CN109815878A - Foothold analysis method and device based on recognition of face - Google Patents
Foothold analysis method and device based on recognition of face Download PDFInfo
- Publication number
- CN109815878A CN109815878A CN201910049340.6A CN201910049340A CN109815878A CN 109815878 A CN109815878 A CN 109815878A CN 201910049340 A CN201910049340 A CN 201910049340A CN 109815878 A CN109815878 A CN 109815878A
- Authority
- CN
- China
- Prior art keywords
- information
- unit
- condition code
- video camera
- passerby
- 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
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
- Closed-Circuit Television Systems (AREA)
Abstract
The present invention provides a kind of using video based on the ground automatic identification of stopping over of recognition of face, analysis identification method.Process is marked by passerby, feature identification is carried out to the owner taken in the specific region for being provided with video camera and distributes virtual identity building passerby library, then, extraction simplification first is carried out to initial data when stopping over point analysis, and then quality screening, virtual identity label is rapidly completed, it is merged into analysis complete table with initial data again again, the point identification of stopping over of each virtual identity is finally realized in the differentiation of then bond area entrance, time and frequency of occurrence.
Description
Technical field
The present invention relates to a kind of data analysing method, refer in particular to a kind of foothold analysis method based on recognition of face and
Device.
Background technique
With Chinese society expanding economy, all kinds of economic activities are increased, and passerby's flowing is more and more subtler, tool
The crowd for having counter-reconnaissance consciousness is even more that can not track.
And it is right, at present everywhere by video monitoring, has the function of passing crowd's face when capture mostly, pass through combination
Intelligent image identification, the structural data after extraction is stored in big data storage system according to time and attribute, so that borrowing
Thus part big data becomes possible to personnel's point analysis of stopping over.
Summary of the invention
The technical problems to be solved by the present invention are: provide it is a kind of establish video monitoring identify based on recognition of face
Foothold analysis method and device.
In order to solve the above-mentioned technical problem, a kind of the technical solution adopted by the present invention are as follows: foothold based on recognition of face
Analysis method, including initialization process, passerby mark process and foothold analysis process;
The initialization process includes the steps that video camera is arranged and video camera information will be answered to charge to camera record table;
The video camera information includes video camera serial number and video camera setting address;
It includes being acquired to passerby that the passerby, which marks process, after then the face characteristic of parsing passerby obtains condition code
The step of after further identification/distribution virtual identity by condition code and virtual identity deposit passerby library;
The foothold analysis process includes step,
B1 original candid photograph record) is obtained;
Described original at least one face picture pair captured in the video flowing that record includes the shooting of at least one video camera
The record information answered;
The record information includes video camera serial number, video camera installation direction attribute, candid photograph serial number, quality point, time
Stamp and condition code;
B2 original capture) is extracted to record the condition code of information in record, capture serial number and mass fractal into comparison information;
B3) comparison information by quality in comparison information point less than setting value abandons;
B4) judge whether condition code has similarity to be greater than threshold value in condition code and passerby library in remaining comparison information,
It is to then follow the steps B5, thens follow the steps B7 if not;
B5 the condition code of maximum similarity) is taken;
B6) virtual identity of condition code in passerby library is marked and gives this feature code, goes to step B9;
B7 a virtual identity) is created;
B8) newly-built virtual identity is marked and gives this feature code, goes to step B9;
B9 the virtual identity for capturing serial number and label of comparison information) is extracted as filter information, and all screenings are believed
Breath generates PID_UID table;
It B10) will be corresponding with the candid photograph serial number of filter information in PID_UID table in the original record information for capturing record
Video camera serial number, video camera installation direction attribute and timestamp fill into filter information in contrast information, finally obtain control
Table;
B11) filter out that virtual identity is identical in table of comparisons control information and according to video camera installation direction determined property is
The control information of entrance;
B12 daily contrasting data further) is screened in control information with each virtual identity, and extracts contrasting data
Timestamp be a control information the latest;
It B13 is) statistics number+1 of video camera serial number in the control information;
It B14 is more than) to set the control information temporary storage of statistical threshold as candidate item by the statistics number of video camera serial number;
B15 the control information of candidate item) is pressed into statistics number Bit-reversed, the control for the preceding setting number for taking number most
Information is defined as the virtual identity of control information according to the corresponding video camera setting address of video camera serial number of control information
Foothold.
Among the above, it includes step that the passerby, which marks process,
A1 at least one face picture in the video flowing of at least one video camera shooting) is obtained;
A2 structuring processing) is carried out to face picture, obtains condition code;
A3) whether condition code has similarity to be greater than threshold value in judging characteristic code and passerby library, be then follow the steps A4, if
It is no to then follow the steps A6;
A4 the condition code of maximum similarity) is taken
A5) virtual identity of condition code in passerby library is marked and gives this feature code, then return step A1;
A6 a virtual identity) is created;
A7) virtual identity is marked and gives this feature code;
A8 it) will be marked in the condition code deposit passerby library of virtual identity, then return to step A1.
Among the above, it includes step that the passerby, which marks process,
A1 at least one face picture and corresponding photographing information in the video flowing of at least one video camera shooting) are obtained;
The photographing information includes video camera serial number, video camera installation direction attribute, candid photograph serial number, quality point and time
Stamp;
A2 structuring processing) is carried out to face picture, obtains condition code, condition code correspondence is added in photographing information;
A3) photographing information by quality in photographing information point less than setting value abandons;
A4) judge that whether condition code has similarity to be greater than threshold value in the condition code and passerby library of photographing information, then executes
Step A5, thens follow the steps A8 if not;
A5 the photographing information of the condition code of maximum similarity) is taken;
A6) judge photographing information whether more than one, be to go to step A7, otherwise go to step A10;
A7 newest one of timestamp in photographing information) is taken, step A8 is gone to;
A8) virtual identity of condition code in passerby library is added in the corresponding photographing information of this feature code, goes to step A9;
A9 the record information correspondence in passerby library) is updated to the photographing information, then return step A1;
A10 a virtual identity) is created;
A11) virtual identity is added in the corresponding photographing information of this feature code, goes to step A12;
A12) photographing information is newly inserted in passerby library, then returns to step A1.
It among the above, further include step between step A8, A9 of passerby's label process,
A891) current virtual identity and condition code are kept in;
A892) judge whether to reach setting time, otherwise return step A1, be to continue step A893;
A893) judge whether similarity is greater than threshold value for the condition code of the virtual identity of caching, be that then nearly retention time stamp is most
A new photographing information goes to step A9, otherwise goes to step A8.
The present invention also provides a kind of foothold analytical equipment based on recognition of face, including initialization module, passerby's label
Module and foothold analysis module;
The initialization module, for video camera to be arranged and video camera information will be answered to charge to camera record table;It is described to take the photograph
Camera information includes video camera serial number and video camera setting address;
Passerby's mark module, for being acquired to passerby, the face characteristic for then parsing passerby obtains condition code
Condition code and virtual identity are stored in passerby library after further identification/distribution virtual identity afterwards;
The foothold analysis module includes,
Original acquiring unit, for obtaining original candid photograph record;The original candid photograph record includes at least one video camera
The corresponding record information of at least one face picture in the video flowing of shooting;The record information includes video camera serial number, takes the photograph
Camera installation direction attribute, candid photograph serial number, quality point, timestamp and condition code;
Condition code extracting unit, for extracting the condition code, candid photograph serial number and the matter that record information in original candid photograph record
Amount point forms comparison information;
Quality screening unit is abandoned for the comparison information by quality in comparison information point less than setting value;
Whether condition code comparing unit has phase with condition code in passerby library for judging in remaining comparison information condition code
It is greater than threshold value like degree, is to go to most like screening unit, if otherwise going to newly-built unit;
Most like screening unit, for taking the condition code of maximum similarity;
Marking unit gives this feature code for marking the virtual identity of condition code in passerby library, goes to generation unit;
Newly-built unit, for creating a virtual identity;
Newly-built marking unit, gives this feature code for marking the virtual identity created, goes to generation unit;
Generation unit, the virtual identity for capturing serial number and label for extracting comparison information are incited somebody to action as filter information
All filter informations generate PID_UID table;
Generation unit is compareed, for will grab in the original record information for capturing record with filter information in PID_UID table
The corresponding video camera serial number of serial number, video camera installation direction attribute and timestamp is clapped to fill into filter information and in contrast believe
Breath, finally obtains the table of comparisons;
Direction screening unit is identical and according to video camera installation side for filtering out virtual identity in table of comparisons control information
It is the control information entered to determined property;
Screening unit the latest, for further screening daily contrasting data in control information with each virtual identity,
And the timestamp for extracting contrasting data is a control information the latest;
Accumulated unit is counted, for the statistics number+1 for video camera serial number in the control information;
Temporary storage location, for being more than to set the control information temporary storage of statistical threshold to wait by the statistics number of video camera serial number
Option;
Foothold marking unit takes number most for the control information of candidate item to be pressed statistics number Bit-reversed
The control information of preceding setting number, is defined as pair according to the corresponding video camera setting address of video camera serial number of control information
According to the foothold of the virtual identity of information.
Among the above, passerby's mark module specifically includes,
Face acquiring unit, at least one face picture in video flowing for obtaining the shooting of at least one video camera;
Structuring processing unit obtains condition code for carrying out structuring processing to face picture;
Whether similarity judging unit has similarity to be greater than threshold value for condition code in judging characteristic code and passerby library,
It is to go to similar screening unit, if otherwise going to identity creates unit;
Similar screening unit, for taking the condition code of maximum similarity
Updating unit gives this feature code for marking the virtual identity of condition code in passerby library, then returns to face and obtains
Take unit;
Identity creates unit, for creating a virtual identity;
Identity marks unit gives this feature code for marking the virtual identity;
Passerby library updating unit, the condition code for that will be marked with virtual identity are stored in passerby library, then return to face
Acquiring unit.
Among the above, passerby's mark module includes,
Face acquiring unit, for obtain at least one video camera shooting video flowing at least one face picture and
Corresponding photographing information;The photographing information includes video camera serial number, video camera installation direction attribute, captures serial number, quality point
And timestamp;
Structuring processing unit obtains condition code, condition code is corresponded to and is added for carrying out structuring processing to face picture
Into photographing information;
Discarding unit is abandoned for the photographing information by quality in photographing information point less than setting value;
Similarity judging unit, for judging whether the condition code of photographing information has similarity big with condition code in passerby library
It in threshold value, is to go to similar screening unit, if otherwise going to identity creates unit;
Similar screening unit, the photographing information of the condition code for taking maximum similarity;
Quantity judging unit, for judge photographing information whether more than one, be to go to newest acquiring unit, otherwise turn
Unit is created to identity;
Newest acquiring unit goes to updating unit for taking newest one of timestamp in photographing information;
Updating unit, for the virtual identity of condition code in passerby library to be added in the corresponding photographing information of this feature code,
Go to passerby library updating unit;
Passerby library updating unit is then returned for the record information correspondence in passerby library to be updated to the photographing information
Face acquiring unit;
Identity creates unit, for creating a virtual identity;
Identity marks unit goes to passerby library for the virtual identity to be added in the corresponding photographing information of this feature code
Updating unit;
Passerby library updating unit then returns to face acquiring unit for newly inserting photographing information in passerby library.
Further, the newly-built unit of the identity of passerby's mark module passes through identity cache unit, time judging unit
And condition code duplicate removal unit goes to identity marks unit,
Identity cache unit, for keeping in current virtual identity and condition code;
Time judging unit reaches setting time for judging whether, otherwise goes to face acquiring unit, be, go to spy
Levy code duplicate removal unit;
Condition code duplicate removal unit, it is then that whether similarity is greater than threshold value to the condition code of the virtual identity for judging to cache
Nearly retention time stabs a newest photographing information and goes to passerby library updating unit, otherwise goes to identity and creates unit.
The beneficial effects of the present invention are: provide it is a kind of using video based on recognition of face stop over ground automatic identification,
Analyze identification method.Process is marked by passerby, the owner taken in the specific region for being provided with video camera is carried out special
Sign, which identifies and distributes virtual identity building passerby library, then first carries out extraction simplification to initial data when stopping over point analysis, into
And quality screening, virtual identity label is rapidly completed, then analysis complete table is merged into initial data again, then combined area
The differentiation of domain entrance, time and frequency of occurrence, finally realizes the point identification of stopping over of each virtual identity.
Detailed description of the invention
Specific structure of the invention is described in detail with reference to the accompanying drawing
Fig. 1 is foothold analysis flow chart diagram of the invention;
Fig. 2 is that the passerby of one embodiment of the invention marks process.
Specific embodiment
In order to describe the technical content, the structural feature, the achieved object and the effect of this invention in detail, below in conjunction with embodiment
And attached drawing is cooperated to be explained in detail.
This patent provides a kind of foothold analysis method based on recognition of face, includingInitialization process、Passerby marks stream JourneyAndFoothold analysis process;
It is describedInitialization process,
Include the steps that video camera is arranged and video camera information will be answered to charge to camera record table;The video camera information packet
Include video camera serial number and video camera setting address;
It is describedPasserby marks process,
Including being acquired to passerby, the face characteristic for then parsing passerby obtains further identification/distribution after condition code
The step of after virtual identity by condition code and virtual identity (UUID) deposit passerby library;
It is describedFoothold analysis process, referring to Fig. 1, including step,
B1 original candid photograph record) is obtained;
Described original at least one face picture pair captured in the video flowing that record includes the shooting of at least one video camera
The record information answered;
The record information includes video camera serial number, video camera installation direction attribute, candid photograph serial number, quality point, time
Stamp and condition code;
Example:
Original candid photograph record citing table
B2 original capture) is extracted to record the condition code of information in record, capture serial number and mass fractal into comparison information;
Since original record can be comprising a large amount of record information, and every record information includes at least above-named 6
Field, if therefore directly being screened using original record very big to the expense of hardware.Therefore, by original candid photograph in this step
Each record information is subsequently used for screening condition code in record, quality sub-sieve is elected, together with the candid photograph serial number for index
Comparison information is formed, deck watch is further obtained, to significantly simplify the data volume of screening, improves operational efficiency.
Example: compareing the original candid photograph record citing table of distance above, and the comparison information table after magically change is as follows
Comparison information | Condition code | Capture serial number | Quality point |
Comparison information 1 | adfebcaecaecer | 8477123 | 75 |
Comparison information 2 | adfebcaefaecer | 8477135 | 58 |
Comparison information 3 | adfebcazcaecec | 8478942 | 68 |
Comparison information 4 | adfebcaeccecer | 8479150 | 69 |
Comparison information 5 | adfebcwecaecez | 8487133 | 45 |
Comparison information 6 | adfebcaecaeder | 8489033 | 55 |
Comparison information 7 | adfebcaecaeces | 8497423 | 62 |
B3) comparison information by quality in comparison information point less than setting value abandons;
Example: the original candid photograph record citing table of distance above is compareed, after this step (usually by the low rejecting of quality point
With 60 points for boundary)
B4) judge whether condition code has similarity to be greater than threshold value in condition code and passerby library in remaining comparison information,
It is to then follow the steps B5, thens follow the steps B7 if not;
B5 the condition code of maximum similarity) is taken;
The record that the condition code of the comparison information of B4 judgement is greater than threshold value with condition code in passerby library may have a plurality of, this step
Suddenly it is exactly to correspond in this case, chooses a condition code the most similar and carry out subsequent operation.
B6) virtual identity of condition code in passerby library is marked and gives this feature code, goes to step B9;
It is similar with height in passerby library due to occurring condition code in comparison information, by condition code in passerby library
Corresponding virtual identity is assigned to this new condition code.
B7 a virtual identity) is created;
B8) newly-built virtual identity is marked and gives this feature code, goes to step B9;
B9 the virtual identity for capturing serial number and label of comparison information) is extracted as filter information, and all screenings are believed
Breath generates PID_UID table;
So far step, quality screening and virtual identity judgement are completed, therefore further simplify to comparison information,
Close to retain the virtual identity (UID) for capturing serial number (PID) corresponding label, forming one, only there are two the PID_ of project
UID table.
Example: compareing the original candid photograph record citing table of distance above, and the PID_UID table to this step only retains two projects
It B10) will be corresponding with the candid photograph serial number of filter information in PID_UID table in the original record information for capturing record
Video camera serial number, video camera installation direction attribute and timestamp fill into filter information in contrast information, finally obtain control
Table;
Factually show fast mass screening by simplified deck watch's logarithm, distribute virtual body with passerby library comparing
Part, the record information respective items recorded at this time according to PID_UID table with the original candid photograph of acquisition merge (by capturing serial number
Look for corresponding record information), so that obtaining final foothold judges the table of comparisons that need to be used.
Example: compareing the original candid photograph record citing table of distance above, and the table of comparisons to this step is
B11) filter out that virtual identity is identical in table of comparisons control information and according to video camera installation direction determined property is
The control information of entrance;
It according to video camera installation direction is exported against specific region due to containing video camera installation direction in the table of comparisons
Or entrance situation, it is known that its face recognized is into/out specific region.Therefore this step will enter specific region
Carry out retain, these records represent people therein and have come in specific region, according to virtual identity by the record of the same person
Choosing is further deleted, to carry out the differentiation of subsequent stop.
B12 daily contrasting data further) is screened in control information with each virtual identity, and extracts contrasting data
Timestamp be a control information the latest;
The people of corresponding each identical virtual identity, it is available to a record the latest according to timestamp.
It B13 is) statistics number+1 of video camera serial number in the control information;
It B14 is more than) to set the control information temporary storage of statistical threshold as candidate item by the statistics number of video camera serial number;
B15 the control information of candidate item) is pressed into statistics number Bit-reversed, the control for the preceding setting number for taking number most
Information is defined as the virtual identity of control information according to the corresponding video camera setting address of video camera serial number of control information
Foothold.
The beneficial effects of the present invention are: provide it is a kind of using video based on recognition of face stop over ground automatic identification,
Analyze identification method.Process is marked by passerby, the owner taken in the specific region for being provided with video camera is carried out special
Sign, which identifies and distributes virtual identity building passerby library, then first carries out extraction simplification to initial data when stopping over point analysis, into
And quality screening, virtual identity label is rapidly completed, then analysis complete table is merged into initial data again, then combined area
The differentiation of domain entrance, time and frequency of occurrence, finally realizes the point identification of stopping over of each virtual identity.
Embodiment 1
Further, the passerby marks process to may particularly include step,
A1 at least one face picture in the video flowing of at least one video camera shooting) is obtained;
A2 structuring processing) is carried out to face picture, obtains condition code;
A3) whether condition code has similarity to be greater than threshold value in judging characteristic code and passerby library, be then follow the steps A4, if
It is no to then follow the steps A6;
A4 the condition code of maximum similarity) is taken
A5) virtual identity of condition code in passerby library is marked and gives this feature code, then return step A1;
A6 a virtual identity) is created;
A7) virtual identity is marked and gives this feature code;
A8 it) will be marked in the condition code deposit passerby library of virtual identity, then return to step A1.
The foundation in passerby library mainly depends on the face picture of front end acquisition, obtains condition code after handling by structuring,
To carry out the classification of virtual identity according to the similarity of condition code.Further, by obtaining condition code and passerby library to new
Have the comparison of record, implements update mark and newly-built preservation respectively, it is ensured that one " virtual identity " only has one in passerby library
A condition code.
Embodiment 2
Further, referring to fig. 2, it includes step that the passerby, which marks process,
A1 at least one face picture and corresponding photographing information in the video flowing of at least one video camera shooting) are obtained;
The photographing information includes video camera serial number, video camera installation direction attribute, candid photograph serial number, quality point and time
Stamp;
Example: the following are schematical photographing information tables
A2 structuring processing) is carried out to face picture, obtains condition code, condition code correspondence is added in photographing information;
A3) photographing information by quality in photographing information point less than setting value abandons;
A4) judge that whether condition code has similarity to be greater than threshold value in the condition code and passerby library of photographing information, then executes
Step A5, thens follow the steps A8 if not;
A5 the photographing information of the condition code of maximum similarity) is taken;
A6) judge photographing information whether more than one, be to go to step A7, otherwise go to step A10;
A7 newest one of timestamp in photographing information) is taken;
A8) virtual identity of condition code in passerby library is added in the corresponding photographing information of this feature code, goes to step A9;
A9 the record information correspondence in passerby library) is updated to the photographing information, then return step A1;
A10 a virtual identity) is created;
A11) virtual identity is added in the corresponding photographing information of this feature code, goes to step A12;
A12) photographing information is newly inserted in passerby library, then returns to step A1.
Embodiment 3
It further include step between step A8, A9 that above-mentioned passerby marks process,
A891) current virtual identity and condition code are kept in;
A892) judge whether to reach setting time, otherwise return step A1, be to continue step A893;
A893) judge whether similarity is greater than threshold value for the condition code of the virtual identity of caching, be that then nearly retention time stamp is most
A new photographing information goes to step A9, otherwise goes to step A8.
The method of re-scheduling when present embodiments providing newly-built virtual identity.Since multithreading can be used in efficient application
Condition code comparison is carried out, therefore multiple threads may be will appear to be that the same newly-increased face is judged to correspond to newly-built to practical
Virtual identity, the case where so as to cause multiple records are established to a people.It, will be (logical in setting time in order to avoid such case
Often can be set as one day), newly-built virtual identity is all first kept in, i.e., is not stored in passerby library, then carries out similarity judgement
Re-scheduling further retains newest mode re-scheduling according to timestamp when necessary.
The present invention also provides a kind of foothold analytical equipment based on recognition of face, including initialization module, passerby's label
Module and foothold analysis module;
The initialization module, for video camera to be arranged and video camera information will be answered to charge to camera record table;It is described to take the photograph
Camera information includes video camera serial number and video camera setting address;
Passerby's mark module, for being acquired to passerby, the face characteristic for then parsing passerby obtains condition code
Condition code and virtual identity are stored in passerby library after further identification/distribution virtual identity afterwards;
The foothold analysis module includes,
Original acquiring unit, for obtaining original candid photograph record;The original candid photograph record includes at least one video camera
The corresponding record information of at least one face picture in the video flowing of shooting;The record information includes video camera serial number, takes the photograph
Camera installation direction attribute, candid photograph serial number, quality point, timestamp and condition code;
Condition code extracting unit, for extracting the condition code, candid photograph serial number and the matter that record information in original candid photograph record
Amount point forms comparison information;
Quality screening unit is abandoned for the comparison information by quality in comparison information point less than setting value;
Whether condition code comparing unit has phase with condition code in passerby library for judging in remaining comparison information condition code
It is greater than threshold value like degree, is to go to most like screening unit, if otherwise going to newly-built unit;
Most like screening unit, for taking the condition code of maximum similarity;
Marking unit gives this feature code for marking the virtual identity of condition code in passerby library, goes to generation unit;
Newly-built unit, for creating a virtual identity;
Newly-built marking unit, gives this feature code for marking the virtual identity created, goes to generation unit;
Generation unit, the virtual identity for capturing serial number and label for extracting comparison information are incited somebody to action as filter information
All filter informations generate PID_UID table;
Generation unit is compareed, for will grab in the original record information for capturing record with filter information in PID_UID table
The corresponding video camera serial number of serial number, video camera installation direction attribute and timestamp is clapped to fill into filter information and in contrast believe
Breath, finally obtains the table of comparisons;
Direction screening unit is identical and according to video camera installation side for filtering out virtual identity in table of comparisons control information
It is the control information entered to determined property;
Screening unit the latest, for further screening daily contrasting data in control information with each virtual identity,
And the timestamp for extracting contrasting data is a control information the latest;
Accumulated unit is counted, for the statistics number+1 for video camera serial number in the control information;
Temporary storage location, for being more than to set the control information temporary storage of statistical threshold to wait by the statistics number of video camera serial number
Option;
Foothold marking unit takes number most for the control information of candidate item to be pressed statistics number Bit-reversed
The control information of preceding setting number, is defined as pair according to the corresponding video camera setting address of video camera serial number of control information
According to the foothold of the virtual identity of information.
The beneficial effects of the present invention are: provide it is a kind of using video based on recognition of face stop over ground automatic identification,
Analyze identity device.Process is marked by passerby, the owner taken in the specific region for being provided with video camera is carried out special
Sign, which identifies and distributes virtual identity building passerby library, then first carries out extraction simplification to initial data when stopping over point analysis, into
And quality screening, virtual identity label is rapidly completed, then analysis complete table is merged into initial data again, then combined area
The differentiation of domain entrance, time and frequency of occurrence, finally realizes the point identification of stopping over of each virtual identity.
Embodiment 4
Among the above, passerby's mark module specifically includes,
Face acquiring unit, at least one face picture in video flowing for obtaining the shooting of at least one video camera;
Structuring processing unit obtains condition code for carrying out structuring processing to face picture;
Whether similarity judging unit has similarity to be greater than threshold value for condition code in judging characteristic code and passerby library,
It is to go to similar screening unit, if otherwise going to identity creates unit;
Similar screening unit, for taking the condition code of maximum similarity
Updating unit gives this feature code for marking the virtual identity of condition code in passerby library, then returns to face and obtains
Take unit;
Identity creates unit, for creating a virtual identity;
Identity marks unit gives this feature code for marking the virtual identity;
Passerby library updating unit, the condition code for that will be marked with virtual identity are stored in passerby library, then return to face
Acquiring unit.
Embodiment 5
Among the above, passerby's mark module includes,
Face acquiring unit, for obtain at least one video camera shooting video flowing at least one face picture and
Corresponding photographing information;The photographing information includes video camera serial number, video camera installation direction attribute, captures serial number, quality point
And timestamp;
Structuring processing unit obtains condition code, condition code is corresponded to and is added for carrying out structuring processing to face picture
Into photographing information;
Discarding unit is abandoned for the photographing information by quality in photographing information point less than setting value;
Similarity judging unit, for judging whether the condition code of photographing information has similarity big with condition code in passerby library
It in threshold value, is to go to similar screening unit, if otherwise going to identity creates unit;
Similar screening unit, the photographing information of the condition code for taking maximum similarity;
Quantity judging unit, for judge photographing information whether more than one, be to go to newest acquiring unit, otherwise turn
Unit is created to identity;
Newest acquiring unit goes to updating unit for taking newest one of timestamp in photographing information;
Updating unit, for the virtual identity of condition code in passerby library to be added in the corresponding photographing information of this feature code,
Go to passerby library updating unit;
Passerby library updating unit is then returned for the record information correspondence in passerby library to be updated to the photographing information
Face acquiring unit;
Identity creates unit, for creating a virtual identity;
Identity marks unit goes to passerby library for the virtual identity to be added in the corresponding photographing information of this feature code
Updating unit;
Passerby library updating unit then returns to face acquiring unit for newly inserting photographing information in passerby library.
Embodiment 6
Further, on the basis of embodiment 5, the identity of passerby's mark module creates unit and passes through identity cache list
Member, time judging unit and condition code duplicate removal unit go to identity marks unit,
Identity cache unit, for keeping in current virtual identity and condition code;
Time judging unit reaches setting time for judging whether, otherwise goes to face acquiring unit, be, go to spy
Levy code duplicate removal unit;
Condition code duplicate removal unit, it is then that whether similarity is greater than threshold value to the condition code of the virtual identity for judging to cache
Nearly retention time stabs a newest photographing information and goes to passerby library updating unit, otherwise goes to identity and creates unit.
The art of this patent, by carrying out intellectual analysis by video of the AI artificial intelligence deep learning algorithm to video camera,
When someone passes through in picture, real-time grasp shoot is carried out to face, and be abstracted to face picture, extracts face characteristic code letter
Breath stores face original image and condition code information picture servers and distributed database server cluster to system
In.The movable passerby library information in certain administrative region is preserved in system, face picture including entire population and from picture
In the condition code information that extracts.The target face picture of candid photograph is subjected to condition code extraction by big data platform, then
Whole condition codes progress similarity in the condition code of target face and population library is compared and (calculates every picture and target person
The distance of face), similarity is greater than certain threshold value (0.83) and quality and the condition code higher than 60 is divided to be classified as the same person's
Condition code, the picture lower than quality point 60 all discard, are not involved in analysis, take the highest population library of the similarity of the condition of satisfaction
In the face unique identification of the UUID as target face, finally based on the UUID and period to capture in library retrieval with
The people that target face occurs simultaneously, then according to the direction of video camera, (positioning passerby is gone out from cell, is still entered from the outside
Cell), the picture that latest time in every day enters cell from the outside is extracted, the Target Photo of foothold is defined, finally counts
In nearest 30 days time cycle, the corresponding address where appearance 5 times or more target video camera ID (the same UUID) is pressed
TOP3 is taken after the most sequence of number, foundation is to take out three entered from the outside cell frequency of occurrence at most from more to less to take the photograph
Camera is defined as the foothold of the UUID.
The above description is only an embodiment of the present invention, is not intended to limit the scope of the invention, all to utilize this hair
Equivalent structure or equivalent flow shift made by bright specification and accompanying drawing content is applied directly or indirectly in other relevant skills
Art field, is included within the scope of the present invention.
Claims (8)
1. a kind of foothold analysis method based on recognition of face, it is characterised in that: mark process including initialization process, passerby
And foothold analysis process;
The initialization process includes the steps that video camera is arranged and video camera information will be answered to charge to camera record table;It is described
Video camera information includes video camera serial number and video camera setting address;
It includes being acquired to passerby that the passerby, which marks process, and the face characteristic for then parsing passerby obtains condition code laggard one
The step of after step identification/distribution virtual identity by condition code and virtual identity deposit passerby library;
The foothold analysis process includes step,
B1 original candid photograph record) is obtained;
Described original at least one face picture captured in the video flowing that record includes the shooting of at least one video camera is corresponding
Record information;
The record information include video camera serial number, video camera installation direction attribute, capture serial number, quality point, timestamp and
Condition code;
B2 original capture) is extracted to record the condition code of information in record, capture serial number and mass fractal into comparison information;
B3) comparison information by quality in comparison information point less than setting value abandons;
B4) judge whether condition code has similarity to be greater than threshold value in condition code and passerby library in remaining comparison information, be then
Step B5 is executed, thens follow the steps B7 if not;
B5 the condition code of maximum similarity) is taken;
B6) virtual identity of condition code in passerby library is marked and gives this feature code, goes to step B9;
B7 a virtual identity) is created;
B8) newly-built virtual identity is marked and gives this feature code, goes to step B9;
B9 the virtual identity for capturing serial number and label of comparison information) is extracted as filter information, by the life of all filter informations
At PID_UID table;
B10) by camera shooting corresponding with the candid photograph serial number of filter information in PID_UID table in the original record information for capturing record
Machine serial number, video camera installation direction attribute and timestamp fill into filter information in contrast information, finally obtain the table of comparisons;
B11) filter out the table of comparisons control information in virtual identity it is identical and according to video camera installation direction determined property for enter
Control information;
B12) further with each virtual identity control information in screen daily contrasting data, and extract contrasting data when
Between stamp be a control information the latest;
It B13 is) statistics number+1 of video camera serial number in the control information;
It B14 is more than) to set the control information temporary storage of statistical threshold as candidate item by the statistics number of video camera serial number;
B15 the control information of candidate item) is pressed into statistics number Bit-reversed, the control letter for the preceding setting number for taking number most
Breath is defined as the virtual identity of control information according to the corresponding video camera setting address of video camera serial number of control information
Foothold.
2. the foothold analysis method based on recognition of face as described in claim 1, it is characterised in that: the passerby marks stream
Journey includes step,
A1 at least one face picture in the video flowing of at least one video camera shooting) is obtained;
A2 structuring processing) is carried out to face picture, obtains condition code;
A3) whether judging characteristic code has similarity to be greater than threshold value with condition code in passerby library, is to then follow the steps A4, if otherwise
Execute step A6;
A4 the condition code of maximum similarity) is taken
A5) virtual identity of condition code in passerby library is marked and gives this feature code, then return step A1;
A6 a virtual identity) is created;
A7) virtual identity is marked and gives this feature code;
A8 it) will be marked in the condition code deposit passerby library of virtual identity, then return to step A1.
3. the foothold analysis method based on recognition of face as described in claim 1, it is characterised in that: the passerby marks stream
Journey includes step,
A1 at least one face picture and corresponding photographing information in the video flowing of at least one video camera shooting) are obtained;
The photographing information includes video camera serial number, video camera installation direction attribute, candid photograph serial number, quality point and timestamp;
A2 structuring processing) is carried out to face picture, obtains condition code, condition code correspondence is added in photographing information;
A3) photographing information by quality in photographing information point less than setting value abandons;
A4) judge that whether condition code has similarity to be greater than threshold value in the condition code and passerby library of photographing information, thens follow the steps
A5 thens follow the steps A8 if not;
A5 the photographing information of the condition code of maximum similarity) is taken;
A6) judge photographing information whether more than one, be to go to step A7, otherwise go to step A10;
A7 newest one of timestamp in photographing information) is taken, step A8 is gone to;
A8) virtual identity of condition code in passerby library is added in the corresponding photographing information of this feature code, goes to step A9;
A9 the record information correspondence in passerby library) is updated to the photographing information, then return step A1;
A10 a virtual identity) is created;
A11) virtual identity is added in the corresponding photographing information of this feature code, goes to step A12;
A12) photographing information is newly inserted in passerby library, then returns to step A1.
4. the foothold analysis method based on recognition of face as claimed in claim 3, it is characterised in that: the passerby marks stream
It further include step between step A8, A9 of journey,
A891) current virtual identity and condition code are kept in;
A892) judge whether to reach setting time, otherwise return step A1, be to continue step A893;
A893) judge whether similarity is greater than threshold value for the condition code of the virtual identity of caching, be that then nearly retention time stamp is newest
One photographing information goes to step A9, otherwise goes to step A8.
5. a kind of foothold analytical equipment based on recognition of face, it is characterised in that: including initialization module, passerby's mark module
And foothold analysis module;
The initialization module, for video camera to be arranged and video camera information will be answered to charge to camera record table;The video camera
Information includes video camera serial number and video camera setting address;
Passerby's mark module, for being acquired to passerby, then it is laggard to obtain condition code for the face characteristic of parsing passerby
One step, which is identified/distributed, is stored in passerby library for condition code and virtual identity after virtual identity;
The foothold analysis module includes,
Original acquiring unit, for obtaining original candid photograph record;The original candid photograph record includes that at least one video camera is shot
Video flowing in the corresponding record information of at least one face picture;The record information includes video camera serial number, video camera
Installation direction attribute, candid photograph serial number, quality point, timestamp and condition code;
Condition code extracting unit divides for extracting the original condition code for recording information in record, candid photograph serial number and the quality of capturing
Form comparison information;
Quality screening unit is abandoned for the comparison information by quality in comparison information point less than setting value;
Condition code comparing unit, for judging whether condition code has similarity in condition code and passerby library in remaining comparison information
It greater than threshold value, is to go to most like screening unit, if otherwise going to newly-built unit;
Most like screening unit, for taking the condition code of maximum similarity;
Marking unit gives this feature code for marking the virtual identity of condition code in passerby library, goes to generation unit;
Newly-built unit, for creating a virtual identity;
Newly-built marking unit, gives this feature code for marking the virtual identity created, goes to generation unit;
Generation unit, the virtual identity for capturing serial number and label for extracting comparison information will own as filter information
Filter information generates PID_UID table;
Generation unit is compareed, the candid photograph stream in the record information for recording original candid photograph with filter information in PID_UID table
The corresponding video camera serial number of water number, video camera installation direction attribute and timestamp fill into filter information in contrast information, most
The table of comparisons is obtained eventually;
Direction screening unit is identical and according to video camera installation direction category for filtering out virtual identity in table of comparisons control information
Property is judged as the control information of entrance;
Screening unit the latest for further screening daily contrasting data in control information with each virtual identity, and mentions
Take the timestamp of contrasting data for a control information the latest;
Accumulated unit is counted, for the statistics number+1 for video camera serial number in the control information;
Temporary storage location, for being more than to set the control information temporary storage of statistical threshold as candidate by the statistics number of video camera serial number
?;
Foothold marking unit is set before taking number most for the control information of candidate item to be pressed statistics number Bit-reversed
The control information for determining number is defined as control letter according to the corresponding video camera setting address of video camera serial number of control information
The foothold of the virtual identity of breath.
6. the foothold analytical equipment based on recognition of face as claimed in claim 5, it is characterised in that: the passerby marks mould
Block specifically includes,
Face acquiring unit, at least one face picture in video flowing for obtaining the shooting of at least one video camera;
Structuring processing unit obtains condition code for carrying out structuring processing to face picture;
Whether similarity judging unit has similarity to be greater than threshold value, is then for condition code in judging characteristic code and passerby library
Similar screening unit is gone to, if otherwise going to identity creates unit;
Similar screening unit, for taking the condition code of maximum similarity
Updating unit gives this feature code for marking the virtual identity of condition code in passerby library, then returns to face and obtains list
Member;
Identity creates unit, for creating a virtual identity;
Identity marks unit gives this feature code for marking the virtual identity;
Passerby library updating unit, the condition code for that will be marked with virtual identity are stored in passerby library, then return to face acquisition
Unit.
7. the foothold analytical equipment based on recognition of face as claimed in claim 5, it is characterised in that: the passerby marks mould
Block includes,
Face acquiring unit, at least one face picture and correspondence in video flowing for obtaining the shooting of at least one video camera
Photographing information;The photographing information includes video camera serial number, video camera installation direction attribute, to capture serial number, quality point timely
Between stab;
Structuring processing unit obtains condition code, condition code correspondence is added to bat for carrying out structuring processing to face picture
It takes the photograph in information;
Discarding unit is abandoned for the photographing information by quality in photographing information point less than setting value;
Similarity judging unit, for judging whether the condition code of photographing information has similarity to be greater than threshold with condition code in passerby library
Value, be to go to similar screening unit, if otherwise going to identity creates unit;
Similar screening unit, the photographing information of the condition code for taking maximum similarity;
Quantity judging unit, for judge photographing information whether more than one, be to go to newest acquiring unit, otherwise go to body
The newly-built unit of part;
Newest acquiring unit goes to updating unit for taking newest one of timestamp in photographing information;
Updating unit is gone to for the virtual identity of condition code in passerby library to be added in the corresponding photographing information of this feature code
Passerby library updating unit;
Passerby library updating unit then returns to face for the record information correspondence in passerby library to be updated to the photographing information
Acquiring unit;
Identity creates unit, for creating a virtual identity;
Identity marks unit goes to the update of passerby library for the virtual identity to be added in the corresponding photographing information of this feature code
Unit;
Passerby library updating unit then returns to face acquiring unit for newly inserting photographing information in passerby library.
8. the foothold analytical equipment based on recognition of face as claimed in claim 7, it is characterised in that: the passerby marks mould
The identity of block creates unit and goes to identity marks list by identity cache unit, time judging unit and condition code duplicate removal unit
Member,
Identity cache unit, for keeping in current virtual identity and condition code;
Time judging unit reaches setting time for judging whether, otherwise goes to face acquiring unit, be, go to condition code
Duplicate removal unit;
Condition code duplicate removal unit, it is closely to protect that whether similarity is greater than threshold value to the condition code of the virtual identity for judging to cache
It stays the newest photographing information of timestamp to go to passerby library updating unit, otherwise goes to identity and create unit.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910049340.6A CN109815878B (en) | 2019-01-18 | 2019-01-18 | Foothold analysis method and device based on face recognition |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910049340.6A CN109815878B (en) | 2019-01-18 | 2019-01-18 | Foothold analysis method and device based on face recognition |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109815878A true CN109815878A (en) | 2019-05-28 |
CN109815878B CN109815878B (en) | 2021-12-14 |
Family
ID=66603568
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910049340.6A Active CN109815878B (en) | 2019-01-18 | 2019-01-18 | Foothold analysis method and device based on face recognition |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109815878B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111291216A (en) * | 2020-02-28 | 2020-06-16 | 罗普特科技集团股份有限公司 | Method and system for analyzing foothold based on face structured data |
CN112232178A (en) * | 2020-10-13 | 2021-01-15 | 重庆紫光华山智安科技有限公司 | Region footfall point judgment method, system, device and medium based on portrait gathering |
CN112241672A (en) * | 2019-07-19 | 2021-01-19 | 杭州海康威视数字技术股份有限公司 | Method and device for associating identity data, electronic equipment and storage medium |
CN113496444A (en) * | 2020-03-19 | 2021-10-12 | 杭州海康威视***技术有限公司 | Method, device and system for identifying foothold |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP0654749A2 (en) * | 1993-11-22 | 1995-05-24 | Hitachi Europe Limited | An image processing method and apparatus |
CN101834801A (en) * | 2010-05-20 | 2010-09-15 | 哈尔滨工业大学 | Data caching and sequencing on-line processing method based on cache pool |
CN105741598A (en) * | 2014-12-09 | 2016-07-06 | 深圳中兴力维技术有限公司 | Suspect vehicle foothold analysis and processing method and device thereof |
US9552522B2 (en) * | 2014-01-11 | 2017-01-24 | Verint Systems Ltd. | Counting and monitoring method using face detection |
CN106529373A (en) * | 2015-09-10 | 2017-03-22 | 上海银晨智能识别科技有限公司 | Human face recognition monitoring alarm method and system |
CN106998444A (en) * | 2017-02-14 | 2017-08-01 | 北京中科天云科技有限公司 | A kind of big data face monitoring system and device |
CN107729795A (en) * | 2016-08-11 | 2018-02-23 | 汤棋 | A kind of information processing manner with face recognition |
CN108197565A (en) * | 2017-12-29 | 2018-06-22 | 深圳英飞拓科技股份有限公司 | Target based on recognition of face seeks track method and system |
CN108256443A (en) * | 2017-12-28 | 2018-07-06 | 深圳英飞拓科技股份有限公司 | A kind of personnel positioning method, system and terminal device |
CN108875835A (en) * | 2018-06-26 | 2018-11-23 | 北京旷视科技有限公司 | Object foothold determines method, apparatus, electronic equipment and computer-readable medium |
-
2019
- 2019-01-18 CN CN201910049340.6A patent/CN109815878B/en active Active
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP0654749A2 (en) * | 1993-11-22 | 1995-05-24 | Hitachi Europe Limited | An image processing method and apparatus |
CN101834801A (en) * | 2010-05-20 | 2010-09-15 | 哈尔滨工业大学 | Data caching and sequencing on-line processing method based on cache pool |
US9552522B2 (en) * | 2014-01-11 | 2017-01-24 | Verint Systems Ltd. | Counting and monitoring method using face detection |
CN105741598A (en) * | 2014-12-09 | 2016-07-06 | 深圳中兴力维技术有限公司 | Suspect vehicle foothold analysis and processing method and device thereof |
CN106529373A (en) * | 2015-09-10 | 2017-03-22 | 上海银晨智能识别科技有限公司 | Human face recognition monitoring alarm method and system |
CN107729795A (en) * | 2016-08-11 | 2018-02-23 | 汤棋 | A kind of information processing manner with face recognition |
CN106998444A (en) * | 2017-02-14 | 2017-08-01 | 北京中科天云科技有限公司 | A kind of big data face monitoring system and device |
CN108256443A (en) * | 2017-12-28 | 2018-07-06 | 深圳英飞拓科技股份有限公司 | A kind of personnel positioning method, system and terminal device |
CN108197565A (en) * | 2017-12-29 | 2018-06-22 | 深圳英飞拓科技股份有限公司 | Target based on recognition of face seeks track method and system |
CN108875835A (en) * | 2018-06-26 | 2018-11-23 | 北京旷视科技有限公司 | Object foothold determines method, apparatus, electronic equipment and computer-readable medium |
Non-Patent Citations (2)
Title |
---|
施梅芳 等: "《大学信息技术试验教程》", 31 August 2019, 苏州大学出版社 * |
李胜广 等: "人工智能在公安卡口中的应用", 《中国安防》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112241672A (en) * | 2019-07-19 | 2021-01-19 | 杭州海康威视数字技术股份有限公司 | Method and device for associating identity data, electronic equipment and storage medium |
CN112241672B (en) * | 2019-07-19 | 2024-05-03 | 杭州海康威视数字技术股份有限公司 | Identity data association method and device, electronic equipment and storage medium |
CN111291216A (en) * | 2020-02-28 | 2020-06-16 | 罗普特科技集团股份有限公司 | Method and system for analyzing foothold based on face structured data |
CN111291216B (en) * | 2020-02-28 | 2022-06-14 | 罗普特科技集团股份有限公司 | Method and system for analyzing foothold based on face structured data |
CN113496444A (en) * | 2020-03-19 | 2021-10-12 | 杭州海康威视***技术有限公司 | Method, device and system for identifying foothold |
CN112232178A (en) * | 2020-10-13 | 2021-01-15 | 重庆紫光华山智安科技有限公司 | Region footfall point judgment method, system, device and medium based on portrait gathering |
Also Published As
Publication number | Publication date |
---|---|
CN109815878B (en) | 2021-12-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109815878A (en) | Foothold analysis method and device based on recognition of face | |
CN106997629B (en) | Access control method, apparatus and system | |
CN101253535B (en) | Image retrieving apparatus and image search method | |
US8570376B1 (en) | Method and system for efficient sampling of videos using spatiotemporal constraints for statistical behavior analysis | |
CN108921008B (en) | Portrait identification method and device and electronic equipment | |
CN106204646A (en) | Multiple mobile object tracking based on BP neutral net | |
CN105469105A (en) | Cigarette smoke detection method based on video monitoring | |
CN109492604A (en) | Faceform's characteristic statistics analysis system | |
CN111291682A (en) | Method and device for determining target object, storage medium and electronic device | |
CN111460940A (en) | Stranger foot drop point studying and judging method and system | |
CN109684961A (en) | Population management's method and device based on recognition of face | |
CN110287856A (en) | A kind of security personnel's behavior analysis system, method and device | |
CN110287370A (en) | Suspect's method for tracing, device and storage medium based on field shoe print | |
CN108171135A (en) | Method for detecting human face, device and computer readable storage medium | |
CN111814690A (en) | Target re-identification method and device and computer readable storage medium | |
CN110019963B (en) | Method for searching suspect relation personnel | |
CN110826390A (en) | Video data processing method based on face vector characteristics | |
CN109086717B (en) | Violent behavior detection system and method based on human skeleton and motion signal characteristics | |
CN109063617A (en) | A kind of analysis method and tool of face recognition algorithms | |
CN112686226A (en) | Big data management method and device based on gridding management and electronic equipment | |
Hernández-Carrascosa et al. | TGCRBNW: A dataset for runner bib number detection (and recognition) in the wild | |
CN116866664A (en) | Time-segmentation-based monitoring video storage method | |
CN111784742A (en) | Cross-lens tracking method and device for pedestrians | |
CN114863364B (en) | Security detection method and system based on intelligent video monitoring | |
DE102009048117A1 (en) | Method and device for detecting a misdetection of an object in an 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 |