by the facial image of target person and the video human face image progress in the video frame of acquisition Match, using matched video human face image as target facial image, can be obtained simultaneously by the process for acquiring target facial image The temporal information of corresponding target facial image, and the position based on picture pick-up device can determine the position letter of the target facial image Breath.

Specifically, identify that the video human face image in video frame can recognize video by recognition of face real-time performance After facial image, neural network or other dividing methods can also be applied, video human face image is divided from video frame It cuts, using obtained independent video human face image as target facial image, in order to subsequently identify association face；And due to obtaining The target facial image taken be in order to identify target face, therefore, it is necessary to by the face in facial image and target face into Row matching, the similarity of two faces can be obtained based on the distance between character pair that neural network obtains or angle, will Similarity is more than the facial image of setting similarity as target facial image.

In a specific example of present invention association face identification method the various embodiments described above, based on regarding in video frame Frequency facial image obtains target facial image, can also include：

According to the relevant information of corresponding target person from the video human face in video frame, the mesh of corresponding relevant information is obtained Mark facial image；Relevant information is used to distinguish each target person.

Specifically, relevant information can include name, identification card number, pre-assigned identification number of target person etc., directly It connected relevant information and was matched with the video human face image that picture pick-up device obtains, and obtained target facial image.

Another embodiment of present invention association face identification method, on the basis of the various embodiments described above, doubtful face With target person same position is appeared in the same or similar time；

Operation 102 includes：

Obtain the location information and temporal information captured in record；

The time interval of corresponding temporal information is obtained according to temporal information；Time interval is to be distinguished centered on temporal information Extend setting time forwards, backwards；

Based on the corresponding associated video frame sequence in the corresponding picture pick-up device acquisition time section of location information；Associated video frame Sequence includes at least one associated video frame；

Doubtful facial image is obtained based on associated video frame, the doubtful facial image based on acquisition obtains doubtful image set.

In the present embodiment, in order to find out the time front and rear camera shooting in suspect's (target facial image) appearance of corresponding position The face that equipment is captured, can be according to actual conditions, and it is front and rear certain time to set time range, such as：Time range is set For (T_i-T,T_i+ T), then in T_i(temporal information) before and after the moment, Y_iThe face set that place occurs is denoted as F (T_i±T,Y_i), to every The face that occurs in a face set carries out re-scheduling, i.e., everyone only deposits a candid photograph picture, and the face set after re-scheduling is denoted as F_{Re-scheduling}(T_i±T,Y_i)。

In a specific example of present invention association face identification method the various embodiments described above, further include：

Exist in response to corresponding two time intervals of a location information and overlap；

Using the time that two time intervals overlap as coincidence section；

A corresponding doubtful figure being stored in two time intervals based on the doubtful facial image for overlapping section acquisition In image set.

When it is implemented, since time interval is a period, it is understood that there may be two time intervals have what is partially overlapped Situation, if at this point, two doubtful images will be recorded without distinguishing in the doubtful facial image in section is overlapped respectively It concentrates, the record of doubtful facial image number can be had an impact in this way, the target facial image for leading to output is inaccurate, because This, is recorded only in the doubtful facial image in overlapping section in a doubtful image set.

In a specific example of present invention association face identification method the various embodiments described above, exist and overlap two times Section includes first time section and the second time interval, and each doubtful facial image corresponds to an acquisition time；

A corresponding doubtful figure being stored in two time intervals based on the doubtful facial image for overlapping section acquisition In image set, including：

It obtains based on the corresponding acquisition time of doubtful facial image for overlapping section acquisition；

It is less than or equal to acquisition time and the in response to the difference of the temporal information in acquisition time and first time section The difference of temporal information in two time intervals, by the corresponding doubtful image set in doubtful facial image deposit first time section In；

It is more than acquisition time and the second time in response to the difference of the temporal information in acquisition time and first time section Doubtful facial image is stored in the corresponding doubtful image set of the second time interval by the difference of the temporal information in section.

On the basis of above-described embodiment, in order to determine the doubtful facial image overlapped in section being stored in first time area Between corresponding doubtful image set or the corresponding doubtful image set of the second time interval, need judge obtain the doubtful facial image Temporal information of the acquisition time apart from first time section be near or the temporal information of the second time interval of distance is near.Specifically Ground, if suspect is in the same localities, appearance is multiple, such as in T₁And T₂There is (T in time₁<T₂), and (T₁-T,T₁+ T) with (T₂-T,T₂+ T) there are coincidence in two periods, then in T₁To T₂People (the acquisition time T of appearance is put in period at this point_j), if T_j-T₁<T₂-T_j, then this face be included into (T₁-T,T₁+ T) period, otherwise it is included into (T₂-T,T₂+T)。

In a specific example of present invention association face identification method the various embodiments described above, obtained based on associated video frame Doubtful facial image is taken, including：

Using recognition of face network, the facial image in recognition of face acquisition associated video frame is carried out to associated video frame；

Re-scheduling is carried out to the facial image of acquisition, obtains at least one facial image as doubtful facial image.

Specifically, identify that the facial image in associated video image can recognize people by recognition of face real-time performance After face image, neural network or other dividing methods can also be applied, facial image is split from video image, and In order to ensure the independent facial image that will be obtained to be needed to carry out not to doubtful facial image repeat count in doubtful image set Re-scheduling, as doubtful facial image, in order to which the number subsequently occurred to doubtful facial image counts.

Optionally, in a specific example of present invention association face identification method the various embodiments described above, to acquisition Facial image carries out re-scheduling, obtains at least one facial image as doubtful facial image, can also include：

Facial image based on acquisition establishes face track；Face track includes at least one face figure of corresponding same people Picture；

The doubtful facial image of corresponding face track is obtained based on face track；Each doubtful facial image corresponds to an institute State face track.

By establishing face track to each face, a facial image is extracted in face track as doubtful face figure Picture, the doubtful facial image specifically obtained are to obtain top-quality facial image in facial image, specifically can be by right Face images in face track are ranked up based on quality from getting well to going bad, and are sequentially obtained top-quality facial image and are made For doubtful facial image.

Optionally, in a specific example of present invention association face identification method the various embodiments described above, based on face Track obtains the doubtful facial image of corresponding face track, including：

Face images in face track based on quality from getting well to going bad are ranked up, are sequentially obtained top-quality Facial image is as doubtful facial image.

In one or more embodiments, optionally, re-scheduling is carried out to the facial image of acquisition, obtains at least one face Image as doubtful facial image, including：

Similarity-rough set is carried out to the facial image of acquisition, at least two faces of preset value are greater than or equal to from similarity A facial image is obtained in image as doubtful facial image.

Wherein, doubtful facial image is greater than or equal to best in quality at least two facial images of preset value for similarity Facial image.

Specifically, the quality of facial image can according to the size of face, angle, whether block etc. and to be sentenced It is disconnected.The size of face by deep learning algorithm carry out Face datection, orient face frame, according to the length and width pixel of face frame into Row judges.Whether angle is blocked and can be judged using deep learning, for example judge facial angle, in training data Different facial angles is represented with different values, by having the learning method of supervision, the parameter update mode of back transfer, training Model.After the completion of model training, picture is inputted, you can provide quality evaluation as a result, essence or a kind of process of classification.

In a specific example of present invention association face identification method the various embodiments described above, to the facial image of acquisition Re-scheduling is carried out, at least one facial image is obtained as doubtful facial image, can also include：

Cluster operation is performed to obtaining facial image, obtains at least one cluster image set；It is wrapped in each cluster image set Include at least one facial image；

Top-quality facial image is obtained as doubtful facial image from cluster image set.

In the present embodiment, the facial image of the identical doubtful face of correspondence is polymerized to by a set by cluster, based on one In a cluster image set, each doubtful face only needs a doubtful facial image, and therefore, it is necessary to be obtained from the set that polymerization obtains Top-quality facial image is obtained as doubtful facial image, subsequently to be matched with target person.

Another embodiment of present invention association face identification method, on the basis of the various embodiments described above, 103 packet of operation It includes：

The number that the number and doubtful face occurred based on target face is occurred in each doubtful image set, calculates doubtful people The association probability of face and target face；

It is greater than or equal to setting probability in response to the association probability of doubtful face and target facial image, exports doubtful face Association personage of the corresponding personage of image as target person.

In the present embodiment, the number that occurs in each doubtful image set by the doubtful facial image of same people, it may be determined that The number that the doubtful same pedestrian and target person occur simultaneously, and time that be target person occurred of the quantity of all doubtful image sets Number, the ratio of number that doubtful same pedestrian and target person are occurred is as association probability.The specific formula for calculating association probability is such as Shown in formula (1)：

P(X_i)=N (X_i)/N (S) formula (1)

Wherein, N (X_i) refer to candid photograph doubtful same pedestrian doubtful facial image X_iNumber (same place, same a period of time of appearance Between section only count once), N (S) refer to suspect target facial image S occur number, if P (X_i)=100% represents several at this A doubtful same pedestrian in place goes together with suspect.

In a specific example of present invention association face identification method the various embodiments described above, further include：In response to doubting It is less than setting probability like the association probability of face and target facial image；

The association personage of target person is not exported；

Or the doubtful facial image of the setting quantity that occurrence number is most in each doubtful image set is exported, by doubting for output Like association personage of the corresponding personage of facial image as target person.

It specifically, can be according to tool when the association probability of doubtful face and target facial image both less than sets probability Body situation (based on association probability, it is not necessary to output association personage) selection does not export the association personage of target person, may be used also With (to export based on the big association personage of relevance, association probability is auxiliary) output as the case may be in each doubtful image set Association personage of the doubtful facial image of the most setting quantity of middle occurrence number as target person.

In a specific example of present invention association face identification method the various embodiments described above, occurred based on target face Number and the number that occurs in each doubtful image set of doubtful face, calculate the association probability of doubtful face and target face, Including：

The number of target face appearance is obtained based on the corresponding quantity for capturing record of target person；It is each to capture record pair It should accumulate once；

The number that doubtful facial image based on corresponding target person occurs in each doubtful image set, obtains doubtful face The number of appearance；The doubtful facial image that same people is corresponded in a doubtful image set only preserves one；

The number that the number divided by target face occurred by doubtful face occurs, obtains doubtful face and the target person The association probability of face.

Specifically, it is obtained based on the appearance of target person due to capturing record, target person occurs Number be the corresponding quantity for capturing record, and the number that doubtful facial image will appear then passes through and counts a doubtful face Image appears in be obtained in which doubtful image set, due to each doubtful facial image in a doubtful image set only there are one Image or no image, therefore, by counting the doubtful image set quantity occurred, you can obtain the number that doubtful face occurs.

In a specific example of present invention association face identification method the various embodiments described above, based on corresponding target person The number that occurs in each doubtful image set of doubtful facial image, obtain the number that doubtful face occurs, including：

Judge in each doubtful image set with the presence or absence of the doubtful facial image of corresponding target person；

In response to including the doubtful facial image of at least two correspondence target persons, by the doubtful face of corresponding target person The number that the quantity of the corresponding doubtful image set of image occurs as doubtful face；

In response to including the doubtful facial image of a same people of correspondence, the number that doubtful facial image occurs is 1.

Specifically, it in order to realize the counting to doubtful facial image, needs to judge which the doubtful facial image appears in In doubtful image set, the number of the cumulative doubtful image set occurred can obtain the corresponding number of doubtful facial image.

In a specific example of present invention association face identification method the various embodiments described above, each doubtful image set is judged In with the presence or absence of corresponding target person doubtful facial image, including：

All doubtful facial images in each doubtful image set are obtained, doubtful facial image is subjected to similarity comparison two-by-two；

It is greater than or equal to setting value, two doubtful facial images pair in response to the similarity between two doubtful facial images Answer target person；

It is less than setting value in response to the similarity between two doubtful facial images, two doubtful facial images do not correspond to mesh Mark personage.

Specifically, judge whether two doubtful facial images correspond to target person, it can be by by doubtful facial image two Two carry out similarity judgements, and the similarity between two doubtful facial images can be special by the face obtained based on neural network The distance between sign or the distance between angle-determining, face characteristic or angle can be the distance between face feature vector or Angle-determining, when the similarity of two facial images reaches setting value, it may be determined that the two facial images correspond to same people.

In a specific example of present invention association face identification method the various embodiments described above, each doubtful image set is judged In with the presence or absence of corresponding same people doubtful facial image, including：

Obtain all doubtful facial images in each doubtful image set；

Each doubtful facial image in each cluster set corresponds to target person.

Specifically, judge whether two doubtful facial images correspond to another method of target person, be to all faces Cluster analysis is carried out, each grouping is judged as same person, and clustering method includes Kmeans, DBSCAN and other innovatory algorithms. K-means algorithms receive parameter k；Then n data object being previously entered is divided into k cluster so that being obtained Cluster meets：Object similarity in same cluster is higher；And the object similarity in different clusters is smaller；Clustering similarity is One " center object " (center of attraction) is obtained come what is calculated using the mean value of object in each cluster.DBSCAN (Density-Based Spatial Clustering of Applications with Noise) is that a comparison has representative The density-based algorithms of property.From dividing and hierarchy clustering method is different, cluster is defined as point that density is connected most by it Big collection and can find arbitrary shape being cluster with region division highdensity enough in the spatial database of noise The cluster of shape.

One of ordinary skill in the art will appreciate that：Realizing all or part of step of above method embodiment can pass through The relevant hardware of program instruction is completed, and aforementioned program can be stored in a computer read/write memory medium, the program When being executed, step including the steps of the foregoing method embodiments is performed；And aforementioned storage medium includes：ROM, RAM, magnetic disc or light The various media that can store program code such as disk.

Fig. 2 is the structure diagram of present invention association face identification device one embodiment.The device of the embodiment can be used In the above-mentioned each method embodiment of the realization present invention.As shown in Fig. 2, the device of the embodiment includes：

Acquiring unit 21 is recorded, is recorded for obtaining corresponding each capture of target person, each record of capturing includes position Information and temporal information.

Doubtful acquiring unit 22 is recorded for being directed to each capture, is obtained based on corresponding location information and temporal information Doubtful image set.

Wherein, doubtful image set includes at least one doubtful facial image, and doubtful facial image includes corresponding doubtful people Face.

Determination unit 23 is associated with, for the number occurred in each doubtful image set based on each doubtful face, determines target The association personage of personage.

In a specific example of present invention association face identification device above-described embodiment, acquiring unit 21, tool are recorded Body is used to record based on corresponding each capture of picture pick-up device acquisition target person, and the location information captured in record is set including camera shooting Standby position captures the temporal information in record and includes the period that picture pick-up device captures target person.

In a specific example of present invention association face identification device above-described embodiment, capture record and further include target Facial image；

The record acquiring unit 21, including：

Photographing module, for being based on picture pick-up device acquisition sequence of frames of video；Sequence of frames of video includes at least one video frame, Video frame includes video human face image；

Face acquisition module, for obtaining target facial image based on the video human face image in video frame.

In a specific example of present invention association face identification device above-described embodiment, face acquisition module, specifically For utilizing recognition of face network, the video human face image in recognition of face acquisition video frame is carried out to video frame；By video people Face in face image is matched with the facial image of mesh target person, in response to the face and target in video human face image The facial image similarity of personage is more than setting similarity, using video human face image as target facial image.

In a specific example of present invention association face identification device above-described embodiment, face acquisition module, specifically For from the video human face in video frame, obtaining the target person of corresponding relevant information according to the relevant information of corresponding target person Face image；Relevant information is used to distinguish each target person.

Another embodiment of present invention association face identification device, on the basis of the various embodiments described above, doubtful face With target person same position is appeared in the same or similar time；

Doubtful acquiring unit 22, including：

Interval determination module, for obtaining the time interval of corresponding temporal information according to temporal information；Time interval be with Extend setting time centered on temporal information forward and backward；

Related acquisition module, for being based on the corresponding associated video in the corresponding picture pick-up device acquisition time section of location information Frame sequence；Associated video frame sequence includes at least one associated video frame；

Doubtful determining module obtains doubtful facial image, the doubtful face figure based on acquisition for being based on associated video frame As obtaining doubtful image set.

In a specific example of present invention association face identification device above-described embodiment, doubtful acquiring unit 22, also Including：

Section overlaps module, is overlapped for existing in response to corresponding two time intervals of a location information；By two The time that time interval overlaps is as coincidence section；

Doubtful memory module, for being stored in two time intervals based on the doubtful facial image for overlapping section acquisition In one corresponding doubtful image set.

In a specific example of present invention association face identification device above-described embodiment, exist and overlap two time zones Between include first time section and the second time interval, each doubtful facial image corresponds to an acquisition time；

Doubtful memory module, during specifically for obtaining based on the corresponding acquisition of doubtful facial image that section obtains is overlapped Between；

In a specific example of present invention association face identification device the various embodiments described above, doubtful determining module, packet It includes：

Face recognition module for utilizing recognition of face network, carries out associated video frame recognition of face acquisition correlation and regards Facial image in frequency frame；

Re-scheduling module for carrying out re-scheduling to the facial image of acquisition, obtains at least one facial image as doubtful people Face image.

In a specific example of present invention association face identification device the various embodiments described above, re-scheduling module, including：

Module is established in track, and face track is established for the facial image based on acquisition；It is same that face track includes correspondence At least one facial image of people；

Track re-scheduling module, for obtaining the doubtful facial image of corresponding face track based on face track；It is each doubtful Facial image corresponds to a face track.

In a specific example of present invention association face identification device the various embodiments described above, track re-scheduling module, tool Body sequentially obtains top-quality people for being ranked up to the face images in face track based on quality from getting well to going bad Face image is as doubtful facial image.

In one or more embodiments, optionally, re-scheduling module carries out similar specifically for the facial image to acquisition Degree compares, and is greater than or equal at least two facial images of preset value from similarity and obtains a facial image as doubtful people Face image；Doubtful facial image is greater than or equal to top-quality face at least two facial images of preset value for similarity Image.

In a specific example of present invention association face identification device the various embodiments described above, re-scheduling module is specific to use In performing cluster operation to obtaining facial image, at least one cluster image set is obtained；Each cluster image set is included at least One facial image；Top-quality facial image is obtained as doubtful facial image from cluster image set.

Another embodiment of present invention association face identification device, on the basis of the various embodiments described above, association determines Unit 23, including：

Number acquisition module, number and doubtful face for being occurred based on target face are occurred in each doubtful image set Number, calculate the association probability of doubtful face and target face；

Output module is associated with, for being greater than or equal to setting in response to the association probability of doubtful face and target facial image Probability exports association personage of the corresponding personage of doubtful facial image as target person.

In the present embodiment, the number that occurs in each doubtful image set by the doubtful facial image of same people, it may be determined that The number that the doubtful same pedestrian and target person occur simultaneously, and time that be target person occurred of the quantity of all doubtful image sets Number, the ratio of number that doubtful same pedestrian and target person are occurred is as association probability.The specific formula for calculating association probability is such as Shown in formula (1) above.

In a specific example of present invention association face identification device the various embodiments described above, output module is associated with, also For being less than setting probability in response to the association probability of doubtful face and target facial image；The affiliated person of target person is not exported Object；

In a specific example of present invention association face identification device the various embodiments described above, number acquisition module, packet It includes：

Targeted number module, for obtaining time of target face appearance based on the corresponding quantity for capturing record of target person Number；Each record correspondence of capturing accumulates once；

Doubtful number module occurs for the doubtful facial image based on corresponding target person in each doubtful image set Number obtains the number that doubtful face occurs；The doubtful facial image that same people is corresponded in a doubtful image set only preserves One；

Probability evaluation entity, the number that number divided by target face for occurring by doubtful face occur, is doubted Like face and the association probability of target face.

In a specific example of present invention association face identification device the various embodiments described above, doubtful number module, packet It includes：

Judgment module, for judging in each doubtful image set with the presence or absence of the doubtful facial image of corresponding target person；

Number statistical module, for the doubtful facial image in response to including at least two correspondence target persons, by correspondence The number that the quantity of the corresponding doubtful image set of doubtful facial image of target person occurs as doubtful face；In response to including The doubtful facial image of one same people of correspondence, the number that doubtful facial image occurs are 1.

In a specific example of present invention association face identification device the various embodiments described above, judgment module is specific to use In obtaining all doubtful facial images in each doubtful image set, doubtful facial image is subjected to similarity comparison two-by-two；In response to Similarity between two doubtful facial images is greater than or equal to setting value, and two doubtful facial images correspond to target person；It rings Setting value should be less than in the similarity between two doubtful facial images, two doubtful facial images do not correspond to target person.

In a specific example of present invention association face identification device the various embodiments described above, judgment module is specific to use All doubtful facial images in each doubtful image set of acquisition；Cluster operation is performed to all doubtful facial images, is obtained at least One cluster set；Each doubtful facial image in each cluster set corresponds to target person.

One side according to embodiments of the present invention, a kind of electronic equipment provided, including processor, processor includes this The device of deploying to ensure effective monitoring and control of illegal activities of the invention association any of the above-described embodiment of face identification method.

One side according to embodiments of the present invention, a kind of electronic equipment provided, including：Memory, can for storing Execute instruction；

And processor, for being associated with recognition of face thereby completing the present invention with memory communication to perform executable instruction The operation of any of the above-described embodiment of method.

A kind of one side according to embodiments of the present invention, the computer storage media provided, can for storing computer The instruction of reading, instruction are performed the operation for performing the present invention association any of the above-described embodiment of face identification method.

One side according to embodiments of the present invention, a kind of computer program provided, including computer-readable code, when For computer-readable code when being run in equipment, the processor execution in the equipment is used to implement present invention association recognition of face side The instruction of method any one embodiment.

The embodiment of the present invention additionally provides a kind of electronic equipment, such as can be mobile terminal, personal computer (PC), put down Plate computer, server etc..Below with reference to Fig. 3, it illustrates suitable for being used for realizing the terminal device of the embodiment of the present application or service The structure diagram of the electronic equipment 300 of device：As shown in figure 3, computer system 300 includes one or more processors, communication Portion etc., one or more of processors are for example：One or more central processing unit (CPU) 301 and/or one or more Image processor (GPU) 313 etc., processor can according to the executable instruction being stored in read-only memory (ROM) 302 or From the executable instruction that storage section 308 is loaded into random access storage device (RAM) 303 perform various appropriate actions and Processing.Communication unit 312 may include but be not limited to network interface card, and the network interface card may include but be not limited to IB (Infiniband) network interface card.

Processor can communicate with read-only memory 302 and/or random access storage device 330 to perform executable instruction, It is connected by bus 304 with communication unit 312 and is communicated through communication unit 312 with other target devices, is implemented so as to complete the application The corresponding operation of any one method that example provides records for example, obtaining corresponding each capture of target person, each to capture record packet Include location information and temporal information；Record is captured for each, doubtful figure is obtained based on corresponding location information and temporal information Image set；Based on the number that each doubtful face occurs in each doubtful image set, the association personage of target person is determined.

In addition, in RAM 303, it can also be stored with various programs and data needed for device operation.CPU301、ROM302 And RAM303 is connected with each other by bus 304.In the case where there is RAM303, ROM302 is optional module.RAM303 is stored Executable instruction is written in executable instruction into ROM302 at runtime, and it is above-mentioned logical that executable instruction performs processor 301 The corresponding operation of letter method.Input/output (I/O) interface 305 is also connected to bus 304.Communication unit 312 can be integrally disposed, It may be set to be with multiple submodule (such as multiple IB network interface cards), and in bus link.

I/O interfaces 305 are connected to lower component：Importation 306 including keyboard, mouse etc.；It is penetrated including such as cathode The output par, c 307 of spool (CRT), liquid crystal display (LCD) etc. and loud speaker etc.；Storage section 308 including hard disk etc.； And the communications portion 309 of the network interface card including LAN card, modem etc..Communications portion 309 via such as because The network of spy's net performs communication process.Driver 310 is also according to needing to be connected to I/O interfaces 305.Detachable media 311, such as Disk, CD, magneto-optic disk, semiconductor memory etc. are mounted on driver 310, as needed in order to be read from thereon Computer program be mounted into storage section 308 as needed.

Need what is illustrated, framework as shown in Figure 3 is only a kind of optional realization method, can root during concrete practice The component count amount and type of above-mentioned Fig. 3 are selected, are deleted, increased or replaced according to actual needs；It is set in different function component Put, can also be used it is separately positioned or integrally disposed and other implementations, such as GPU and CPU separate setting or can be by GPU collection Into on CPU, communication unit separates setting, can also be integrally disposed on CPU or GPU, etc..These interchangeable embodiments Each fall within protection domain disclosed by the invention.

Particularly, in accordance with an embodiment of the present disclosure, it may be implemented as computer above with reference to the process of flow chart description Software program.For example, embodiment of the disclosure includes a kind of computer program product, it is machine readable including being tangibly embodied in Computer program on medium, computer program are included for the program code of the method shown in execution flow chart, program code It may include the corresponding instruction of corresponding execution method and step provided by the embodiments of the present application, for example, it is corresponding each to obtain target person Record is captured, each record of capturing includes location information and temporal information；Record is captured for each, is believed based on corresponding position Breath and temporal information obtain doubtful image set；Based on the number that each doubtful face occurs in each doubtful image set, target is determined The association personage of personage.In such embodiments, which can be downloaded by communications portion 309 from network And installation and/or from detachable media 311 be mounted.When the computer program is performed by central processing unit (CPU) 301, Perform the above-mentioned function of being limited in the present processes.

Methods and apparatus of the present invention, equipment may be achieved in many ways.For example, software, hardware, firmware can be passed through Or any combinations of software, hardware, firmware realize methods and apparatus of the present invention, equipment.The step of for method Sequence is stated merely to illustrate, the step of method of the invention is not limited to sequence described in detail above, unless with other Mode illustrates.In addition, in some embodiments, the present invention can be also embodied as recording program in the recording medium, this A little programs include being used to implement machine readable instructions according to the method for the present invention.Thus, the present invention also covering stores to hold The recording medium of the program of row according to the method for the present invention.

Description of the invention provides for the sake of example and description, and is not exhaustively or will be of the invention It is limited to disclosed form.Many modifications and variations are obvious for the ordinary skill in the art.It selects and retouches It states embodiment and is to more preferably illustrate the principle of the present invention and practical application, and those of ordinary skill in the art is enable to manage The solution present invention is so as to design the various embodiments with various modifications suitable for special-purpose.

Claims

1. a kind of association face identification method, which is characterized in that including：

It is recorded for each described capture, doubtful image set is obtained based on the corresponding location information and the temporal information； The doubtful image set includes at least one doubtful facial image, and the doubtful facial image includes corresponding doubtful face；

Based on the number that each doubtful face occurs in each doubtful image set, the association personage of the target person is determined.

2. according to the method described in claim 1, it is characterized in that, the corresponding each candid photograph record of the acquisition target person, packet It includes：

Corresponding each capture of target person is obtained based on picture pick-up device to record, the location information captured in record includes described The position of picture pick-up device, the temporal information captured in record include the time that the picture pick-up device captures the target person Section.

3. according to the method described in claim 2, it is characterized in that, the record of capturing further includes target facial image；

Sequence of frames of video is acquired based on picture pick-up device；The sequence of frames of video includes at least one video frame, in the video frame Including video human face image；

4. according to the method described in claim 3, it is characterized in that, the video human face image based in the video frame obtains The target facial image is obtained, including：

Using recognition of face network, the video human face image in the recognition of face acquisition video frame is carried out to the video frame；

Face in the video human face image with the facial image of target person described in mesh is matched, is regarded in response to described The facial image similarity of face and the target person in frequency facial image is more than setting similarity, by the video human face Image is as the target facial image.

5. according to the method described in claim 3, it is characterized in that, the video human face image based in the video frame obtains The target facial image is obtained, including：

According to the relevant information of the correspondence target person from the video human face in the video frame, the corresponding correlation is obtained The target facial image of information；The relevant information is used to distinguish each target person.

6. a kind of association face identification device, which is characterized in that including：

Acquiring unit is recorded, is recorded for obtaining corresponding each capture of target person, each record of capturing includes position letter Breath and temporal information；

Doubtful acquiring unit is recorded for being directed to each described capture, is believed based on the corresponding location information and the time Breath obtains doubtful image set；The doubtful image set includes at least one doubtful facial image, and the doubtful facial image includes Corresponding doubtful face；

Determination unit is associated with, for the number occurred in each doubtful image set based on each doubtful face, determines the mesh Mark the association personage of personage.

7. a kind of electronic equipment, which is characterized in that including processor, the processor includes the affiliated person described in claim 6 Face identification device.

8. a kind of electronic equipment, which is characterized in that including：Memory, for storing executable instruction；

And processor, for communicating to perform the executable instruction so as to complete claim 1 to 5 times with the memory The operation of the association face identification method of meaning one.

9. a kind of computer storage media, for storing computer-readable instruction, which is characterized in that described instruction is performed When perform claim require 1 to 5 any one described in be associated with face identification method operation.

10. a kind of computer program, including computer-readable code, which is characterized in that when the computer-readable code is being set During standby upper operation, the processor execution in the equipment, which is used to implement described in claim 1 to 5 any one, is associated with recognition of face The instruction of method.