CN108664853A - Method for detecting human face and device - Google Patents
Method for detecting human face and device Download PDFInfo
- Publication number
- CN108664853A CN108664853A CN201710203886.3A CN201710203886A CN108664853A CN 108664853 A CN108664853 A CN 108664853A CN 201710203886 A CN201710203886 A CN 201710203886A CN 108664853 A CN108664853 A CN 108664853A
- Authority
- CN
- China
- Prior art keywords
- area
- interest
- humanoid
- present frame
- face
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
Abstract
A kind of method for detecting human face of present invention offer and device.The method includes：Present frame is pre-processed, present frame gray figure is obtained；Humanoid detection is carried out to the present frame gray figure using humanoid grader；Judge whether the present frame gray figure detects humanoid region, when detecting humanoid region, select area-of-interest of the humanoid region as Face datection, otherwise finds the area-of-interest of the Face datection of present frame；Face datection is carried out to the area-of-interest using face grader；Face testing result is orderly preserved according to registration.The present invention can improve the detection efficiency of Face datection.
Description
Technical field
The present invention relates to mode identification technology more particularly to a kind of method for detecting human face and device.
Background technology
Recognition of face refers to comparing the computer technology that face visual signature information carries out identity authentication using analysis, is mesh
The emphasis of preceding artificial intelligence and pattern-recognition is widely used in national security, military security, identification, bank and sea
The fields such as monitoring, access control system, the video conference of pass.
Face datection is the key link in recognition of face, and Face datection refers to the image given for any one width, is adopted
It is scanned for certain strategy with determine whether if it is to return containing face the position of face, size and
Posture.
Common method for detecting human face traverses entire image at this stage, is examined to entire image using grader
It surveys, efficiency is low and easily causes flase drop.
Invention content
Method for detecting human face and device provided by the invention, can improve the efficiency of Face datection.
In a first aspect, the present invention provides a kind of method for detecting human face, including：
Present frame is pre-processed, present frame gray figure is obtained；
Humanoid detection is carried out to the present frame gray figure using humanoid grader；
Judge whether the present frame gray figure detects humanoid region, when detecting humanoid region, selects the people
Otherwise area-of-interest of the shape region as Face datection finds the area-of-interest of the Face datection of present frame；
Face datection is carried out to the area-of-interest using face grader；
Face testing result is orderly preserved according to registration.
Optionally, described to include to the humanoid detection of present frame gray figure progress using humanoid grader：
Multiple dimensioned scaling is carried out to the present frame gray figure according to the first image scale factor, obtains multiple scaling gray scales
Figure；
Extract the feature of the multiple scaling gray-scale map；
Detect all gray-scale map child windows of each scaling gray-scale map successively using the first detection window；
Judge whether each gray-scale map child window is humanoid region；
Clustering is carried out to all humanoid regions detected, obtains final humanoid region.
Optionally, the area-of-interest of the Face datection for finding present frame includes：
Increase an offset on the basis of position coordinates of the testing result of the former frame of the present frame of preservation, obtains
The position coordinates of prediction result；
The prediction result is expanded, the area-of-interest of the Face datection of present frame is obtained.
Optionally, described to include to area-of-interest progress Face datection using face grader：
Multiple dimensioned scaling is carried out to the area-of-interest according to the second image scale factor, it is interested to obtain multiple scalings
Region；
Extract the feature of the multiple scaling area-of-interest；
Detect all area-of-interest child windows of each scaling area-of-interest successively using the second detection window；
Judge whether each area-of-interest child window is human face region；
Clustering is carried out to all human face regions detected, obtains final human face region.
Optionally, described face testing result is orderly preserved according to registration to include：
Compare the registration of each testing result of present frame and each testing result of former frame successively；
The testing result of each target of preservation is updated according to registration comparison result.
Second aspect, the present invention provide a kind of human face detection device, including：
Preprocessing module obtains present frame gray figure for being pre-processed to present frame；
Humanoid detection module, for carrying out humanoid detection to the present frame gray figure using humanoid grader；
Judgment module, for judging whether present frame gray figure detects humanoid region；
First area-of-interest determining module, for when detecting humanoid region, selecting the humanoid region as people
The area-of-interest of face detection；
Second area-of-interest determining module, the face for when not detecting humanoid region, finding present frame are examined
The area-of-interest of survey；
Face detection module, for carrying out Face datection to the area-of-interest using face grader；
Preserving module, for orderly preserving face testing result according to registration.
Optionally, the humanoid detection module includes：
First unit for scaling, for carrying out multiple dimensioned contracting to the present frame gray figure according to the first image scale factor
It puts, obtains multiple scaling gray-scale maps；
First extraction unit, the feature for extracting the multiple scaling gray-scale map；
First detection unit, all gray scales for detecting each scaling gray-scale map successively using the first detection window
Figure child window；
First judging unit, for judging whether each gray-scale map child window is humanoid region；
First cluster cell obtains final humanoid area for carrying out clustering to all humanoid regions detected
Domain.
Optionally, the second area-of-interest determining module includes：
Computing unit, for increasing by one on the basis of the position coordinates of the testing result of the former frame of the present frame of preservation
A offset obtains the position coordinates of prediction result；
Expansion unit obtains the area-of-interest of the Face datection of present frame for expanding the prediction result.
Optionally, the face detection module includes：
Second unit for scaling, for carrying out multiple dimensioned scaling to the area-of-interest according to the second image scale factor,
Obtain multiple scaling area-of-interests；
Second extraction unit, the feature for extracting the multiple scaling area-of-interest；
Second detection unit, for detecting all of each scaling area-of-interest successively using the second detection window
Area-of-interest child window；
Second judgment unit, for judging whether each area-of-interest child window is human face region；
Second cluster cell obtains final face area for carrying out clustering to all human face regions detected
Domain.
Optionally, the preserving module includes：
Comparing unit, the registration of each testing result of each testing result and former frame for comparing present frame successively；
Updating unit, for being updated to the testing result of each target of preservation according to registration comparison result.
Method for detecting human face and device provided by the invention, first pass through humanoid detection, find the interested of image to be detected
Then region carries out Face datection again to area-of-interest, reduce the traversal range of Face datection, compared with prior art,
Humanoid detection and Face datection are combined, while being not entirely dependent on humanoid detection again, improves the detection effect of Face datection
Rate, while reducing false drop rate.
Description of the drawings
Fig. 1 is the flow chart of the method for detecting human face of an embodiment of the present invention；
Fig. 2 is the structural schematic diagram of the human face detection device of an embodiment of the present invention；
Fig. 3 is the structural schematic diagram of the humanoid detection module of the human face detection device in Fig. 2；
Fig. 4 is the structural schematic diagram of the second area-of-interest determining module of the human face detection device in Fig. 2；
Fig. 5 is the structural schematic diagram of the face detection module of the human face detection device in Fig. 2；
Fig. 6 is the structural schematic diagram of the preserving module of the human face detection device in Fig. 2.
Specific implementation mode
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is only
It is only a part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill
The every other embodiment that personnel are obtained without making creative work, shall fall within the protection scope of the present invention.
The embodiment of the present invention provides a kind of method for detecting human face, as shown in Figure 1, the method includes：
S11, present frame is pre-processed, obtains present frame gray figure.
Input needs the video frame images detected, using image to be detected as present frame.The original image of input is all color
Present frame is converted into gray level image by color image by coloured image.
S12, humanoid detection is carried out to the present frame gray figure using humanoid grader.
Before starting detection, humanoid data model and human face data model are loaded, i.e., reads humanoid grader and people respectively
The data of face grader are saved in corresponding memory.The humanoid grader refers to the classification trained by humanoid sample database
Device, the face classification device refer to the grader trained by Face Sample Storehouse.
Specifically, include the following steps：
1) feature for extracting the present frame gray figure, calculates characteristic value.
2) detection is begun stepping through at the upper left corner (0,0) of the present frame gray figure using the first detection window, successively
All gray-scale map child windows of the present frame gray figure are detected, when first detection window is with the humanoid classifier training
Window size be consistent；
For any one child window, by first weak typing of the first order strong classifier of the humanoid grader
Device is counted, if the characteristic value of child window is less than the threshold value of the Weak Classifier, is directly exited detection, is otherwise continued to judge next
Weak Classifier, until by this grade of strong classifier, being further continued for entering next strong classifier, until exiting detection or by complete
Portion's strong classifier.Each child window when exiting detection, can all return the child window by current strong classifier series,
Under normal circumstances, the step-length of traversal is set as step, i.e., offset next time adds step on the basis of previous, if currently
The series for the strong classifier that child window is returned when exiting detection is no more than 2 grades, i.e., has only passed through 1 grade or 2 grades of strong classifiers, then
The step-length of traversal is changed to 2*step temporarily, directly skips next child window, that is to say, that, it is believed that current sub-window it is next
Child window is that the possibility of target is minimum, is not detected directly.
3) judge whether each gray-scale map child window is humanoid region, when the gray-scale map child window passes through the people
When shape grader, the gray-scale map child window is humanoid region；
Only when a gray-scale map child window passes sequentially through whole strong classifiers, which is considered as just people
Shape region simultaneously preserves.
4) factor (the first image scale factor) scales the present frame gray figure according to set proportion, repeat 1)~
3)；
5) step 4) is repeated, until present frame gray figure zooms to the minimum value of setting, usual minimum value is set as one
The size of first detection window, at this time the first detection window size as image to be detected, it is only necessary to one-time detection.
By above-mentioned steps, all people's shape region can be detected in present frame gray figure and is preserved.
6) clustering is carried out to the humanoid region detected.
Registration is considered the same target in 70% or more humanoid region and is merged, and merges equal using taking
The principle of value.Remaining humanoid region is exactly final humanoid region after merging.
Humanoid region for only detecting 1 time is considered as flase drop and is removed.
S13, judge whether the present frame gray figure detects humanoid region.If detecting humanoid region, execute
Otherwise S14 executes S15.
S14, area-of-interest of the humanoid region detected as Face datection is selected.
S15, find present frame Face datection area-of-interest.
When not detecting humanoid region, the testing result according to some target in continuous multiple frames before calculates one at this time
A offset calculates an offset first by taking the testing result of two continuous frames before present frame as an example,
Offset expression formula is：
Shift=(center_0-center_1) * 0.8；
Wherein shift is offset, and center_0 is the central point of the testing result of the former frame of present frame, center_1
For the central point of the testing result of the second frame before present frame.
Then the position coordinates of the prediction result of present frame are：
Objects_roi=Rect (r.x+shifit.x, r.y+shifit.y, r, width, r.height)；
Wherein r indicates that the position of the testing result of former frame, r.x indicate that the x coordinate of testing result, shifit.x indicate x
The offset of coordinate, r.y indicate that the y-coordinate of testing result, shifit.y indicate that the offset of y-coordinate, r.width indicate detection
As a result width, r.height indicate the height of testing result.
The prediction result of the present frame suitably expand (such as being expanded to original 2 times) as our present frames
The area-of-interest of Face datection.
S16, Face datection is carried out to the area-of-interest using face grader.
Specifically, include the following steps：
11) feature for extracting the area-of-interest, calculates characteristic value；
12) the second detection window is used to detect all child windows of the area-of-interest, the second detection window successively
Window size when mouth is trained with the face classification device is consistent；
For any one child window, by first weak typing of the first order strong classifier of the face classification device
Device is counted, if the characteristic value of child window is less than the threshold value of the Weak Classifier, is directly exited detection, is otherwise continued to judge next
Weak Classifier, until by this grade of strong classifier, being further continued for entering next strong classifier, until exiting detection or by complete
Portion's strong classifier.Each child window when exiting detection, can all return the child window by current strong classifier series,
Under normal circumstances, the step-length of traversal is set as step, i.e., offset next time adds step on the basis of previous, if currently
The series for the strong classifier that child window is returned when exiting detection is no more than 2 grades, i.e., has only passed through 1 grade or 2 grades of strong classifiers, then
The step-length of traversal is changed to 2*step temporarily, directly skips next child window, that is to say, that, it is believed that current sub-window it is next
Child window is that the possibility of target is minimum, is not detected directly.
13) judge whether each child window is human face region, when the area-of-interest child window passes through the people
When face grader, the area-of-interest child window is human face region；
Only when a child window passes sequentially through whole strong classifiers, which is considered as just human face region and preserves
Get off.
14) factor (the second image scale factor) scales the area-of-interest according to set proportion, repeat 11)~
13)；
15) step 14) is repeated, until the area-of-interest zooms to the minimum value of setting, usual minimum value is set as
The size of one the second detection window, at this time the second detection window size as image to be detected, it is only necessary to which one-time detection is
It can.
By above-mentioned steps, all people's face region can be detected in the area-of-interest and is preserved.
16) clustering is carried out to the human face region detected.
Registration is considered the same target in 70% or more human face region and is merged, and merges equal using taking
The principle of value.Remaining human face region is exactly final human face region after merging.
S17, face testing result is orderly preserved according to registration.
For first frame, preserve in all testing results to the storage array vector of corresponding target, such as the inspection of target 1
Result to be surveyed to be saved in vector1, the testing result of target 2 is saved in vector2, and so on, it, will since the second frame
The testing result of present frame carries out the comparison of registration with the testing result of the former frame of preservation successively, and registration expression formula is such as
Under：
Coincidence_ij=min (s_i_area/now_i_area, s_i_area/pre_j_area)；
Wherein, coincidence_ij indicates registration, i-th of testing result of s_i_erea present frames and former frame
The area of the lap of j-th of testing result, now_i_area are the area of i-th of testing result of present frame, pre_
J_area is the area of j-th of testing result of former frame (the known result belongs to target 1)；
If the registration of the two is more than first threshold, first threshold can be set as needed, such as be set as 60%, then recognize
It is same target for them, to be updated in the testing result of present frame using i-th of testing result of present frame as target 1
Where target 1 storage array vector1 (less than 3 backward plus, reach 3 and just remove first, it is subsequent successively toward Forward
Position).If the coincidence of multiple testing results of present frame and some testing result (assuming that the result belongs to target 2) of former frame
Degree be more than first threshold, then select the highest testing result of registration as target 2 present frame testing result simultaneously
Where more fresh target 2 storage array vector2 (less than 3 backward plus, reach 3 and just remove first, it is subsequent past successively
Anterior displacement).
If the testing result of present frame can not find with the testing result of former frame have coincidence (i.e. all registrations are equal
Less than first threshold), then it is assumed that it is fresh target and is saved to the storage array of target N (1,2...... successively toward heel row)
In vectorN, in case follow-up use.
It should be noted that if before some the target k that is consecutively detected the not new coordinates regional of a certain frame with
Correspondence, previously stored historical information (i.e. the corresponding vector of target k) still will not be eliminated, and only continuous multiple frames are all
Do not detect corresponding coordinates regional and the target area have it is larger overlap, the target be just considered target disappear to by its
The historical information (i.e. the corresponding vector of target k) of preservation is removed.
Method for detecting human face provided in an embodiment of the present invention first passes through humanoid detection, finds the interested of image to be detected
Then region carries out Face datection again to area-of-interest, reduce the traversal range of Face datection, compared with prior art,
Humanoid detection and Face datection are combined, while being not entirely dependent on humanoid detection again, improves the detection effect of Face datection
Rate, while reducing false drop rate.
The embodiment of the present invention also provides a kind of human face detection device, as shown in Fig. 2, described device includes：
Preprocessing module 21 obtains present frame gray figure for being pre-processed to present frame；
Humanoid detection module 22, for carrying out humanoid detection to the present frame gray figure using humanoid grader；
Judgment module 23, for judging whether present frame gray figure detects humanoid region；
First area-of-interest determining module 24, for when detecting humanoid region, select the humanoid region as
The area-of-interest of Face datection；
Second area-of-interest determining module 25, the face for when not detecting humanoid region, finding present frame
The area-of-interest of detection；
Face detection module 26, for carrying out Face datection to the area-of-interest using face grader；
Preserving module 27, for orderly preserving face testing result according to registration.
Optionally, as shown in figure 3, the humanoid detection module 22 includes：
First unit for scaling 221, it is multiple dimensioned for being carried out to the present frame gray figure according to the first image scale factor
Scaling, obtains multiple scaling gray-scale maps；
First extraction unit 222, the feature for extracting the multiple scaling gray-scale map；
First detection unit 223, for detecting all of each scaling gray-scale map successively using the first detection window
Gray-scale map child window；
First judging unit 224, for judging whether each gray-scale map child window is humanoid region；
First cluster cell 225 obtains final humanoid for carrying out clusterings to all humanoid regions detected
Region.
Optionally, as shown in figure 4, the second area-of-interest determining module 25 includes：
Computing unit 251, for increasing on the basis of the position coordinates of the testing result of the former frame of the present frame of preservation
Add an offset, obtains the position coordinates of prediction result；
Expansion unit 252 obtains the region of interest of the Face datection of present frame for expanding the prediction result
Domain.
Optionally, as shown in figure 5, the face detection module 26 includes：
Second unit for scaling 261, for carrying out multiple dimensioned contracting to the area-of-interest according to the second image scale factor
It puts, obtains multiple scaling area-of-interests；
Second extraction unit 262, the feature for extracting the multiple scaling area-of-interest；
Second detection unit 263, for detecting each scaling area-of-interest successively using the second detection window
All area-of-interest child windows；
Second judgment unit 264, for judging whether each area-of-interest child window is human face region；
Second cluster cell 265 obtains final face for carrying out clustering to all human face regions detected
Region.
Optionally, as shown in fig. 6, the preserving module 27 includes：
Comparing unit 271, the coincidence of each testing result of each testing result and former frame for comparing present frame successively
Degree；
Updating unit 272, for being updated to the testing result of each target of preservation according to registration comparison result.
Human face detection device provided in an embodiment of the present invention first passes through humanoid detection, finds the interested of image to be detected
Then region carries out Face datection again to area-of-interest, reduce the traversal range of Face datection, compared with prior art,
Humanoid detection and Face datection are combined, while being not entirely dependent on humanoid detection again, improves the detection effect of Face datection
Rate, while reducing false drop rate.
One of ordinary skill in the art will appreciate that realizing all or part of flow in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, the program can be stored in a computer read/write memory medium
In, the program is when being executed, it may include such as the flow of the embodiment of above-mentioned each method.Wherein, the storage medium can be magnetic
Dish, CD, read-only memory (Read-Only Memory, ROM) or random access memory (Random Access
Memory, RAM) etc..
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, all answer by the change or replacement that can be readily occurred in
It is included within the scope of the present invention.Therefore, protection scope of the present invention should be subject to the protection scope in claims.
Claims (10)
1. a kind of method for detecting human face, which is characterized in that including：
Present frame is pre-processed, present frame gray figure is obtained；
Humanoid detection is carried out to the present frame gray figure using humanoid grader；
Judge whether the present frame gray figure detects humanoid region, when detecting humanoid region, selects the humanoid area
Otherwise area-of-interest of the domain as Face datection finds the area-of-interest of the Face datection of present frame；
Face datection is carried out to the area-of-interest using face grader；
Face testing result is orderly preserved according to registration.
2. according to the method described in claim 1, it is characterized in that, described use humanoid grader to the present frame gray figure
Carrying out humanoid detection includes：
Multiple dimensioned scaling is carried out to the present frame gray figure according to the first image scale factor, obtains multiple scaling gray-scale maps；
Extract the feature of the multiple scaling gray-scale map；
Detect all gray-scale map child windows of each scaling gray-scale map successively using the first detection window；
Judge whether each gray-scale map child window is humanoid region；
Clustering is carried out to all humanoid regions detected, obtains final humanoid region.
3. according to the method described in claim 1, it is characterized in that, the area-of-interest of the Face datection for finding present frame
Including：
Increase an offset on the basis of position coordinates of the testing result of the former frame of the present frame of preservation, is predicted
As a result position coordinates；
The prediction result is expanded, the area-of-interest of the Face datection of present frame is obtained.
4. according to the method described in claim 1, it is characterized in that, it is described using face grader to the area-of-interest into
Row Face datection includes：
Multiple dimensioned scaling is carried out to the area-of-interest according to the second image scale factor, obtains multiple scaling region of interest
Domain；
Extract the feature of the multiple scaling area-of-interest；
Detect all area-of-interest child windows of each scaling area-of-interest successively using the second detection window；
Judge whether each area-of-interest child window is human face region；
Clustering is carried out to all human face regions detected, obtains final human face region.
5. according to the method described in claim 1, it is characterized in that, described orderly preserve face testing result packet according to registration
It includes：
Compare the registration of each testing result of present frame and each testing result of former frame successively；
The testing result of each target of preservation is updated according to registration comparison result.
6. a kind of human face detection device, which is characterized in that including：
Preprocessing module obtains present frame gray figure for being pre-processed to present frame；
Humanoid detection module, for carrying out humanoid detection to the present frame gray figure using humanoid grader；
Judgment module, for judging whether present frame gray figure detects humanoid region；
First area-of-interest determining module, for when detecting humanoid region, selecting the humanoid region to be examined as face
The area-of-interest of survey；
The second area-of-interest determining module, the Face datection for when not detecting humanoid region, finding present frame
Area-of-interest；
Face detection module, for carrying out Face datection to the area-of-interest using face grader；
Preserving module, for orderly preserving face testing result according to registration.
7. device according to claim 6, which is characterized in that the humanoid detection module includes：
First unit for scaling is obtained for carrying out multiple dimensioned scaling to the present frame gray figure according to the first image scale factor
To multiple scaling gray-scale maps；
First extraction unit, the feature for extracting the multiple scaling gray-scale map；
First detection unit, all gray-scale maps for detecting each scaling gray-scale map successively using the first detection window
Window；
First judging unit, for judging whether each gray-scale map child window is humanoid region；
First cluster cell obtains final humanoid region for carrying out clustering to all humanoid regions detected.
8. device according to claim 6, which is characterized in that the second area-of-interest determining module includes：
Computing unit, for increasing by one on the basis of the position coordinates of the testing result of the former frame of the present frame of preservation partially
Shifting amount obtains the position coordinates of prediction result；
Expansion unit obtains the area-of-interest of the Face datection of present frame for expanding the prediction result.
9. device according to claim 6, which is characterized in that the face detection module includes：
Second unit for scaling is obtained for carrying out multiple dimensioned scaling to the area-of-interest according to the second image scale factor
Multiple scaling area-of-interests；
Second extraction unit, the feature for extracting the multiple scaling area-of-interest；
Second detection unit, institute's thoughts for detecting each scaling area-of-interest successively using the second detection window are emerging
Interesting region child window；
Second judgment unit, for judging whether each area-of-interest child window is human face region；
Second cluster cell obtains final human face region for carrying out clustering to all human face regions detected.
10. device according to claim 6, which is characterized in that the preserving module includes：
Comparing unit, the registration of each testing result of each testing result and former frame for comparing present frame successively；
Updating unit, for being updated to the testing result of each target of preservation according to registration comparison result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710203886.3A CN108664853B (en) | 2017-03-30 | 2017-03-30 | Face detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710203886.3A CN108664853B (en) | 2017-03-30 | 2017-03-30 | Face detection method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108664853A true CN108664853A (en) | 2018-10-16 |
| CN108664853B CN108664853B (en) | 2022-05-27 |
Family
ID=63785551
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710203886.3A Active CN108664853B (en) | 2017-03-30 | 2017-03-30 | Face detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108664853B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113242428A (en) * | 2021-04-15 | 2021-08-10 | 中南大学 | ROI (region of interest) -based post-processing acceleration method in video conference scene |
| CN114924645A (en) * | 2022-05-18 | 2022-08-19 | 上海庄生晓梦信息科技有限公司 | Interaction method and system based on gesture recognition |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101216885A (en) * | 2008-01-04 | 2008-07-09 | 中山大学 | Passerby face detection and tracing algorithm based on video |
| US20150081302A1 (en) * | 2011-05-05 | 2015-03-19 | At&T Intellectual Property I, L.P. | System and method for dynamic facial features for speaker recognition |
| CN105069813A (en) * | 2015-07-20 | 2015-11-18 | 阔地教育科技有限公司 | Stable moving target detection method and device |
| CN105354549A (en) * | 2015-11-02 | 2016-02-24 | 南京理工大学 | Rapid pedestrian detection method based on objectness estimation |
| CN105702056A (en) * | 2016-04-20 | 2016-06-22 | 李勇 | Human face identifying method |
| CN105809114A (en) * | 2016-02-29 | 2016-07-27 | 深圳市智美达科技股份有限公司 | Face detection method and apparatus |
-
2017
- 2017-03-30 CN CN201710203886.3A patent/CN108664853B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101216885A (en) * | 2008-01-04 | 2008-07-09 | 中山大学 | Passerby face detection and tracing algorithm based on video |
| US20150081302A1 (en) * | 2011-05-05 | 2015-03-19 | At&T Intellectual Property I, L.P. | System and method for dynamic facial features for speaker recognition |
| CN105069813A (en) * | 2015-07-20 | 2015-11-18 | 阔地教育科技有限公司 | Stable moving target detection method and device |
| CN105354549A (en) * | 2015-11-02 | 2016-02-24 | 南京理工大学 | Rapid pedestrian detection method based on objectness estimation |
| CN105809114A (en) * | 2016-02-29 | 2016-07-27 | 深圳市智美达科技股份有限公司 | Face detection method and apparatus |
| CN105702056A (en) * | 2016-04-20 | 2016-06-22 | 李勇 | Human face identifying method |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113242428A (en) * | 2021-04-15 | 2021-08-10 | 中南大学 | ROI (region of interest) -based post-processing acceleration method in video conference scene |
| CN113242428B (en) * | 2021-04-15 | 2024-03-15 | 中南大学 | Post-processing acceleration method based on ROI (region of interest) in video conference scene |
| CN114924645A (en) * | 2022-05-18 | 2022-08-19 | 上海庄生晓梦信息科技有限公司 | Interaction method and system based on gesture recognition |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108664853B (en) | 2022-05-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102609686B (en) | Pedestrian detection method | |
| CN107316036B (en) | Insect pest identification method based on cascade classifier | |
| CN101981582B (en) | Method and apparatus for detecting object | |
| CN101178770B (en) | Image detection method and apparatus | |
| CN109829467A (en) | Image labeling method, electronic device and non-transient computer-readable storage medium | |
| CN100561505C (en) | A kind of image detecting method and device | |
| CN107273832B (en) | License plate recognition method and system based on integral channel characteristics and convolutional neural network | |
| CN105260749B (en) | Real-time target detection method based on direction gradient binary pattern and soft cascade SVM | |
| CN103093212A (en) | Method and device for clipping facial images based on face detection and face tracking | |
| CN109191488B (en) | Target tracking system and method based on CSK and TLD fusion algorithm | |
| CN100561501C (en) | A kind of image detecting method and device | |
| CN107633226A (en) | A kind of human action Tracking Recognition method and system | |
| CN110008899B (en) | Method for extracting and classifying candidate targets of visible light remote sensing image | |
| CN105868708A (en) | Image object identifying method and apparatus | |
| Ruta et al. | Detection, tracking and recognition of traffic signs from video input | |
| CN105303163B (en) | A kind of method and detection device of target detection | |
| CN110008900A (en) | A kind of visible remote sensing image candidate target extracting method by region to target | |
| CN106023159A (en) | Disease spot image segmentation method and system for greenhouse vegetable leaf | |
| CN113378675A (en) | Face recognition method for simultaneous detection and feature extraction | |
| CN108664853A (en) | Method for detecting human face and device | |
| CN111881775B (en) | Real-time face recognition method and device | |
| CN111582057B (en) | Face verification method based on local receptive field | |
| Mitsui et al. | Object detection by joint features based on two-stage boosting | |
| CN110334703B (en) | Ship detection and identification method in day and night image | |
| CN111402185B (en) | Image detection method and device |
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 |