CN110232323A - A kind of parallel method for quickly identifying of plurality of human faces for crowd and its device - Google Patents
A kind of parallel method for quickly identifying of plurality of human faces for crowd and its device Download PDFInfo
- Publication number
- CN110232323A CN110232323A CN201910395419.4A CN201910395419A CN110232323A CN 110232323 A CN110232323 A CN 110232323A CN 201910395419 A CN201910395419 A CN 201910395419A CN 110232323 A CN110232323 A CN 110232323A
- Authority
- CN
- China
- Prior art keywords
- frame
- human face
- face region
- video frame
- video
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/40—Analysis of texture
- G06T7/41—Analysis of texture based on statistical description of texture
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/53—Recognition of crowd images, e.g. recognition of crowd congestion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30196—Human being; Person
- G06T2207/30201—Face
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Human Computer Interaction (AREA)
- General Health & Medical Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
Abstract
A kind of parallel method for quickly identifying of plurality of human faces for crowd provided by the embodiments of the present application and its device, wherein method includes: to obtain target monitoring video, and the target monitoring video includes multi-frame video frame, includes multiple human face regions in every frame video frame;For the video frame of the preset quantity in the target monitoring video, multiple human face regions in every frame video frame are determined, and establish in adjacent video frames the incidence relation for corresponding to human face region;Human face region in every frame video frame is analyzed, determines the identification degree of the human face region in every frame video frame;The target human face region for carrying out feature extraction and identification is determined from every frame video frame according to the identification degree;Feature extraction and identification are carried out to the target human face region.The parallel method for quickly identifying of plurality of human faces for crowd and its device of the embodiment of the present application can synchronize identification to the face in monitor video picture, avoid delay present in face recognition process and omit.
Description
Technical field
This application involves wisdom field of security technology more particularly to a kind of plurality of human faces for crowd quick identification sides parallel
Method and its device.
Background technique
Video monitoring is security and guard technology means common in actual life.With popularizing for intellectualized technology, for front end
The video pictures of monitoring device acquisition carry out automatic piece identity and are identified as current important developing direction, wherein being mainly
It is realized based on the extraction of face characteristic and identification.Especially AT STATION, the cities such as square, airport, commercial street are public
Space, by the monitor video image expansion identification towards larger field range, emphasis that can rapidly in locking crowd
Object promotes security protection efficiency and specific aim, maintains public order and public security.
Feature extraction is carried out to face and identification needs stronger operational capability.In particular, above-mentioned have wide-angle
Often there are multiple human face regions in the monitor video picture of field range, if concurrent operation needs very powerful hardware
Configuration, hardware device in practice are difficult to reach.Due to the limitation of operational capability, wherein one can only be locked in multiple faces
A human face region carries out feature extraction and identification, carries out the identification of next face again after an identification is completed, this
Sample will could be completed to result in video monitoring in this way to whole recognitions of face in monitor video picture by very long delay
The real-time that backstage obtains piece identity's recognition result in picture is poor, also likely causes to omit.In addition, since video is drawn
The renewal frequency in face be it is very fast, reach -30 frame of 10 frame per second, there are continuitys for the content of each frame video pictures, existing
There is no adequately utilize this continuity to face recognition technology.
Summary of the invention
In view of this, the purpose of the application be to propose a kind of parallel method for quickly identifying of the plurality of human faces for crowd and its
Device, to solve in the prior art to delay very long present in whole face recognition process in monitor video picture, together
When may cause in monitor video picture the technical issues of the omission of recognition of face.
Based on above-mentioned purpose, the first aspect of the application proposes a kind of plurality of human faces for crowd and quickly knows parallel
Other method, comprising:
Target monitoring video is obtained, the target monitoring video includes multi-frame video frame, includes multiple in every frame video frame
Human face region;
For the video frame of the preset quantity in the target monitoring video, multiple face areas in every frame video frame are determined
Domain, and the incidence relation for corresponding to human face region is established in adjacent video frames;
Human face region in every frame video frame is analyzed, determines can recognize for the human face region in every frame video frame
Degree;
The target face area for carrying out feature extraction and identification is determined from every frame video frame according to the identification degree
Domain;
Feature extraction and identification are carried out to the target human face region.
In some embodiments, multiple human face regions in the every frame video frame of the determination, comprising:
Edge detection is carried out to the image frame in every frame video frame, it will be in every frame video frame according to the quantity of closure edge
Image frame be divided into multiple regions, to each region carry out texture recognition, extract the textural characteristics in each region, will mention
Region after taking textural characteristics is based on textural characteristics and matches with faceform, determines multiple face areas in every frame video frame
Domain.
In some embodiments, the image frame in every frame video frame carries out edge detection, according to closure edge
Quantity the image frame in every frame video frame is divided into multiple regions, comprising:
Convolution is made with Gauss mask to the image frame in every frame video frame, the image frame in every frame video frame is carried out
Smoothing processing;
The ladder of each pixel of the image frame in every frame video frame after calculating smoothing processing using Sobel operator
Degree;
Retain the maximum of gradient intensity on each pixel of the image frame in every frame video frame, deletes other values;
Set on each pixel of the image frame in every frame video frame the threshold value upper bound of the maximum of gradient intensity and
The pixel that the maximum of gradient intensity is greater than the threshold value upper bound is confirmed as boundary, by the pole of gradient intensity by threshold value lower bound
Big value is greater than the threshold value lower bound and is confirmed as weak boundary less than the pixel in the threshold value upper bound, and the maximum of gradient intensity is small
Non- boundary is confirmed as in the pixel of the threshold value lower bound;
The weak boundary being connected with the boundary is confirmed into boundary, other weak boundaries are confirmed as non-boundary, thus will be by
The region that boundary surrounds is determined as the region of the image frame in every frame video frame.
It is in some embodiments, described to establish in adjacent video frames the incidence relation for corresponding to human face region, comprising:
Judge whether the variable quantity that the relative position of human face region is corresponded in adjacent video frames is less than preset threshold, and by phase
The human face region that the variable quantity of relative position is less than preset threshold in adjacent video frame is determined as with incidence relation.
In some embodiments, the human face region in every frame video frame is analyzed, and is determined in every frame video frame
Human face region identification degree, comprising:
Human face region in every frame video frame is analyzed, determines the shooting angle of the human face region in every frame video frame
Degree, size, the size of average brightness and shelter, generate the identification degree vector of human face region, according to the identification degree to
Amount determines the identification degree of the human face region in every frame video frame at a distance from standard vector.
In some embodiments, described determined from every frame video frame according to the identification degree carries out feature extraction and body
The target human face region of part identification, specifically includes:
The human face region that identification degree in every frame video frame is greater than preset threshold is determined as target human face region, generates mesh
Human face region collection is marked, so that the target human face region that the target human face region is concentrated includes the human face region in frame video frame.
In some embodiments, further includes:
For the target human face region with incidence relation, chooses and can recognize the maximum target human face region of angle value the most most
Whole target human face region.
Based on above-mentioned purpose, in the second aspect of the application, it is also proposed that a kind of plurality of human faces for crowd is fast parallel
Fast identification device, comprising:
Target monitoring video acquiring module, for obtaining target monitoring video, the target monitoring video includes multiframe view
Frequency frame includes multiple human face regions in every frame video frame;
Human face region determining module determines every for the video frame for the preset quantity in the target monitoring video
Multiple human face regions in frame video frame, and the incidence relation for corresponding to human face region is established in adjacent video frames;
Identification degree determining module determines every frame video frame for analyzing the human face region in every frame video frame
In human face region identification degree;
Target face area determination module, for determining that carrying out feature mentions from every frame video frame according to the identification degree
Take the target human face region with identification;
Identification module, for carrying out feature extraction and identification to the target human face region.
In some embodiments, the human face region determining module, is specifically used for:
Edge detection is carried out to the image frame in every frame video frame, it will be in every frame video frame according to the quantity of closure edge
Image frame be divided into multiple regions, to each region carry out texture recognition, extract the textural characteristics in each region, will mention
Region after taking textural characteristics is based on textural characteristics and matches with faceform, determines multiple face areas in every frame video frame
Domain judges whether the variable quantity that the relative position of human face region is corresponded in adjacent video frames is less than preset threshold, and by adjacent view
The human face region that the variable quantity of relative position is less than preset threshold in frequency frame is determined as with incidence relation.
In some embodiments, the identification degree determining module, is specifically used for:
Human face region in every frame video frame is analyzed, determines the shooting angle of the human face region in every frame video frame
Degree, size, the size of average brightness and shelter, generate the identification degree vector of human face region, according to the identification degree to
Amount determines the identification degree of the human face region in every frame video frame at a distance from standard vector.
A kind of parallel method for quickly identifying of plurality of human faces for crowd provided by the embodiments of the present application and its device, wherein side
Method includes: to obtain target monitoring video, and the target monitoring video includes multi-frame video frame, includes multiple people in every frame video frame
Face region;For the video frame of the preset quantity in the target monitoring video, multiple face areas in every frame video frame are determined
Domain, and the incidence relation for corresponding to human face region is established in adjacent video frames;Human face region in every frame video frame is analyzed,
Determine the identification degree of the human face region in every frame video frame;It is determined from every frame video frame according to the identification degree and carries out spy
Sign extracts the target human face region with identification;Feature extraction and identification are carried out to the target human face region.This Shen
Please embodiment the parallel method for quickly identifying of the plurality of human faces for crowd and its device, can be to the face in monitor video picture
Identification is synchronized, delay present in face recognition process is avoided and is omitted.
Detailed description of the invention
By reading a detailed description of non-restrictive embodiments in the light of the attached drawings below, the application's is other
Feature, objects and advantages will become more apparent upon:
Fig. 1 is the flow chart of the parallel method for quickly identifying of the plurality of human faces for crowd of the embodiment of the present application one;
Fig. 2 is the structural schematic diagram of the quick identification device parallel of the plurality of human faces for crowd of the embodiment of the present application two;
Fig. 3 is the human face region cross hairs signal of the parallel method for quickly identifying of the plurality of human faces for crowd of the embodiment of the present application one
Figure.
Specific embodiment
The application is described in further detail with reference to the accompanying drawings and examples.It is understood that this place is retouched
The specific embodiment stated is used only for explaining related invention, rather than the restriction to the invention.It also should be noted that in order to
Convenient for description, part relevant to related invention is illustrated only in attached drawing.
It should be noted that in the absence of conflict, the features in the embodiments and the embodiments of the present application can phase
Mutually combination.The application is described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.
Specifically, as shown in Figure 1, being the parallel method for quickly identifying of the plurality of human faces for crowd of the embodiment of the present application one
Flow chart.From figure 1 it appears that one embodiment as the application, the plurality of human faces for crowd is quickly known parallel
Other method, may comprise steps of:
S101: target monitoring video is obtained, the target monitoring video includes multi-frame video frame, includes in every frame video frame
Multiple human face regions.
The parallel method for quickly identifying of plurality of human faces for crowd of the embodiment of the present application, can be applied to public security protection or nothing
The fields such as people's monitoring, by being mounted on the video capture device in the biggish region of flow of the people, such as with video capture function
Camera acquires the movable video of crowd behaviour.Wherein, camera can use wide-angle lens, to increase mesh collected
Mark the field range of monitor video.When carrying out recognition of face to collected video, the method that can use the present embodiment.Tool
Body, it, can be using one section of video therein or all videos as target monitoring video, to described for collected video
Target monitoring video carries out plurality of human faces and quickly identifies parallel, so that the whole plurality of human faces that carries out to video quickly identifies parallel.It is logical
In normal situation, since the flow of the people in video is larger, the length of target monitoring video takes be advisable for 3 seconds, certainly, for flow of the people
The length of less video, target monitoring video can also slightly extend, such as 5 seconds, 10 seconds, can be to following step S102
In counted for the total quantity of human face region of target monitoring video extraction, and the mesh is adjusted according to count value dynamic
The setting duration of monitor video is marked, such as the length of initial target monitoring video takes 3 seconds, if in subsequent step S102
In in the target monitoring video counting of human face region be greater than preset first face amount threshold, then by the target monitoring video
Time span shorten to such as 2 seconds;, whereas if the count value of the human face region total quantity of target monitoring video extraction
Less than preset second face amount threshold, then the time span of the target monitoring video can be extended for such as 5 seconds or
Person 10 seconds.Alternatively, also can be set according to actual needs the length of the target monitoring video.The target monitoring video can
It include multiple human face regions in every frame video frame to include multi-frame video frame.
S102: it for the video frame of the preset quantity in the target monitoring video, determines multiple in every frame video frame
Human face region, and the incidence relation for corresponding to human face region is established in adjacent video frames.
In the present embodiment, after getting target monitoring video, for the preset quantity in the target monitoring video
Video frame, may further determine that multiple human face regions in every frame video frame of the target monitoring video.Monitor video
General each second includes 10 to 30 frame pictures, by taking 20 frames as an example, in order not to keep the calculation amount of single identification excessive, and usual situation
Under, it can choose 3 seconds namely 60 frame pictures identify multiple human face regions in every frame video frame.To human face region
During being identified, first have to determine multiple human face regions present in every frame video frame.Specifically, every frame can be regarded
Image frame in frequency frame carries out edge detection, is divided into the image frame in every frame video frame according to the quantity of closure edge
Multiple regions carry out texture recognition to each region, the textural characteristics in each region are extracted, by the area after texture feature extraction
Domain is based on textural characteristics and matches with faceform, determines multiple human face regions in every frame video frame.Wherein, every frame is regarded
Image frame in frequency frame carries out edge detection, can make convolution with Gauss mask to the image frame in every frame video frame, right
Image frame in every frame video frame is smoothed;In every frame video frame after calculating smoothing processing using Sobel operator
Image frame each pixel gradient;Retain gradient intensity on each pixel of the image frame in every frame video frame
Maximum, delete other values;Set the maximum of gradient intensity on each pixel of the image frame in every frame video frame
The threshold value upper bound and threshold value lower bound, by the maximum of gradient intensity be greater than the threshold value upper bound pixel be confirmed as boundary, will
The maximum of gradient intensity is greater than the threshold value lower bound and is confirmed as weak boundary less than the pixel in the threshold value upper bound, and gradient is strong
The pixel that the maximum of degree is less than the threshold value lower bound is confirmed as non-boundary;The weak boundary being connected with the boundary is confirmed into side
Other weak boundaries are confirmed as non-boundary by boundary, so that the region surrounded by boundary is determined as the image in every frame video frame
The region of picture.By edge detection, the enclosed region in the image frame in every frame video frame, the enclosed region can be extracted
Such as can be face, article, clothes or other there are the graphic edges of color difference.
It, can textural characteristics based on enclosed region and predetermined faceform for the enclosed region extracted
It is matched, and then determines the human face region in the image frame in every frame video frame.Textural characteristics are for illumination variation, angle
Degree offset is all insensitive, therefore can have and well adapt to changing capability, and the calculation method of textural characteristics is as follows: will extract
The boundary rectangle of any one enclosed region come is decomposed into N × N number of subregion, and the value range of N is 5-20;For wherein
Each sub-regions, for each pixel extraction in the subregion by center pixel of the pixel including the pixel upper left,
Upper, upper right, the right side, bottom right, under, lower-left, left side adjacent pixel 3 × 3 block of pixels;The image texture characteristic value T of the center pixelc
Are as follows:
Wherein icIndicate the grey scale pixel value of center pixel, ipThe grey scale pixel value for indicating adjacent pixel, according to upper left, it is upper,
Upper right, the right side, bottom right, under, the sequence of lower-left, a left side, the value of p is successively by 1 to 8;And
That is, in 3 × 3 block of pixels, using the gray value of center pixel as threshold value, by the ash of 8 adjacent pixels
Angle value is compared with it, if adjacent pixel gray value is more than or equal to center pixel gray value, which is marked as
1, otherwise the adjacent pixel is labeled as 0.In this way, 8 adjacent pixels in 3 × 3 block of pixels compared can produce 8 numerical value be 0
Or 1 label, according to upper left, upper, upper right, the right side, bottom right, under, lower-left, a left side sequence by the corresponding tag arrangement of adjacent pixel
For one 8 binary numbers, it is T which, which is converted into the decimal system,c, centered on pixel image texture it is special
Value indicative, and reflect with this value the texture information of the block of pixels.For each of N × N number of subregion subregion, obtain
The wherein image texture characteristic value of each pixel, and then carry out the histogram system of the subregion pixel image texture eigenvalue
Meter, obtains the histogram data of each subregion;The histogram data of whole subregions is combined, the data set of formation
Cooperation is the textural characteristics of the enclosed region.
By the textural characteristics of each enclosed region, the texture having with predetermined reflection face class enclosed region is special
The faceform of sign matches, detailed process are as follows: is trained using the feature classifiers of SVM support vector machines principle, shape
At the disaggregated model of face and non-human textural characteristics;Specifically, in the training stage, this feature classifier is worked as from video frame
In, a part of frame is extracted as training sample, such as in the installation and debugging stage of monitoring system, randomly selects 1000 frame video frames
As training sample;It, can be existing for mode manually demarcates in video frame for the video frame as training sample
Each face enclosed region, and according to the textural characteristics of approach presented above extraction face enclosed region;In turn, it will train
The textural characteristics of face enclosed region in sample input the feature classifiers of the SVM support vector machines, execute face closure
The training of region recognition;After training is completed, for the enclosed region textural characteristics of every frame video frame in target monitoring video,
This feature classifier is inputted, judges that each enclosed region is human face region or non-face area according to the output result of classifier
Domain.
After determining the human face region in the image frame in every frame video frame, due in two adjacent frame video frames pair
In the time difference answered, too big change will not occur for the offset of human face region, still by taking 20 frame per second as an example, then adjacent video frames
Time difference be 0.05 second, in 0.05 second, the offset of human face region does not have too big change, therefore, can set phase
Then the threshold value of the offset of human face region in adjacent video frame judges the relative position that human face region is corresponded in adjacent video frames
Whether variable quantity is less than preset threshold, and the variable quantity of relative position in adjacent video frames is less than to the human face region of preset threshold
It is determined as with incidence relation, i.e., the same human face region in adjacent video frames has incidence relation, in this way, to face area
When domain carries out feature extraction and identification, it can know to avoid repeated characteristic extraction is carried out to same human face region with identity
Not.In addition, be that feature extraction and identification are carried out to human face region easy to identify in every frame video frame in the present embodiment,
And feature extraction and identification are then carried out in other video frames for human face region not easy to identify, therefore, it is necessary to build
Found incidence relation of the same human face region in different video frame.
S103: analyzing the human face region in every frame video frame, and determine the human face region in every frame video frame can
Resolution.
In the present embodiment, it when determining multiple human face regions in every frame video frame, and establishes corresponding in adjacent video frames
After the incidence relation of human face region, the human face region in every frame video frame can be analyzed, be determined in every frame video frame
The identification degree of human face region.Specifically, the human face region in every frame video frame can be analyzed, determines every frame video frame
In human face region shooting angle, size, the size of average brightness and shelter, generate the identification degree of human face region to
Amount, according to the identification degree of the human face region determined at a distance from the identification degree vector and standard vector in every frame video frame.
For example, can determine its identification degree vector Xi=(ci, si, li, ri), i=to each of every frame video frame face region
The number of 1,2,3 ... wherein i table human face region, ci are the shooting angle parameter for the human face region that number is i, and si is that number is
The size parameter of the human face region of i, li are the average luminance parameter for the human face region that number is i, and ri is the face that number is i
The size parameter of the shelter in region, no shelter are then denoted as 0, are then normalized, make to ci, si, li, ri
The corresponding numerical value of ci, si, li and ri in an order of magnitude, then calculate identification degree vector Xi and standard vector X0 away from
From, and corresponding human face region identification degree vector Xi being determined as at a distance from standard vector X0 in every frame video frame can
Resolution.Make face in the Y direction with difference in face camera lens optical axis, because bowing or facing upward head wherein it is possible to set face
Angle deviating camera lens optical axis, face deviate with different angle the cross template in a variety of situations such as camera lens optical axis in X-direction, each
Cross template corresponds to scheduled deviation angle parameter, is formed according to eyes line in actual human face region and nose middle line
Cross hairs and each cross template in X, the differential seat angle of Y-direction, the smallest cross template of the sum of differential seat angle is determined, by the cross
The corresponding deviation angle parameter of template is determined as shooting angle ci, for example, showing the ten of an actual human face region in Fig. 3
Wordline.The size parameter si of human face region can use the cartographic represenation of area of face enclosed region.The average luminance parameter of human face region
Li can be indicated with the average brightness value of face enclosed region pixel.The size parameter ri of shelter can be used for face closed area
The occlusion area size of domain overlapping indicates.
S104: the target person for carrying out feature extraction and identification is determined from every frame video frame according to the identification degree
Face region.
It in the present embodiment, can will be every after determining the identification degree of the corresponding human face region in every frame video frame
The human face region that identification degree is greater than preset threshold in frame video frame is determined as target human face region, generates target human face region
Collection, so that the target human face region that the target human face region is concentrated includes whole human face regions in each frame video frame.For example,
In the video frame of the i-th frame to the i-th+n frame, there are human face region F1-Fm;According to the identification degree of human face region in each frame,
It can determine that being extracted in the i-th frame with the human face region of identification is F1, F5;The human face region for extracting and identifying in i+1 frame
It is F2, F3 ... is repeated the above process, until the human face region in each frame for extracting and identifying covers whole F1-Fm.
S105: feature extraction and identification are carried out to the target human face region.
After determining target human face region, it can use method in the prior art and spy carried out to the target human face region
Sign is extracted and identification, no longer illustrates here.
Due in the present embodiment, carry out feature extraction and identification is all the higher human face region of identification degree,
Therefore calculation resources can utmostly be saved and improve recognition speed.
The parallel method for quickly identifying of plurality of human faces for crowd of the embodiment of the present application, can be in monitor video picture
Face synchronizes identification, avoids delay present in face recognition process and omits.
In addition, an alternative embodiment as the application can also include: in the above-described embodiments
For the target human face region with incidence relation, chooses and can recognize the maximum target human face region of angle value the most most
Whole target human face region.Since identification degree of the same human face region in different video frames is all higher than preset threshold, it is
It avoids carrying out duplicate feature extraction and identification to same human face region, it can by same human face region in different views
Identification degree in frequency frame compares, and then chooses the target person that can recognize that the maximum target human face region of angle value is the most final
Face region, and feature extraction and identification are carried out to final target human face region, and for other and the final mesh
Marking human face region has the target human face region of incidence relation then without feature extraction and identification.
The parallel method for quickly identifying of plurality of human faces for crowd of the present embodiment, can obtain similar with above-described embodiment
Technical effect, which is not described herein again.
As shown in Fig. 2, being that the structure of the quick identification device parallel of the plurality of human faces for crowd of the embodiment of the present application two is shown
It is intended to.The plurality of human faces for crowd of the present embodiment quick identification device parallel may include:
Target monitoring video acquiring module 201, for obtaining target monitoring video, the target monitoring video includes multiframe
Video frame includes multiple human face regions in every frame video frame.
Specifically, the target monitoring video acquiring module 201 for example can be in above-described embodiment, and there is video to clap
The camera of camera shooting function can carry out the movable video of crowd behaviour by being installed in the biggish region of flow of the people
Acquisition.Specifically, it for collected video, can be regarded using one section of video therein or all videos as target monitoring
Frequently, it carries out plurality of human faces to the target monitoring video quickly to identify parallel, so that the whole progress plurality of human faces to video is fast parallel
Speed identification.Under normal conditions, since the flow of the people in video is larger, the length of target monitoring video takes be advisable for 3 seconds, certainly,
The length of the video less for flow of the people, target monitoring video can also slightly extend, such as 5 seconds, 10 seconds, or can also be with
The length of the target monitoring video is set according to actual needs.The target monitoring video may include multi-frame video frame, often
It include multiple human face regions in frame video frame.
Human face region determining module 202 is determined for the video frame for the preset quantity in the target monitoring video
Multiple human face regions in every frame video frame, and the incidence relation for corresponding to human face region is established in adjacent video frames.
Specifically, in the present embodiment, after getting target monitoring video, for pre- in the target monitoring video
If the video frame of quantity, it may further determine that multiple human face regions in every frame video frame of the target monitoring video.Prison
Controlling video general each second includes 10 to 30 frame pictures, by taking 20 frames as an example, in order not to keep the calculation amount of single identification excessive, usually
In the case of, it can choose 3 seconds namely 60 frame pictures identify multiple human face regions in every frame video frame.To face
During region is identified, first have to determine multiple human face regions present in every frame video frame.It specifically, can be to every
Image frame in frame video frame carries out edge detection, is drawn the image frame in every frame video frame according to the quantity of closure edge
It is divided into multiple regions, texture recognition is carried out to each region, the textural characteristics in each region are extracted, after texture feature extraction
Region be based on textural characteristics matched with faceform, determine multiple human face regions in every frame video frame.
After determining the human face region in the image frame in every frame video frame, due in two adjacent frame video frames pair
In the time difference answered, too big change will not occur for the offset of human face region, still by taking 20 frame per second as an example, then adjacent video frames
Time difference be 0.05 second, in 0.05 second, the offset of human face region does not have too big change, therefore, can set phase
Then the threshold value of the offset of human face region in adjacent video frame judges the relative position that human face region is corresponded in adjacent video frames
Whether variable quantity is less than preset threshold, and the variable quantity of relative position in adjacent video frames is less than to the human face region of preset threshold
It is determined as with incidence relation, i.e., the same human face region in adjacent video frames has incidence relation, in this way, to face area
When domain carries out feature extraction and identification, it can know to avoid repeated characteristic extraction is carried out to same human face region with identity
Not.In addition, be that feature extraction and identification are carried out to human face region easy to identify in every frame video frame in the present embodiment,
And feature extraction and identification are then carried out in other video frames for human face region not easy to identify, therefore, it is necessary to build
Found incidence relation of the same human face region in different video frame.
Identification degree determining module 203 determines every frame video for analyzing the human face region in every frame video frame
The identification degree of human face region in frame.
In the present embodiment, it when determining multiple human face regions in every frame video frame, and establishes corresponding in adjacent video frames
After the incidence relation of human face region, the human face region in every frame video frame can be analyzed, be determined in every frame video frame
The identification degree of human face region.Specifically, the human face region in every frame video frame can be analyzed, determines every frame video frame
In human face region shooting angle, size, the size of average brightness and shelter, generate the identification degree of human face region to
Amount, according to the identification degree of the human face region determined at a distance from the identification degree vector and standard vector in every frame video frame.
For example, can determine its identification degree vector Xi=(ci, si, li, ri), i to each of every frame video frame face region
The number of=1,2,3 ... wherein i table human face regions, ci are the shooting angle for the human face region that number is i, and si is that number is i
Human face region size, li is the average brightness for the human face region that number is i, and ri is that the human face region that number is i blocks
The size of object, no shelter are then denoted as 0, and then ci, si, li, ri are normalized, so that ci, si, li and ri couple
Then the numerical value answered calculates identification degree vector Xi at a distance from standard vector X0 in an order of magnitude, and by identification degree to
Amount Xi is determined as the identification degree of the corresponding human face region in every frame video frame at a distance from standard vector X0.
Target face area determination module 204 carries out spy for determining from every frame video frame according to the identification degree
Sign extracts the target human face region with identification.
It in the present embodiment, can will be every after determining the identification degree of the corresponding human face region in every frame video frame
The human face region that identification degree is greater than preset threshold in frame video frame is determined as target human face region, generates target human face region
Collection, so that the target human face region that the target human face region is concentrated includes the human face region in frame video frame.For example, i-th
Frame is into the video frame of the i-th+n frame, and there are human face region F1-Fm;It, can be true according to the identification degree of human face region in each frame
Being extracted in fixed i-th frame with the human face region of identification is F1, F5;The human face region for extracting and identifying in i+1 frame is F2,
F3 ... is repeated the above process, until the human face region in each frame for extracting and identifying covers whole F1-Fm.
Identification module 205, for carrying out feature extraction and identification to the target human face region.
The plurality of human faces for crowd of the present embodiment quick identification device parallel, can be to the face in monitor video picture
Identification is synchronized, delay present in face recognition process is avoided and is omitted.
Above description is only the preferred embodiment of the application and the explanation to institute's application technology principle.Those skilled in the art
Member is it should be appreciated that invention scope involved in the application, however it is not limited to technology made of the specific combination of above-mentioned technical characteristic
Scheme, while should also cover in the case where not departing from foregoing invention design, it is carried out by above-mentioned technical characteristic or its equivalent feature
Any combination and the other technical solutions formed.Such as features described above has similar function with (but being not limited to) disclosed herein
Can technical characteristic replaced mutually and the technical solution that is formed.
Claims (10)
1. a kind of parallel method for quickly identifying of plurality of human faces for crowd characterized by comprising
Target monitoring video is obtained, the target monitoring video includes multi-frame video frame, includes multiple faces in every frame video frame
Region;
For the video frame of the preset quantity in the target monitoring video, multiple human face regions in every frame video frame are determined,
And the incidence relation for corresponding to human face region is established in adjacent video frames;
Human face region in every frame video frame is analyzed, determines the identification degree of the human face region in every frame video frame;
The target human face region for carrying out feature extraction and identification is determined from every frame video frame according to the identification degree;
Feature extraction and identification are carried out to the target human face region.
2. the method according to claim 1, wherein multiple human face regions in the every frame video frame of the determination,
Include:
Edge detection is carried out to the image frame in every frame video frame, according to the quantity of closure edge by the figure in every frame video frame
As picture is divided into multiple regions, texture recognition is carried out to each region, the textural characteristics in each region is extracted, line will be extracted
Region after reason feature is based on textural characteristics and matches with faceform, determines multiple human face regions in every frame video frame.
3. according to the method described in claim 2, it is characterized in that, the image frame in every frame video frame carries out edge
Detection, is divided into multiple regions for the image frame in every frame video frame according to the quantity of closure edge, comprising:
Convolution is made with Gauss mask to the image frame in every frame video frame, the image frame in every frame video frame is carried out smooth
Processing;
The gradient of each pixel of the image frame in every frame video frame after calculating smoothing processing using Sobel operator;
Retain the maximum of gradient intensity on each pixel of the image frame in every frame video frame, deletes other values;
Set the threshold value upper bound of the maximum of gradient intensity and threshold value on each pixel of the image frame in every frame video frame
The pixel that the maximum of gradient intensity is greater than the threshold value upper bound is confirmed as boundary, by the maximum of gradient intensity by lower bound
The pixel for being less than the threshold value upper bound greater than the threshold value lower bound is confirmed as weak boundary, and the maximum of gradient intensity is less than institute
The pixel for stating threshold value lower bound is confirmed as non-boundary;
The weak boundary being connected with the boundary is confirmed into boundary, other weak boundaries are confirmed as non-boundary, thus will be by boundary
The region surrounded is determined as the region of the image frame in every frame video frame.
4. according to the method described in claim 3, it is characterized in that, described establish in adjacent video frames the pass for corresponding to human face region
Connection relationship, comprising:
Judge whether the variable quantity that the relative position of human face region is corresponded in adjacent video frames is less than preset threshold, and by adjacent view
The human face region that the variable quantity of relative position is less than preset threshold in frequency frame is determined as with incidence relation.
5. according to the method described in claim 4, it is characterized in that, the human face region in every frame video frame divides
Analysis, determines the identification degree of the human face region in every frame video frame, comprising:
Human face region in every frame video frame is analyzed, determines the shooting angle, big of the human face region in every frame video frame
The size of small, average brightness and shelter generates the identification degree vector of human face region, according to the identification degree vector and mark
The distance of quasi- vector determines the identification degree of the human face region in every frame video frame.
6. according to the method described in claim 5, it is characterized in that, described true from every frame video frame according to the identification degree
Surely the target human face region for carrying out feature extraction and identification, specifically includes:
The human face region that identification degree in every frame video frame is greater than preset threshold is determined as target human face region, generates target person
Face region collection, so that the target human face region that the target human face region is concentrated includes the human face region in frame video frame.
7. according to the method described in claim 6, it is characterized by further comprising:
For the target human face region with incidence relation, chooses and can recognize that the maximum target human face region of angle value is the most final
Target human face region.
8. a kind of plurality of human faces for crowd quick identification device parallel characterized by comprising
Target monitoring video acquiring module, for obtaining target monitoring video, the target monitoring video includes multi-frame video frame,
It include multiple human face regions in every frame video frame;
Human face region determining module determines every frame view for the video frame for the preset quantity in the target monitoring video
Multiple human face regions in frequency frame, and the incidence relation for corresponding to human face region is established in adjacent video frames;
Identification degree determining module determines in every frame video frame for analyzing the human face region in every frame video frame
The identification degree of human face region;
Target face area determination module, for according to the identification degree from every frame video frame determine carry out feature extraction with
The target human face region of identification;
Identification module, for carrying out feature extraction and identification to the target human face region.
9. device according to claim 8, which is characterized in that the human face region determining module is specifically used for:
Edge detection is carried out to the image frame in every frame video frame, according to the quantity of closure edge by the figure in every frame video frame
As picture is divided into multiple regions, texture recognition is carried out to each region, the textural characteristics in each region is extracted, line will be extracted
Region after reason feature is based on textural characteristics and matches with faceform, determines multiple human face regions in every frame video frame,
Judge whether the variable quantity that the relative position of human face region is corresponded in adjacent video frames is less than preset threshold, and by adjacent video frames
The human face region that the variable quantity of middle relative position is less than preset threshold is determined as with incidence relation.
10. device according to claim 9, which is characterized in that the identification degree determining module is specifically used for:
Human face region in every frame video frame is analyzed, determines the shooting angle, big of the human face region in every frame video frame
The size of small, average brightness and shelter generates the identification degree vector of human face region, according to the identification degree vector and mark
The distance of quasi- vector determines the identification degree of the human face region in every frame video frame.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910395419.4A CN110232323A (en) | 2019-05-13 | 2019-05-13 | A kind of parallel method for quickly identifying of plurality of human faces for crowd and its device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910395419.4A CN110232323A (en) | 2019-05-13 | 2019-05-13 | A kind of parallel method for quickly identifying of plurality of human faces for crowd and its device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110232323A true CN110232323A (en) | 2019-09-13 |
Family
ID=67860587
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910395419.4A Pending CN110232323A (en) | 2019-05-13 | 2019-05-13 | A kind of parallel method for quickly identifying of plurality of human faces for crowd and its device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110232323A (en) |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110633648A (en) * | 2019-08-21 | 2019-12-31 | 重庆特斯联智慧科技股份有限公司 | Face recognition method and system in natural walking state |
CN111145189A (en) * | 2019-12-26 | 2020-05-12 | 成都市喜爱科技有限公司 | Image processing method, image processing device, electronic equipment and computer readable storage medium |
CN111507245A (en) * | 2020-04-15 | 2020-08-07 | 海信集团有限公司 | Embedded system and method for face detection |
CN111539253A (en) * | 2020-03-24 | 2020-08-14 | 深圳英飞拓科技股份有限公司 | Face detection method, device, terminal and storage medium |
CN112101305A (en) * | 2020-05-12 | 2020-12-18 | 杭州宇泛智能科技有限公司 | Multi-path image processing method and device and electronic equipment |
CN112200084A (en) * | 2020-10-10 | 2021-01-08 | 华航高科(北京)技术有限公司 | Face recognition method and device for video stream, electronic equipment and storage medium |
CN112614160A (en) * | 2020-12-24 | 2021-04-06 | 中标慧安信息技术股份有限公司 | Multi-object face tracking method and system |
CN113382046A (en) * | 2021-05-27 | 2021-09-10 | 青岛海信智慧生活科技股份有限公司 | Method and device for changing face information in community |
CN115272923A (en) * | 2022-07-22 | 2022-11-01 | 华中科技大学同济医学院附属协和医院 | Intelligent identification method and system based on big data platform |
WO2023173785A1 (en) * | 2022-03-18 | 2023-09-21 | 珠海优特电力科技股份有限公司 | Access permission verification method, device, and system and identity authentication terminal |
GB2620664A (en) * | 2022-03-18 | 2024-01-17 | Zhuhai Unitech Power Tech Co | Access permission verification method, device, and system and identity authentication terminal |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20130044923A1 (en) * | 2008-12-05 | 2013-02-21 | DigitalOptics Corporation Europe Limited | Face Recognition Using Face Tracker Classifier Data |
CN107346426A (en) * | 2017-07-10 | 2017-11-14 | 深圳市海清视讯科技有限公司 | A kind of face information collection method based on video camera recognition of face |
CN108171207A (en) * | 2018-01-17 | 2018-06-15 | 百度在线网络技术(北京)有限公司 | Face identification method and device based on video sequence |
CN108416336A (en) * | 2018-04-18 | 2018-08-17 | 特斯联(北京)科技有限公司 | A kind of method and system of intelligence community recognition of face |
CN109034013A (en) * | 2018-07-10 | 2018-12-18 | 腾讯科技(深圳)有限公司 | A kind of facial image recognition method, device and storage medium |
CN109299690A (en) * | 2018-09-21 | 2019-02-01 | 浙江中正智能科技有限公司 | A method of video real-time face accuracy of identification can be improved |
CN109359548A (en) * | 2018-09-19 | 2019-02-19 | 深圳市商汤科技有限公司 | Plurality of human faces identifies monitoring method and device, electronic equipment and storage medium |
-
2019
- 2019-05-13 CN CN201910395419.4A patent/CN110232323A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20130044923A1 (en) * | 2008-12-05 | 2013-02-21 | DigitalOptics Corporation Europe Limited | Face Recognition Using Face Tracker Classifier Data |
CN107346426A (en) * | 2017-07-10 | 2017-11-14 | 深圳市海清视讯科技有限公司 | A kind of face information collection method based on video camera recognition of face |
CN108171207A (en) * | 2018-01-17 | 2018-06-15 | 百度在线网络技术(北京)有限公司 | Face identification method and device based on video sequence |
CN108416336A (en) * | 2018-04-18 | 2018-08-17 | 特斯联(北京)科技有限公司 | A kind of method and system of intelligence community recognition of face |
CN109034013A (en) * | 2018-07-10 | 2018-12-18 | 腾讯科技(深圳)有限公司 | A kind of facial image recognition method, device and storage medium |
CN109359548A (en) * | 2018-09-19 | 2019-02-19 | 深圳市商汤科技有限公司 | Plurality of human faces identifies monitoring method and device, electronic equipment and storage medium |
CN109299690A (en) * | 2018-09-21 | 2019-02-01 | 浙江中正智能科技有限公司 | A method of video real-time face accuracy of identification can be improved |
Cited By (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110633648B (en) * | 2019-08-21 | 2020-09-11 | 重庆特斯联智慧科技股份有限公司 | Face recognition method and system in natural walking state |
CN110633648A (en) * | 2019-08-21 | 2019-12-31 | 重庆特斯联智慧科技股份有限公司 | Face recognition method and system in natural walking state |
CN111145189B (en) * | 2019-12-26 | 2023-08-08 | 成都市喜爱科技有限公司 | Image processing method, apparatus, electronic device, and computer-readable storage medium |
CN111145189A (en) * | 2019-12-26 | 2020-05-12 | 成都市喜爱科技有限公司 | Image processing method, image processing device, electronic equipment and computer readable storage medium |
CN111539253A (en) * | 2020-03-24 | 2020-08-14 | 深圳英飞拓科技股份有限公司 | Face detection method, device, terminal and storage medium |
CN111539253B (en) * | 2020-03-24 | 2024-03-05 | 深圳英飞拓仁用信息有限公司 | Face detection method, device, terminal and storage medium |
CN111507245A (en) * | 2020-04-15 | 2020-08-07 | 海信集团有限公司 | Embedded system and method for face detection |
CN112101305A (en) * | 2020-05-12 | 2020-12-18 | 杭州宇泛智能科技有限公司 | Multi-path image processing method and device and electronic equipment |
CN112200084A (en) * | 2020-10-10 | 2021-01-08 | 华航高科(北京)技术有限公司 | Face recognition method and device for video stream, electronic equipment and storage medium |
CN112614160B (en) * | 2020-12-24 | 2021-08-31 | 中标慧安信息技术股份有限公司 | Multi-object face tracking method and system |
CN112614160A (en) * | 2020-12-24 | 2021-04-06 | 中标慧安信息技术股份有限公司 | Multi-object face tracking method and system |
CN113382046A (en) * | 2021-05-27 | 2021-09-10 | 青岛海信智慧生活科技股份有限公司 | Method and device for changing face information in community |
CN113382046B (en) * | 2021-05-27 | 2022-07-01 | 青岛海信智慧生活科技股份有限公司 | Method and device for changing face information in community |
WO2023173785A1 (en) * | 2022-03-18 | 2023-09-21 | 珠海优特电力科技股份有限公司 | Access permission verification method, device, and system and identity authentication terminal |
GB2620664A (en) * | 2022-03-18 | 2024-01-17 | Zhuhai Unitech Power Tech Co | Access permission verification method, device, and system and identity authentication terminal |
CN115272923A (en) * | 2022-07-22 | 2022-11-01 | 华中科技大学同济医学院附属协和医院 | Intelligent identification method and system based on big data platform |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110232323A (en) | A kind of parallel method for quickly identifying of plurality of human faces for crowd and its device | |
EP2959454B1 (en) | Method, system and software module for foreground extraction | |
CN103729858B (en) | A kind of video monitoring system is left over the detection method of article | |
Kumar et al. | An efficient approach for detection and speed estimation of moving vehicles | |
CN104036278B (en) | The extracting method of face algorithm standard rules face image | |
JP5271227B2 (en) | Crowd monitoring device, method and program | |
WO2019114145A1 (en) | Head count detection method and device in surveillance video | |
CN102609724B (en) | Method for prompting ambient environment information by using two cameras | |
JP5726596B2 (en) | Image monitoring device | |
Prabhakar et al. | A novel design for vehicle license plate detection and recognition | |
CN103530638A (en) | Method for matching pedestrians under multiple cameras | |
CN104463232A (en) | Density crowd counting method based on HOG characteristic and color histogram characteristic | |
CN102542552A (en) | Frontlighting and backlighting judgment of video images and detection method of shooting time | |
Abbas et al. | Realization of multiple human head detection and direction movement using Raspberry Pi | |
CN104318216B (en) | Across the identification matching process of blind area pedestrian target in video monitoring | |
CN104899559B (en) | A kind of rapid pedestrian detection method based on video monitoring | |
Das et al. | Human face detection in color images using HSV color histogram and WLD | |
KR101711328B1 (en) | Method for classifying children and adult by using head and body height in images obtained from camera such as CCTV | |
Ilao et al. | Crowd estimation using region-specific HOG With SVM | |
CN104615985B (en) | A kind of recognition methods of human face similarity degree | |
JP7092616B2 (en) | Object detection device, object detection method, and object detection program | |
KR102150661B1 (en) | Method and Apparatus for Preprocessing Image for Detecting Objects | |
Chen et al. | Taxi drivers' smoking behavior detection in traffic monitoring video | |
CN107527357B (en) | Face datection positioning and method for real time tracking in Violent scene | |
KR20160017152A (en) | Gender Classification Based on Binary Haar Cascade |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190913 |