CN107016322A - A kind of method and device of trailing personnel analysis - Google Patents
A kind of method and device of trailing personnel analysis Download PDFInfo
- Publication number
- CN107016322A CN107016322A CN201610058847.4A CN201610058847A CN107016322A CN 107016322 A CN107016322 A CN 107016322A CN 201610058847 A CN201610058847 A CN 201610058847A CN 107016322 A CN107016322 A CN 107016322A
- Authority
- CN
- China
- Prior art keywords
- picture
- candid photograph
- personnel
- point
- information
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/30—Scenes; Scene-specific elements in albums, collections or shared content, e.g. social network photos or video
-
- 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
- G06V40/172—Classification, e.g. identification
- G06V40/173—Classification, e.g. identification face re-identification, e.g. recognising unknown faces across different face tracks
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Human Computer Interaction (AREA)
- Image Analysis (AREA)
- Alarm Systems (AREA)
Abstract
The application provides the method and device that a kind of trailing personnel analyze, and this method includes, and searches for each point picture of capturing and is simultaneously compared with suspect's picture of establishment, filters out human face similarity degree more than the face picture of first threshold and candid photograph point information and preserve；Screening in point, which is captured, from preservation trailed for first suspect's time not less than trailing time threshold personnel；The trailing personnel filtered out are grouped by human face similarity degree, a pictures are selected from every group as personnel's retrieving image is trailed, face picture and corresponding candid photograph point information that human face similarity degree in all pictures for capturing point is more than first threshold is searched out；The candid photograph point information for trailing personnel in each packet is captured into point information with the screening of corresponding trailing personnel retrieving image to be compared, the candid photograph point information of omission is found out, judges whether trailing personnel are that suspect works in partnership by trailing ratio.Also include a corresponding device.The present invention can reduce the artificial investigation amount of criminal detective, and reduction is manually judged by accident and failed to judge.
Description
Technical field
The application is related to the present invention relates to big data searching field, more particularly to suspect and its trails personnel
Motion trail analysis method and device.
Background technology
Usual offender is clique's crime, and perhaps criminal investigation department only knows several in clique personnel
Individual suspect, therefore want criminal gang catching all in one draft more difficult.For example, working as criminal gang
During crime, generally all share out the work and help one another, and criminal detective is being found that several suspects have deployed to arrest,
Other cliques personnel will necessarily be startled, it is not easy to arrest criminal gang personnel totally.
In view of the situation, criminal detective utilizes Video Supervision Technique, to the trailing personnel of known suspect
Manually investigated, filter out other cliques personnel.But such mode, criminal detective need processing
Data volume it is big, and easily cause and manually judge by accident and fail to judge, can not equally ensure that criminal gang is total
Arrest.
The content of the invention
In view of this, the application provides a kind of method and device of trailing personnel analysis.This method and correspondence dress
The artificial investigation amount of criminal detective that can reduce is put, and reduction is manually judged by accident and failed to judge.
Specifically, the application is achieved by the following technical solution：
A kind of method of trailing personnel analysis, captures what point was captured by each in setting range and setting time
Face picture is compared with the suspect's picture established, and human face similarity degree is more than to the face picture of first threshold
And corresponding point information of capturing saves as first suspect's information；
First suspect's time of trailing is filtered out from the candid photograph point in first suspect's information not less than tail
Trailing is saved as with the personnel of time threshold, and by the trailing personnel filtered out and correspondence candid photograph point information
Personal information；
The trailing personnel trailed in personal information are grouped by human face similarity degree, each face in being often grouped is calculated
Picture quality score, using quality score highest face picture in every group as personnel's retrieving image is trailed, is searched
It is all in rope setting range and setting time to capture the face picture that point is captured, filter out human face similarity degree
More than the face picture and corresponding candid photograph point information of first threshold；
The candid photograph point information of personnel will be trailed in each packet with passing through corresponding trailing personnel retrieving image
Screening is captured point information and is compared, and finds out the candid photograph point information of omission；
After the candid photograph point information of omission is found out, check whether the candid photograph point of omission is captured to the first suspect,
Update first suspect's information；
Statistics candid photograph simultaneously in setting range is accounted for the candid photograph point of the first suspect and trailing personnel always grabs
The ratio in point is clapped, when ratio is more than given threshold, judges that trailing personnel work in partnership for suspect.
It is preferred that, first suspect's information is combined on map and is trailed point information is captured in personal information,
The first suspect is drawn by the time of candid photograph and trails the movement locus of personnel.
It is preferred that, the trailing personnel trailed in personal information are grouped by human face similarity degree, concretely comprised the following steps,
The face picture trailed in personal information is chosen, is carried out with trailing remaining picture in information table
Human face similarity degree compares, if human face similarity degree is more than Second Threshold, is placed in same packet.
It is preferred that, as two capture point collaboration capture described in face picture, specifically include, grab when first
Clap point to capture to after face picture, judge whether the face picture captured meets standard faces picture condition,
Check that the second candid photograph point captures information if being unsatisfactory for, judge whether second capture a face picture captured
Meet standard faces picture condition, if the face picture that the second candid photograph point is captured is eligible, preserving should
Face picture and corresponding candid photograph point information, if the face picture that the second candid photograph point is captured is ineligible,
Then selection preserves the face picture and corresponding candid photograph point information closer to standard faces picture condition in two.
It is preferred that, it is 30 ° -60 ° to set first to capture point with the second angled scope of camera lens institute for capturing point.
It is preferred that, described face picture is captured simultaneously by multiple cameras, filters out the face figure captured
Piece meets standard faces picture condition, and closest to the face picture of standard faces picture, wherein, it is described
Multiple camera shooting angles covering each angle of candid photograph personnel.
A kind of trailing personnel analytical equipment, the device includes：
First suspect's information generating unit, is arrived for each camera in setting range and setting time to be captured
Face picture with establish suspect's picture compared, by human face similarity degree be more than first threshold face figure
Piece and corresponding candid photograph point information save as first suspect's information；
Personal information generation unit is trailed, for being sieved from the candid photograph point of first suspect's information generating unit
The personnel that the first suspect is not above trailing time threshold the time that trail are selected, and by the trailing filtered out
Personnel and correspondence capture point information and save as trailing personal information；
Trailing personnel, which omit, captures point storage unit, for the trailing personnel that will be trailed in personal information
It is grouped by human face similarity degree, each face picture quality score in every group is calculated, with quality highest in every packet
Face picture as trail personnel's retrieving image, search setting range and setting time in it is all capture
The picture that point is captured, filters out face picture of the human face similarity degree more than first threshold and corresponding candid photograph
Point information；The candid photograph point information that personnel are trailed in each packet is schemed with being retrieved by corresponding trailing personnel
Piece screening is captured point information and is compared, and finds out the candid photograph point information of omission；
First suspect's information updating unit, for after the candid photograph point information of omission is found out, checking omission
Candid photograph point whether capture to the first suspect, update the information in first suspect's generation unit；
Suspect works in partnership judging unit, for count in setting range and meanwhile capture to the first suspect with
The candid photograph point of trailing personnel accounts for total ratio captured in point, when ratio is more than given threshold, judges the tail
Work in partnership with personnel for suspect.
It is preferred that, trace generator unit, for combining first suspect's information and tail on map
With point information is captured in personal information, draw the first suspect by the time of candid photograph and trail the motion of personnel
Track.
It is preferred that, the trailing personnel trailed in personal information are grouped by human face similarity degree, concretely comprised the following steps,
Choose and trail a face picture in personal information, pedestrian is entered with trailing remaining picture in information table
Face similarity degree compares, if human face similarity degree is more than Second Threshold, is placed in same packet.
It is preferred that, as two capture point collaboration capture described in face picture, specifically include, grab when first
Clap point to capture to after face picture, judge whether the face picture captured meets standard faces picture condition,
Check that the second candid photograph point captures information if being unsatisfactory for, judge whether second capture a face picture captured
Meet standard faces picture condition, if the face picture that the second candid photograph point is captured is eligible, preserving should
Face picture and corresponding candid photograph point information, if the face picture that the second candid photograph point is captured is ineligible,
Then selection preserves the face picture and corresponding candid photograph point information closer to standard faces picture condition in two.
Compared to prior art, the present invention program finds trailing personal information by first suspect's information,
And corresponding trailing face picture is searched in setting range using human face similarity degree, find out omission and capture point,
And by analyze trailing personnel and the first suspect the identical quantity for capturing point of process judge that trailing personnel are
No is suspect partner.
Brief description of the drawings
Fig. 1 is the implementation process figure of the application；
Fig. 2 is the first suspect of the application exemplary embodiment and trails personnel motion trail figure；
Fig. 3 is two or more schematic diagram of a scenario for capturing point candid photograph in the application exemplary embodiment；
Embodiment
Here exemplary embodiment will be illustrated in detail, its example is illustrated in the accompanying drawings.Following retouches
State when being related to accompanying drawing, unless otherwise indicated, the same numbers in different accompanying drawings represent same or analogous key element.
Embodiment described in following exemplary embodiment does not represent all embodiment party consistent with the application
Formula.On the contrary, they be only be described in detail in such as appended claims, the application some in terms of phase one
The example of the apparatus and method of cause.
It is the purpose only merely for description specific embodiment in term used in this application, and is not intended to be limiting this
Application." one kind ", " described " of singulative used in the application and appended claims
It is also intended to "the" including most forms, unless context clearly shows that other implications.It is also understood that
Term "and/or" used herein refer to and comprising it is one or more it is associated list any of project or
It is possible to combination.
It will be appreciated that though various letters may be described using term first, second, third, etc. in the application
Breath, but these information should not necessarily be limited by these terms.These terms are only used for same type of information area each other
Separate.For example, in the case where not departing from the application scope, the first information can also be referred to as the second information,
Similarly, the second information can also be referred to as the first information.Depending on linguistic context, word as used in this
" if " can be construed to " ... when " or " when ... when " or " in response to determine ".
To solve the technical problem mentioned in background technology, the present invention provides a kind of side of trailing personnel analysis
Method.With reference to Fig. 1, Fig. 2 and Fig. 3, the implementation of the present invention is illustrated with an embodiment.
First, criminal detective utilizes the picture of the suspect established to setting time and setting model
Interior all monitoring datas for capturing point are enclosed to be retrieved.It should be noted that " setting range " here
The scope of investigation that " setting time " Main Basiss criminal detective delimit when deploying the investigation of crime case
And the investigation period.For example, occurring in several days of crime case to occurring this section of crime case
Carry out investigations.Here candid photograph point is commonly referred to as the candid photograph camera on road, such as bayonet socket camera.
By searching for the face pictures that all candid photograph points are captured in setting time section, it will scheme with the suspect of establishment
Piece gets off compared to the picture and corresponding candid photograph point information record that human face similarity degree is more than first threshold.It is logical
It can often preserve in table 1 as first suspect's information.
As shown in figure 3, having 1-12 candid photograph camera in the present embodiment in the range of the candid photograph of setting, search
The rope time is 2016.1.170：00 arrives 2016.1.1712:00, it is assumed that first threshold is 80%, then can be with
Obtain table 1 as follows：
| Capture point | The candid photograph time | Similarity |
| 1 | 2016.1.1702:15 | 85% |
| 2 | 2016.1.1702:30 | 87% |
| 9 | 2016.1.1703:00 | 89% |
| 10 | 2016.1.1703:15 | 87% |
| 11 | 2016.1.1705:45 | 90% |
| 12 | 2016.1.1706:15 | 85% |
| 7 | 2016.1.1706:30 | 83% |
| 8 | 2016.1.1707:00 | 87% |
Table 1
For convenience of the behavior for analyzing the first suspect, map can be combined by the first suspect according to the time of candid photograph
The position for passing through candid photograph point is marked on map, the movement locus of the first suspect is formed
L1：1-2-9-10-11-12-7-8.
The time that the first suspect of trailing is filtered out in the candid photograph point contained from first suspect's packet does not surpass
The personnel for trailing time threshold are crossed, and by the trailing face picture filtered out and corresponding candid photograph point information
It is stored in table 2 as trailing personal information.
Assuming that in the present embodiment, it is 30 minutes to trail time threshold.Therefore table 2 is exemplified below：
| Picture | Capture point | The candid photograph time |
| Picture 1 | 1 | 2016.1.1702:20 |
| Picture 2 | 2 | 2016.1.1702:35 |
| Picture 3 | 2 | 2016.1.1702:37 |
| Picture 4 | 9 | 2016.1.1703:14 |
| Picture 5 | 9 | 2016.1.1703:15 |
| Picture 6 | 10 | 2016.1.1703:23 |
| Picture 7 | 10 | 2016.1.1703:25 |
| Picture 8 | 11 | 2016.1.1706:00 |
| Picture 9 | 12 | 2016.1.1706:20 |
| Picture 10 | 7 | 2016.1.1706:40 |
| Picture 11 | 7 | 2016.1.1706:45 |
| Picture 12 | 8 | 2016.1.1707:05 |
| Picture 13 | 8 | 2016.1.1707:10 |
Table 2
It is grouped to trailing personal information in table 2 by human face similarity degree.For example, choosing in table 2
Trail face picture for first to be compared with other pictures in table, will be greater than the trailing personnel of Second Threshold
Information is selected as same packet.By that analogy, all trailing personnel in table 2 are carried out more afterwards one by one
Personnel are grouped.The trailing number of times of trailing personnel can be counted from packet.By by the trailing personnel of component cloth
Information is stored in table 3.Record has the corresponding human face similarity degree of every pictures in each packet.
Assuming that Second Threshold is 85%, then table 3 is grouped as follows：
| Picture | Capture point | The candid photograph time | Human face similarity degree | Packet |
| Picture 1 | 1 | 2016.1.1702:20 | 1 | |
| Picture 2 | 2 | 2016.1.1702:35 | 89% | |
| Picture 11 | 7 | 2016.1.1706:45 | 90% | |
| Picture 12 | 8 | 2016.1.1707:05 | 88% | |
| Picture 3 | 2 | 2016.1.1702:37 | 2 | |
| Picture 5 | 9 | 2016.1.1703:15 | 87% | |
| Picture 7 | 10 | 2016.1.1703:25 | 90% | |
| Picture 8 | 11 | 2016.1.1706:00 | 91% | |
| … | … | … | … | 3 |
| … | … | … | … | … |
Check trailing personnel motion trail for convenience herein, map can be combined and trail people by each in table 3
Member draws out the movement locus L2 of each trailing personnel by its candid photograph point candid photograph time on map.
Citing, packet 1:1-2-7-8, packet 2：2-9-10-11.
Certainly, because the data trailed in personal information table 2 are to be screened according to time interval,
There can be certain erroneous judgement.For example, there is a part only has stretch journey identical road with the first suspect
People, and this erroneous judgement can all have very big influence to the input and efficiency of solving a case of manpower of solving a case.Therefore
Need to delete this part erroneous judgement data.
Herein, applicant calculates face picture in each packet to improve the accuracy of retrieval result
Quality score, the parameter mainly considered includes：Face size, image definition, image intensities.When
Other specification can also be so chosen, for example, image sharpness, image conformity, face bilateral symmetry etc..
Quality score highest picture is picked out from each packet as personnel's retrieving image is trailed, in setting model
The face picture captured with all candid photograph points of search in setting time section is enclosed, will be with trailing personnel's retrieving image
Compared to the face picture and corresponding candid photograph point information record that human face similarity degree is more than first threshold to table 4
In.Still it is grouped and preserves according to each retrieval result in table 4.
| Picture | Capture point | The candid photograph time | Human face similarity degree | Packet |
| Picture 11 | 7 | 2016.1.1706:45 | 90% | 1 |
| Picture 1 | 1 | 2016.1.1702:20 | ||
| Picture 2 | 2 | 2016.1.1702:35 | 89% | |
| Picture 14 | 3 | 2016.1.1703:30 | 88% | |
| Picture 15 | 4 | 2016.1.1704:10 | 90% | |
| Picture 16 | 5 | 2016.1.1705:20 | 86% | |
| Picture 17 | 6 | 2016.1.1705:35 | 91% | |
| Picture 12 | 8 | 2016.1.1707:05 | 88% | |
| Picture 8 | 11 | 2016.1.1706:00 | 91% | 2 |
| Picture 3 | 2 | 2016.1.1702:53 | ||
| Picture 5 | 9 | 2016.1.1703:20 | 87% | |
| Picture 7 | 10 | 2016.1.1703:55 | 90% | |
| Picture 18 | 3 | 2016.1.1704:30 | 85% | |
| Picture 19 | 6 | 2016.1.1705:45 | 90% | |
| … | … | … | … | 3 |
| … | … | … | … | … |
Table 4
Similarly, for ease of the behavior of analysis trailing personnel, map can be combined by 4 every groups of retrieval results of table
Movement locus L3 is drawn according to the candid photograph time for capturing point.
For example, packet 1:1-2-3-4-5-6-7-8, packet 2:2-9-10-3-6-11.
For the candid photograph point of rejecting, it can also mark out and on map, be easy to criminal detective to be grabbed for these
Clap the analysis that point is removed reason.
Due to being scanned for based on trailing personnel's retrieving image in setting range and setting time section, it is searched for
Scope is greater than the trailing demographic data directly selected by table 1.By the candid photograph that table 4 is grouped to personnel
The candid photograph point of point packet personnel corresponding with table 2 is compared, and finds out the candid photograph point of the omission of table 2.In this reality
Applying in example to find out, and it is 3,4,5,6 that point is captured in the omission of packet 1, and point is captured in the omission of packet 2
For 3,6.Then judge whether the candid photograph point omitted has first suspect's picture, if so, grabbing omission
The pictorial information clapped point and captured is added to table 1, updates the first suspect movement locus L1, and check
Candid photograph point active situation of first suspect in omission.Find to capture in packet 1 just like in this example, checking
Point 3 and 6 has the first suspect, therefore updates table 1, and obtains the motion after the first suspect updates
Track L1 '：1-2-9-10-3-6-11-12-7-8.
Can be all as comprehensive as possible by the movement locus of the first suspect and trailing personnel by above-mentioned steps
Analyze come.But have as previously described, in trailing personnel may some be erroneous judgement personnel, need
Delete.There is provided a kind of preferred embodiment to solve this problem as the present invention.Calculate the first suspect and
The candid photograph point quantity that trailing personnel pass through jointly and total ratio for capturing point quantity, judge whether accounting exceedes
Given threshold, when more than given threshold, it is believed that this kind of trailing personnel are partners.Assuming that given threshold is 40%,
The trailing personnel that then may determine that packet 1 in the present embodiment are partners.The present invention reduces trailing personnel prison
Scope is controlled, is conducive to improving criminal detective's case handling efficiency.Certainly filtering out partner personnel can be compared by calculating
Also the realization of the other modes such as number of times can be trailed by counting beyond example, will not be repeated here.
If the first suspect is in the case where capturing the intentional side face of point in such scheme, by that can not obtain
Normal face picture, therefore accurate result of calculation can not relatively be obtained by human face similarity degree.For
This problem is solved, it is The present invention gives two kinds preferred embodiment, specific to combine Fig. 3 explanations.One
The mode of kind is that face picture is captured by two candid photograph point collaboration candid photographs.Specifically, when first captures point A
Capture after face picture, judge whether the face picture captured meets standard faces picture condition, if
It is unsatisfactory for, checks that the second candid photograph point B captures information, judges that the face picture that the second candid photograph point B is captured is
It is no to meet standard faces picture condition, if the face picture that the second candid photograph point B is captured is eligible, protect
The face picture and corresponding candid photograph point information are deposited, if the face picture that the second candid photograph point B is captured is not met
Condition, then selection preserves the face picture in two closer to standard faces picture condition and corresponding candid photograph
Point information.Herein, standard faces picture condition refers to that eyes spacing is more than 30 pixels, face level
Rotational angle [- 15 °, 15 °], face pitch rotation angle [- 10 °, 10 °].The candid photograph mode captures effect
The mode that fruit more only captures a pictures is accurate, and it is also few to consume performance.To ensure to capture face
Accuracy and definition, set first camera with second camera camera lens institute angled scope at 30 ° -60 °
Within it is ideal.
As another preferred embodiment, described face picture by multiple candid photograph points (such as in Fig. 3,
Tetra- candid photograph points of A, B, C, D) capture simultaneously, filter out the face picture captured and meet standard people
Face picture condition, and closest to the face picture of standard faces picture, wherein, described multiple cameras are clapped
Take the photograph angle covering each angle of candid photograph personnel.
Embodiment with the method that foregoing trailing personnel analyze is corresponding, and present invention also provides the personnel of trailing point
The embodiment of the device of analysis.
The embodiment that the application trails the device of personnel's analysis can be applied such as platform in rear end equipment.
Device embodiment can be realized by software, can also be realized by way of hardware or software and hardware combining.
It is the processing by rear end equipment where it as the device on a logical meaning exemplified by implemented in software
Corresponding computer program instructions in nonvolatile memory are read what operation in internal memory was formed by device.Implement
Rear end equipment in example where device can also include other hardware generally according to the actual functional capability of the device,
This is repeated no more.
The device is specifically included, first suspect's information generating unit, by setting range and setting time
The face picture that each camera is captured is compared with the suspect's picture established, and human face similarity degree is more than into first
The face picture of threshold value and corresponding candid photograph point information save as first suspect's information；
Personal information generation unit is trailed, for being sieved from the candid photograph point of first suspect's information generating unit
The personnel that the first suspect is not above trailing time threshold the time that trail are selected, and by the trailing filtered out
Personnel and correspondence capture point information and save as trailing personal information；
Trailing personnel, which omit, captures point storage unit, for the trailing personnel that will be trailed in personal information
It is grouped by human face similarity degree, calculates each face picture quality score in every group, with quality score in every group most
High face picture searches for all in setting range and setting time grab as personnel's retrieving image is trailed
The picture that point is captured is clapped, human face similarity degree is filtered out more than the face picture of first threshold and corresponding grabs
Clap point information；The candid photograph point information of personnel will be trailed in each packet to retrieve with by corresponding trailing personnel
Picture screening is captured point information and is compared, and finds out the candid photograph point information of omission；
First suspect's information updating unit, for after the candid photograph point information of omission is found out, checking omission
Candid photograph point whether capture to the first suspect, update the information in first suspect's generation unit；
Suspect works in partnership judging unit, for count in setting range and meanwhile capture to the first suspect with
The candid photograph point of trailing personnel accounts for total ratio captured in point, when ratio is more than given threshold, judges the tail
Work in partnership with personnel for suspect.
Trace generator unit, for combining first suspect's information on map and trailing personnel's letter
Point information is captured in breath, the first suspect is drawn by the time of candid photograph and trails the movement locus of personnel.
Wherein the trailing personnel trailed in personal information are grouped by human face similarity degree, concretely comprised the following steps,
Choose and trail a face picture in personal information, pedestrian is entered with trailing remaining picture in information table
Face similarity degree compares, if human face similarity degree is more than Second Threshold, is placed in same packet.
As the face picture described in two are captured point collaboration candid photograph, specifically include, when the first candid photograph point is captured
To after face picture, judge whether the face picture captured meets standard faces picture condition, if discontented
It is sufficient then check the second candid photograph point capture information, judge second candid photograph point candid photograph a face picture whether completely meet
Standard faces picture condition, if the face picture that the second candid photograph point is captured is eligible, preserves the face
Picture and corresponding candid photograph point information, if the face picture that the second candid photograph point is captured is ineligible, are selected
Select the face picture and corresponding candid photograph point information for preserving closer standard faces picture condition in two.
The present invention passes through face recognition technology search the by suspect's picture of establishment in setting range
One suspect's information, finds trailing personnel using the corresponding point of capturing of the first suspect, is trailed by analyzing
Personal information, finds omission and captures point, improves the first suspect and trails the movement locus of personnel.Pass through
The coincidence ratio for analyzing the first suspect and trailing personnel candid photograph point judges whether trailing personnel are suspect group
Group.Degree of intelligence is high, the problem of efficiently solving trailing personnel erroneous judgement and fail to judge.
The function of unit and the implementation process of effect specifically refer to correspondence step in the above method in said apparatus
Rapid implementation process, will not be repeated here.
For device embodiment, because it corresponds essentially to embodiment of the method, so related part referring to
The part explanation of embodiment of the method.Device embodiment described above is only schematical, wherein
The unit illustrated as separating component can be or may not be physically separate, be used as unit
The part of display can be or may not be physical location, you can with positioned at a place, or also may be used
To be distributed on multiple NEs.Some or all of module therein can be selected according to the actual needs
To realize the purpose of application scheme.Those of ordinary skill in the art without creative efforts,
It is appreciated that and implements.
The preferred embodiment of the application is the foregoing is only, it is all in this Shen not to limit the application
Within spirit and principle please, any modification, equivalent substitution and improvements done etc. should be included in this Shen
Within the scope of please protecting.
Claims (10)
1. a kind of method of trailing personnel analysis, it is characterised in that
The face picture that point is captured and the suspect established are captured by each in setting range and setting time
Picture compares, and face picture and corresponding candid photograph point information that human face similarity degree is more than first threshold are protected
Save as first suspect's information；
First suspect's time of trailing is filtered out from the candid photograph point in first suspect's information not less than tail
Trailing is saved as with the personnel of time threshold, and by the trailing personnel filtered out and correspondence candid photograph point information
Personal information；
The trailing personnel trailed in personal information are grouped by human face similarity degree, each face in being often grouped is calculated
Picture quality score, using quality score highest face picture in every group as personnel's retrieving image is trailed, is searched
It is all in rope setting range and setting time to capture the picture that point is captured, filter out human face similarity degree and be more than
The face picture of first threshold and corresponding candid photograph point information；
The candid photograph point information of personnel will be trailed in each packet with passing through corresponding trailing personnel retrieving image
Screening is captured point information and is compared, and finds out the candid photograph point information of omission；
After the candid photograph point information of omission is found out, check whether the candid photograph point of omission is captured to the first suspect,
Update first suspect's information；
Statistics candid photograph simultaneously in setting range is accounted for the candid photograph point of the first suspect and trailing personnel always grabs
The ratio in point is clapped, when ratio is more than given threshold, judges that trailing personnel work in partnership for suspect.
2. the method for trailing personnel analysis according to claim 1, it is characterised in that in map
Upper combination the first suspect information and trail point information is captured in personal information, the is drawn by the time of candid photograph
One suspect and the movement locus for trailing personnel.
3. the method for trailing personnel analysis according to claim 1, it is characterised in that will trail
Trailing personnel in personal information are grouped by human face similarity degree, are concretely comprised the following steps,
The face picture trailed in personal information is chosen, is carried out with trailing remaining picture in information table
Human face similarity degree compares, if human face similarity degree is more than Second Threshold, is placed in same packet.
4. the method for trailing personnel analysis according to claim 1, it is characterised in that by two
The face picture described in point collaboration candid photograph is captured, is specifically included, is captured when the first candid photograph point and arrives the face
After picture, judge whether the face picture captured meets standard faces picture condition, looked into if being unsatisfactory for
See that the second candid photograph point captures information, judge whether the face picture that the second candid photograph point is captured meets standard faces
Picture condition, if the face picture that the second candid photograph point is captured is eligible, preserves the face picture and right
The candid photograph point information answered, if the face picture that the second candid photograph point is captured is ineligible, selection preserves two
Closer to the face picture and corresponding candid photograph point information of standard faces picture condition in.
5. the method for trailing personnel according to claim 4 analysis, it is characterised in that set the
The one angled scope of camera lens institute for capturing point and the second candid photograph point is 30 ° -60 °.
6. the method for trailing personnel analysis according to claim 1, it is characterised in that by multiple
Camera captures described face picture simultaneously, filters out the face picture captured and meets standard faces picture
Condition, and closest to the face picture of standard faces picture, wherein, described multiple camera shooting angles
Cover each angle of candid photograph personnel.
7. a kind of trailing personnel analytical equipment, it is characterised in that the device includes,
First suspect's information generating unit, is arrived for each camera in setting range and setting time to be captured
Face picture with establish suspect's picture compared, by human face similarity degree be more than first threshold face figure
Piece and corresponding candid photograph point information save as first suspect's information；
Personal information generation unit is trailed, for being sieved from the candid photograph point of first suspect's information generating unit
The personnel that the first suspect is not above trailing time threshold the time that trail are selected, and by the trailing filtered out
Personnel and correspondence capture point information and save as trailing personal information；
Trailing personnel, which omit, captures point storage unit, for the trailing personnel that will be trailed in personal information
It is grouped by human face similarity degree, calculates each face picture quality score in every group, with quality score in every group most
High face picture searches for all in setting range and setting time grab as personnel's retrieving image is trailed
The picture that point is captured is clapped, human face similarity degree is filtered out more than the face picture of first threshold and corresponding grabs
Clap point information；The candid photograph point information of personnel will be trailed in each packet to retrieve with by corresponding trailing personnel
Picture screening is captured point information and is compared, and finds out the candid photograph point information of omission；
First suspect's information updating unit, for after the candid photograph point information of omission is found out, checking omission
Candid photograph point whether capture to the first suspect, update the information in first suspect's generation unit；
Suspect works in partnership judging unit, for count in setting range and meanwhile capture to the first suspect with
The candid photograph point of trailing personnel accounts for total ratio captured in point, when ratio is more than given threshold, judges the tail
Work in partnership with personnel for suspect.
8. the device of trailing personnel analysis according to claim 7, it is characterised in that motion rail
Mark generation unit, is captured a little for combining first suspect's information on map and trailing in personal information
Information, draws the first suspect by the time of candid photograph and trails the movement locus of personnel.
9. the device of trailing personnel analysis according to claim 7, it is characterised in that will trail
Trailing personnel in personal information are grouped by human face similarity degree, are concretely comprised the following steps,
Choose and trail a face picture in personal information, pedestrian is entered with trailing remaining picture in information table
Face similarity degree compares, if human face similarity degree is more than Second Threshold, is placed in same packet.
10. the device of trailing personnel analysis according to claim 7, it is characterised in that
As the face picture described in two are captured point collaboration candid photograph, specifically include, when the first candid photograph point is captured
To after face picture, judge whether the face picture captured meets standard faces picture condition, if discontented
It is sufficient then check the second candid photograph point capture information, judge second candid photograph point candid photograph a face picture whether meet mark
Quasi- face picture condition, if the face picture that the second candid photograph point is captured is eligible, preserves the face figure
Piece and corresponding candid photograph point information, if the face picture that the second candid photograph point is captured is ineligible, are selected
Preserve face picture and corresponding candid photograph point information closer to standard faces picture condition in two.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610058847.4A CN107016322B (en) | 2016-01-28 | 2016-01-28 | Method and device for analyzing followed person |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610058847.4A CN107016322B (en) | 2016-01-28 | 2016-01-28 | Method and device for analyzing followed person |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107016322A true CN107016322A (en) | 2017-08-04 |
| CN107016322B CN107016322B (en) | 2020-01-14 |
Family
ID=59438993
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610058847.4A Active CN107016322B (en) | 2016-01-28 | 2016-01-28 | Method and device for analyzing followed person |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107016322B (en) |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108229335A (en) * | 2017-12-12 | 2018-06-29 | 深圳市商汤科技有限公司 | It is associated with face identification method and device, electronic equipment, storage medium, program |
| CN109117714A (en) * | 2018-06-27 | 2019-01-01 | 北京旷视科技有限公司 | A kind of colleague's personal identification method, apparatus, system and computer storage medium |
| CN109165637A (en) * | 2018-10-08 | 2019-01-08 | 武汉爱迪科技股份有限公司 | A kind of personal identification method and system based on dynamic video analysis |
| CN109784274A (en) * | 2018-12-29 | 2019-05-21 | 杭州励飞软件技术有限公司 | Identify the method trailed and Related product |
| CN109784199A (en) * | 2018-12-21 | 2019-05-21 | 深圳云天励飞技术有限公司 | Analysis method of going together and Related product |
| CN109858329A (en) * | 2018-12-15 | 2019-06-07 | 深圳壹账通智能科技有限公司 | Anti- trailing method, apparatus, equipment and storage medium based on recognition of face |
| CN110019963A (en) * | 2017-12-11 | 2019-07-16 | 罗普特（厦门）科技集团有限公司 | The searching method of suspect relationship personnel |
| WO2019218824A1 (en) * | 2018-05-15 | 2019-11-21 | 腾讯科技（深圳）有限公司 | Method for acquiring motion track and device thereof, storage medium, and terminal |
| CN110569720A (en) * | 2019-07-31 | 2019-12-13 | 安徽四创电子股份有限公司 | audio and video intelligent identification processing method based on audio and video processing system |
| CN111104915A (en) * | 2019-12-23 | 2020-05-05 | 云粒智慧科技有限公司 | Method, device, equipment and medium for peer analysis |
| CN111382627A (en) * | 2018-12-28 | 2020-07-07 | 成都云天励飞技术有限公司 | Method for judging peer and related products |
| CN111563479A (en) * | 2020-05-26 | 2020-08-21 | 深圳市商汤科技有限公司 | Same-pedestrian duplicate removal method, group analysis method, device and electronic equipment |
| CN111639689A (en) * | 2020-05-20 | 2020-09-08 | 杭州海康威视***技术有限公司 | Face data processing method and device and computer readable storage medium |
| CN112149627A (en) * | 2020-10-19 | 2020-12-29 | 杭州海康威视数字技术股份有限公司 | Method and device for identifying fellow persons, electronic equipment and storage medium |
| WO2020259099A1 (en) * | 2019-06-28 | 2020-12-30 | 深圳市商汤科技有限公司 | Information processing method and device, and storage medium |
| CN113452903A (en) * | 2021-06-17 | 2021-09-28 | 浙江大华技术股份有限公司 | Snapshot equipment, snapshot method and main control chip |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102158595A (en) * | 2011-02-16 | 2011-08-17 | 中兴通讯股份有限公司 | Method and device for realizing burglary prevention of mobile terminal by face recognition |
| CN103456175A (en) * | 2013-09-25 | 2013-12-18 | 武汉烽火众智数字技术有限责任公司 | Accompanying vehicle real-time detection method based on vehicle registration plate recognition and meshing monitoring |
| CN103871248A (en) * | 2014-03-18 | 2014-06-18 | 浙江宇视科技有限公司 | Method and device for analyzing vehicles tailing after suspected vehicle based on track collision |
| US8837805B1 (en) * | 2014-03-11 | 2014-09-16 | Rafael Aviyants | System and method for verification of a banknote |
| CN104616494A (en) * | 2014-12-23 | 2015-05-13 | 浙江宇视科技有限公司 | Method and device for recording and determining target object based on base station and block port |
-
2016
- 2016-01-28 CN CN201610058847.4A patent/CN107016322B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102158595A (en) * | 2011-02-16 | 2011-08-17 | 中兴通讯股份有限公司 | Method and device for realizing burglary prevention of mobile terminal by face recognition |
| CN103456175A (en) * | 2013-09-25 | 2013-12-18 | 武汉烽火众智数字技术有限责任公司 | Accompanying vehicle real-time detection method based on vehicle registration plate recognition and meshing monitoring |
| US8837805B1 (en) * | 2014-03-11 | 2014-09-16 | Rafael Aviyants | System and method for verification of a banknote |
| CN103871248A (en) * | 2014-03-18 | 2014-06-18 | 浙江宇视科技有限公司 | Method and device for analyzing vehicles tailing after suspected vehicle based on track collision |
| CN104616494A (en) * | 2014-12-23 | 2015-05-13 | 浙江宇视科技有限公司 | Method and device for recording and determining target object based on base station and block port |
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110019963A (en) * | 2017-12-11 | 2019-07-16 | 罗普特（厦门）科技集团有限公司 | The searching method of suspect relationship personnel |
| CN108229335A (en) * | 2017-12-12 | 2018-06-29 | 深圳市商汤科技有限公司 | It is associated with face identification method and device, electronic equipment, storage medium, program |
| WO2019218824A1 (en) * | 2018-05-15 | 2019-11-21 | 腾讯科技（深圳）有限公司 | Method for acquiring motion track and device thereof, storage medium, and terminal |
| CN109117714A (en) * | 2018-06-27 | 2019-01-01 | 北京旷视科技有限公司 | A kind of colleague's personal identification method, apparatus, system and computer storage medium |
| CN109165637A (en) * | 2018-10-08 | 2019-01-08 | 武汉爱迪科技股份有限公司 | A kind of personal identification method and system based on dynamic video analysis |
| CN109858329A (en) * | 2018-12-15 | 2019-06-07 | 深圳壹账通智能科技有限公司 | Anti- trailing method, apparatus, equipment and storage medium based on recognition of face |
| CN109784199B (en) * | 2018-12-21 | 2020-11-24 | 深圳云天励飞技术有限公司 | Peer-to-peer analysis method and related product |
| CN109784199A (en) * | 2018-12-21 | 2019-05-21 | 深圳云天励飞技术有限公司 | Analysis method of going together and Related product |
| CN111382627A (en) * | 2018-12-28 | 2020-07-07 | 成都云天励飞技术有限公司 | Method for judging peer and related products |
| CN111382627B (en) * | 2018-12-28 | 2024-03-26 | 成都云天励飞技术有限公司 | Method for judging peer and related products |
| CN109784274A (en) * | 2018-12-29 | 2019-05-21 | 杭州励飞软件技术有限公司 | Identify the method trailed and Related product |
| CN109784274B (en) * | 2018-12-29 | 2021-09-14 | 杭州励飞软件技术有限公司 | Method for identifying trailing and related product |
| WO2020259099A1 (en) * | 2019-06-28 | 2020-12-30 | 深圳市商汤科技有限公司 | Information processing method and device, and storage medium |
| CN110569720B (en) * | 2019-07-31 | 2022-06-07 | 安徽四创电子股份有限公司 | Audio and video intelligent identification processing method based on audio and video processing system |
| CN110569720A (en) * | 2019-07-31 | 2019-12-13 | 安徽四创电子股份有限公司 | audio and video intelligent identification processing method based on audio and video processing system |
| CN111104915A (en) * | 2019-12-23 | 2020-05-05 | 云粒智慧科技有限公司 | Method, device, equipment and medium for peer analysis |
| CN111104915B (en) * | 2019-12-23 | 2023-05-16 | 云粒智慧科技有限公司 | Method, device, equipment and medium for peer analysis |
| CN111639689A (en) * | 2020-05-20 | 2020-09-08 | 杭州海康威视***技术有限公司 | Face data processing method and device and computer readable storage medium |
| CN111639689B (en) * | 2020-05-20 | 2023-07-25 | 杭州海康威视***技术有限公司 | Face data processing method and device and computer readable storage medium |
| CN111563479A (en) * | 2020-05-26 | 2020-08-21 | 深圳市商汤科技有限公司 | Same-pedestrian duplicate removal method, group analysis method, device and electronic equipment |
| CN111563479B (en) * | 2020-05-26 | 2023-11-03 | 深圳市商汤科技有限公司 | Concurrent person weight removing method, partner analyzing method and device and electronic equipment |
| CN112149627A (en) * | 2020-10-19 | 2020-12-29 | 杭州海康威视数字技术股份有限公司 | Method and device for identifying fellow persons, electronic equipment and storage medium |
| CN113452903A (en) * | 2021-06-17 | 2021-09-28 | 浙江大华技术股份有限公司 | Snapshot equipment, snapshot method and main control chip |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107016322B (en) | 2020-01-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107016322A (en) | A kind of method and device of trailing personnel analysis | |
| US8644563B2 (en) | Recognition of faces using prior behavior | |
| CN110263613A (en) | Monitor video processing method and processing device | |
| CN106295618A (en) | A kind of personal identification method and device based on video image | |
| RU2634225C1 (en) | Methods and systems for searching object in video stream | |
| US8842889B1 (en) | System and method for automatic face recognition | |
| US10262209B2 (en) | Method for analyzing video data | |
| CN111898581A (en) | Animal detection method, device, electronic equipment and readable storage medium | |
| DE112012002579T5 (en) | Capture and edit content using biometric data | |
| CN110442742A (en) | Retrieve method and device, processor, electronic equipment and the storage medium of image | |
| CN110781853A (en) | Crowd abnormality detection method and related device | |
| CN111832434B (en) | Campus smoking behavior recognition method under privacy protection and processing terminal | |
| CN109086829A (en) | A kind of method and device that social population administers | |
| CN108230669A (en) | The illegal detection method of road vehicle and system based on big data and cloud analysis | |
| CN109756467A (en) | A kind of recognition methods of fishing website and device | |
| CN112949439A (en) | Method and system for monitoring invasion of personnel in key area of oil tank truck | |
| CN107426132A (en) | The detection method and device of network attack | |
| Garcia et al. | Acquisition of digital images and identification of Aedes Aegypti mosquito eggs using classification and deep learning | |
| JP2022100358A (en) | Search method and device in search support system | |
| CN106570466B (en) | Video classification method and system | |
| CN113190700B (en) | Face snapshot, screening and storage method and system for real estate transaction | |
| Gurrin et al. | An examination of a large visual lifelog | |
| US9805272B1 (en) | Storage system of original frame of monitor data and storage method thereof | |
| CN106777215A (en) | A kind of image searching result display methods and device | |
| KR101683232B1 (en) | System for processing image and supplying related information by using total information of image |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |