CN108334811A - A kind of face image processing process and device - Google Patents
A kind of face image processing process and device Download PDFInfo
- Publication number
- CN108334811A CN108334811A CN201711434909.8A CN201711434909A CN108334811A CN 108334811 A CN108334811 A CN 108334811A CN 201711434909 A CN201711434909 A CN 201711434909A CN 108334811 A CN108334811 A CN 108334811A
- Authority
- CN
- China
- Prior art keywords
- image
- face
- current
- facial image
- current 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
- 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
- G06V10/50—Extraction of image or video features by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
An embodiment of the present invention provides a kind of face image processing process and devices.In embodiments of the present invention, it detects the facial image in video image by seetaface Face datection algorithms the face images occurred in video image can be detected, avoids omitting facial image.Histograms of oriented gradients feature vector has invariance to rotation, scaling and brightness, also there is stability to factors such as visual angle change, light and noises, therefore it is influenced by outside environmental elements small, strong robustness, so, according to histograms of oriented gradients feature vector come judge the facial image tracked and current face's image whether be same personage facial image, judging nicety rate can be improved.
Description
Technical field
The present invention relates to field of computer technology, more particularly to a kind of face image processing process and device.
Background technology
It is social now, in order to people work and life safety precautions and guarantee are provided, often set at critical positions
It is equipped with monitoring camera, the monitor video at critical positions is recorded by monitoring camera, later, arrangement checks that personnel check prison
It controls the monitor video that camera is recorded and whether there is suspicious figure, for example, checking whether that there are fugitive personnel etc..
Wherein, check that personnel when the monitor video that checking monitoring camera is recorded whether there is suspicious figure, need one
Each frame video image of one frame of the frame ground in checking monitoring video, causes to check less efficient and leads to the workload for checking personnel
It is higher.
Check that efficiency and reduction check that the workload of personnel is calculated using meanshift in the prior art to improve
Method or camshift algorithms or mean value track algorithm carry out the personage in video rough tracking, are finally stored in personage and leave
When personage facial image.
However, it is found by the inventors that the prior art can cause the excalation problem of tracking person, and tracking easy tos produce
Discontinuous problem, therefore the accurate tracking difficult to realize to facial image, cannot achieve the complete acquisition to facial image.
Invention content
In order to solve the above technical problems, the embodiment of the present invention shows a kind of face image processing process and device.
In a first aspect, the embodiment of the present invention shows a kind of face image processing process, the method includes：
Current face's image in current video image is detected by seetaface Face datection algorithms；
Determine current location of the current face's image in the current video image and from the current video
Current face's image is extracted in image；
The face for judging whether tracking using particle filter track algorithm according to the current video image
Image；
If there is the facial image for using particle filter track algorithm tracking, current face's image is obtained
Human face characteristic point first direction histogram of gradients feature vector, and, obtain described in the people of facial image that is tracking
The second direction histogram of gradients feature vector of face characteristic point；
Sentenced according to the first direction histogram of gradients feature vector and the second direction histogram of gradients feature vector
The facial image tracked and current face's image whether be same personage facial image；
If the facial image tracked and the facial image that current face's image is same personage, use
Replace the position of the facial image tracked cached in the current location.
In an optional realization method, the method further includes：
If there is no the facial image for using particle filter track algorithm tracking, then in the owner cached
The facial image for same personage with current face's image is judged whether in face image；
If there is no the faces with current face's image for same personage in the face images cached
Image then caches current face's image and the caching current location；
Start in the current location to track current face's image using particle filter track algorithm.
In an optional realization method, the method further includes：
If the facial image tracked is not the facial image of same personage with current face's image,
The facial image for same personage with current face's image is judged whether in the face images cached；
If there is no the faces with current face's image for same personage in the face images cached
Image then caches current face's image and the caching current location；
Start in the current location to track current face's image using particle filter track algorithm.
In an optional realization method, the method further includes：
Judge whether the area of current face's image is more than the face of the facial image tracked cached
Product；
If the area of current face's image is more than the area of the facial image tracked cached, make
The facial image tracked described in having cached is replaced with current face's image.
In an optional realization method, the method further includes：
If not occurring working as forefathers with described in the video image of the preset quantity after the current video image
Face image belongs to the facial image of same personage, stops using the tracking of particle filter track algorithm and current face's image
Belong to the facial image of same personage；
Obtain the geography information of the facial image for belonging to same personage with current face's image cached；
By cached belong to the facial image of same personage with current face's image and cached with it is described
Geography information where current face's image belongs to the facial image of same personage stores in the database；
Delete cached belong to the facial image of same personage with current face's image and delete and cached
Belong to the position of the facial image of same personage with current face's image.
Second aspect, the embodiment of the present invention show that a kind of face image processing device, described device include：
Detection module, for detecting current face's figure in current video image by seetaface Face datection algorithms
Picture；
Determining module, for determine current location of the current face's image in the current video image and from
Current face's image is extracted in the current video image；
First judgment module, for judging whether to calculate using particle filter tracking according to the current video image
The facial image that method is tracking；
First acquisition module, for if there is the facial image for using particle filter track algorithm tracking, obtaining
The first direction histogram of gradients feature vector of the human face characteristic point of current face's image is taken, and, described in acquisition
The second direction histogram of gradients feature vector of the human face characteristic point of the facial image of tracking；
Second judgment module, for according to the first direction histogram of gradients feature vector and the second direction gradient
The facial image that is being tracked described in histogram feature vector determination and current face's image whether be same personage people
Face image；
First replacement module, if being same people for the facial image tracked and current face's image
The facial image of object replaces the position of the facial image tracked described in having cached using the current location.
In an optional realization method, described device further includes：
Third judgment module, for if there is no the facial image for using particle filter track algorithm tracking,
The facial image for same personage with current face's image is then judged whether in the face images cached；
First cache module, if for being not present and current face's image in the face images cached
For the facial image of same personage, then current face's image and the caching current location are cached；
First tracking module, it is described current for starting to track in the current location using particle filter track algorithm
Facial image.
In an optional realization method, described device further includes：
4th judgment module, if not being same for the facial image tracked and current face's image
The facial image of personage then judges whether with current face's image to be same in the face images cached
The facial image of personage；
Second cache module, if for being not present and current face's image in the face images cached
For the facial image of same personage, then current face's image and the caching current location are cached；
Second tracking module, it is described current for starting to track in the current location using particle filter track algorithm
Facial image.
In an optional realization method, described device further includes：
5th judgment module, for judge current face's image area whether be more than cached it is described with
The area of the facial image of track；
Second replacement module, if the area for current face's image is more than cached described and is tracking
The area of facial image replaces the facial image tracked described in having cached using current face's image.
In an optional realization method, described device further includes：
Stopping modular, if for not occurring in the video image of the preset quantity after the current video image
The facial image for belonging to same personage with current face's image, stop using particle filter track algorithm tracking with it is described
Current face's image belongs to the facial image of same personage；
Second acquisition module, for obtaining the facial image for belonging to same personage with current face's image cached
Geography information；
Memory module, for the facial image and for belonging to same personage with current face's image will to have been cached
The geography information storage of caching belonged to current face's image where the facial image of same personage is in the database；
Removing module, for delete cached with current face's image belong to same personage facial image and
Delete the position of the facial image for belonging to same personage with current face's image cached.
Compared with prior art, the embodiment of the present invention includes following advantages：
In embodiments of the present invention, the facial image detected in video image by seetaface Face datection algorithms can
Detected the face images occurred in video image, avoid omitting facial image.Histograms of oriented gradients is special
Sign vector has invariance to rotation, scaling and brightness, also has to factors such as visual angle change, light and noises steady
It is qualitative, therefore small, strong robustness is influenced by outside environmental elements, so, judged according to histograms of oriented gradients feature vector
The facial image that is tracking and current face's image whether be same personage facial image, judging nicety rate can be improved.
Description of the drawings
Fig. 1 is a kind of step flow chart of face image processing process embodiment of the present invention；
Fig. 2 is a kind of structure diagram of face image processing device embodiment of the present invention.
Specific implementation mode
In order to make the foregoing objectives, features and advantages of the present invention clearer and more comprehensible, below in conjunction with the accompanying drawings and specific real
Applying mode, the present invention is described in further detail.
Referring to Fig.1, a kind of step flow chart of face image processing process embodiment of the present invention is shown, it specifically can be with
Include the following steps：
In step S101, the current face in current video image is detected by seetaface Face datection algorithms and is schemed
Picture；
When recording the video of fixed position by monitoring camera, video figure can be constantly acquired by monitoring camera
Picture, whenever collecting a frame video image, it is necessary to pass through seetaface Face datection algorithms and detect collected video image
In with the presence or absence of facial image, using the facial image detected as current face's image, execute step when there are facial image
Rapid S102.
In step s 102, current location of current face's image in current video image is determined and from current video
Current face's image is extracted in image；
It wherein, can be according to the upper left corner of the rectangle frame comprising current face's image when detecting current face's image
The position of pixel, the length of long side and short side length determine the current location of current face's image.
In step s 103, according to current video image judge whether using particle filter track algorithm with
The facial image of track；
In embodiments of the present invention, other videos before current video image are being collected by monitoring camera
When image, if detecting the facial image occurred for the first time in other video images, it is determined that there is face for the first time
Position of the image in other video images, the face that then facial image occurs for the first time in caching and caching first time occurs
Position of the image in other video images, and using particle filter track algorithm in the position start tracking first time go out
Existing facial image.When the facial image of tracking leaves the acquisition range of monitoring camera, then particle filter is stopped using
Track algorithm tracks the facial image for belonging to same personage with the facial image occurred for the first time.
Using particle filter track algorithm to start in the position tracking first time occur facial image when, need
Predicted location area of the facial image that prediction occurs for the first time in next frame video image.
Multiple facial images, and other videos of different facial images may be existed simultaneously in other video images
Position in image is different, therefore, using particle filter track algorithm predict different facial images next frame video
The band of position in image is different.
If facial image is not detected in other video images before current video image, just it is not present at this time
The facial image tracked using particle filter track algorithm, current face's image in current video image is first
The facial image of secondary appearance.
In embodiments of the present invention, previous before current video image is being regarded using particle filter track algorithm
After everyone face image in frequency image tracks respectively, the facial image in previous video image can be predicted respectively
Predicted location area in current video image.
So before obtaining after the current location in current video image of facial image, need judging present bit
Setting in the predicted location area in which of previous video image facial image in current video image.
If current location is located at prediction of a certain facial image in current video image in previous video image
The facial image is then determined as the facial image tracked using particle filter track algorithm by the band of position.
If current location is not located at everyone face image in previous video image in current video image
Predicted location area, it is determined that there is no the facial images tracked using particle filter track algorithm.
If there is no the facial image for using particle filter track algorithm tracking, in step S104, caching
Current face's image and caching current location start tracking current face using particle filter track algorithm in current location
Image；
If there is no the facial image for using particle filter track algorithm tracking, then need in the institute cached
There is the facial image judged whether in facial image with current face's image is same personage；If all what is cached
There is no the facial images with current face's image for same personage in facial image, then current face's image is that occur for the first time
Facial image, therefore, it is necessary to cache current face's image and caching current location, and use particle filter track algorithm
Start to track current face's image in the position.
If there is the facial image for using particle filter track algorithm tracking, in step S105, acquisition is worked as
The first direction histogram of gradients feature vector of the human face characteristic point of preceding facial image, and, obtain the face figure tracked
The second direction histogram of gradients feature vector of the human face characteristic point of picture；
In step s 106, according to first direction histogram of gradients feature vector and second direction histogram of gradients feature to
Amount judge the facial image that is tracking and current face's image whether be same personage facial image；
In embodiments of the present invention, the human face characteristic point in current face's image can be determined using Dlib algorithms libraries
Position, obtains the human face characteristic point in facial image, for example, regression tree is trained using grey scale pixel value, then by multiple regression tree grades
Connection is a cascade classifier, and the human face characteristic point in current face's image is predicted using cascade classifier.
The human face characteristic point in facial image in the embodiment of the present invention includes 68, it is of course also possible to including more
Human face characteristic point, such as 98 or 128 etc., the embodiment of the present invention is not limited this.
Wherein, in embodiments of the present invention, human face characteristic point includes the human face characteristic point of two eyebrow outlines, two eyes
The human face characteristic point of profile, the human face characteristic point of nose profile, the human face characteristic point of face profile and the face of cheek profile
Characteristic point etc..
In embodiments of the present invention, when obtaining the human face characteristic point of facial image, the face to acquisition is generally required
Characteristic point is numbered, and the number of the different human face characteristic points in same facial image is different.
In this way, everyone face characteristic point of the current face's image obtained has respective number, and, acquisition is just
There is respective number in everyone face characteristic point of the facial image of tracking.
In current face's image and the facial image tracked, need the human face characteristic point of same number respectively
Match.
In current face's image, if big with the quantity of the human face characteristic point of the facial image successful match tracked
In predetermined threshold value, it is determined that current face's image belongs to the facial image of same personage with the facial image tracked.
In current face's image, if small with the quantity of the human face characteristic point of the facial image successful match tracked
In or equal to predetermined threshold value, it is determined that current face's image is not belonging to the face figure of same personage with the facial image tracked
Picture.
Predetermined threshold value can be the total quantity of human face characteristic point got from current face's image 70%, 80% or
90 etc., the embodiment of the present invention is not limited this.
By the number A's in the human face characteristic point of the number A in current face's image and the facial image tracked
When human face characteristic point matches, the first direction gradient histogram for calculating the human face characteristic point of the number A in current face's image is needed
Figure feature vector, and, calculate the second direction gradient histogram of the human face characteristic point of the number A in the facial image tracked
Figure feature vector, then calculate between first direction histogram of gradients feature vector and second direction histogram of gradients feature vector
Euclidean distance.
If the Euclidean distance being calculated is less than default Euclidean distance threshold value, it is determined that the number in current face's image
The human face characteristic point successful match of the human face characteristic point of A and the number A in the facial image tracked.
If the Euclidean distance being calculated is greater than or equal to default Euclidean distance threshold value, it is determined that in current face's image
Number A human face characteristic point and the number A in the facial image that is tracking the non-successful match of human face characteristic point.
Wherein, calculate current face's image in number A human face characteristic point first direction histogram of gradients feature to
Amount, including：
Centered on the human face characteristic point of the number A in current face's image, 16 pixel *, 16 pixels are chosen
First pixel region.
By the second pixel region that the first pixel region division is 16 different 4 pixel *, 4 pixels.
For any one the second pixel region, the ladder of 16 pixels in the second pixel region is calculated separately
Direction is spent, gradient direction range can be divided into 12 one's share of expenses for a joint undertaking ranges by ranging from 0~360 ° of gradient direction, ranging from per one's share of expenses for a joint undertaking
30 °, to obtain the subrange where the gradient direction of each pixel in 16 pixels, and then determine gradient side
To the quantity for the pixel being located in each subrange, to obtain the feature vector of one 12 dimension, by the feature of 12 dimensions
Vector is determined as the feature vector in the second pixel region.Other each second pixels are obtained also according to the above method
The feature vector in region.
The combination of eigenvectors in 16 the second pixel regions is obtained into the feature vector of 12*16 dimensions, and as working as
The first direction gradient eigenvector of the human face characteristic point of number A in preceding facial image.
In embodiments of the present invention, the division methods of gradient direction range are not limited to above-mentioned division methods, also can will be terraced
Degree direction scope is divided into 8 one's share of expenses for a joint undertaking ranges, and per ranging from 45 ° of one's share of expenses for a joint undertaking, gradient direction range can be also divided into 10 one's share of expenses for a joint undertaking ranges,
Per ranging from 36 ° etc. of one's share of expenses for a joint undertaking, the embodiment of the present invention is not limited this.
In embodiments of the present invention, the division methods in pixel region are not limited to above-mentioned division methods, also can be by first
Pixel region division is the second pixel region of 64 different 2 pixel *, 2 pixels, also can be by the first pixel
Point region division is the second pixel region etc. of 4 different 8 pixel *, 8 pixels, and the embodiment of the present invention is to this
It is not limited.
Secondly, the second direction histogram of gradients of the human face characteristic point of the number A in the facial image tracked is calculated
Feature vector, the first direction histogram of gradients that may refer to calculate the human face characteristic point of the number A in current face's image are special
The calculation process of vector is levied, it is not described here in detail.
It regard the human face characteristic point obtained by facial modeling algorithm as histograms of oriented gradients feature extraction
Key point so that when being matched to facial image more be directed to face features match, as eyes, nose, face, profile, under
Palestine and Israel and eyebrow etc. match accuracy higher, and matching algorithm speed is fast, to improve tracking efficiency.
If the facial image tracked and the facial image that current face's image is same personage, in step S107
In, the position of the facial image tracked cached is replaced using the current location of current face's image；
So as to which the face figure that particle filter track algorithm continues tracking from current location can be used later
Picture.
If the facial image tracked is not the facial image of same personage with current face's image, in step S108
In, cache current face's image and caching current location, and using particle filter track algorithm current location start with
Track current face's image.
If the facial image tracked is not the facial image of same personage with current face's image, need
The facial image for same personage with current face's image is judged whether in the face images of caching；If having delayed
In the face images deposited there is no with the facial image that current face's image is same personage, then current face's image is the
The facial image once occurred therefore, it is necessary to cache current face's image and caching current location, and uses particle filter
Track algorithm starts to track current face's image in the position.
In embodiments of the present invention, the facial image detected in video image by seetaface Face datection algorithms can
Detected the face images occurred in video image, avoid omitting facial image.Histograms of oriented gradients is special
Sign vector has invariance to rotation, scaling and brightness, also has to factors such as visual angle change, light and noises steady
It is qualitative, therefore small, strong robustness is influenced by outside environmental elements, so, judged according to histograms of oriented gradients feature vector
The facial image that is tracking and current face's image whether be same personage facial image, judging nicety rate can be improved.
In embodiments of the present invention, if what the facial image of a certain personage occurred in the acquisition range of monitoring camera
Time is longer, then the facial image of the personage appears in the multi-frame video image that monitoring camera acquires, if by each
The facial image of the personage in frame video image caches, then can occupy more spatial cache, therefore, in order to save caching
Which space and do not influence to check in the acquisition range of monitoring camera personage occurred after checking personnel, without will be each
The facial image of the personage in frame video image caches, and need to only cache the face of the personage in a wherein frame video image
Image.
Further, in order to enable checking that personnel check whom occurred in the acquisition range of monitoring camera later
When object, the facial detail of personage can be more clearly seen in an alternative embodiment of the invention, in the face for caching the personage
When image, the maximum facial image of the area for caching the personage is needed.
Wherein, in embodiments of the present invention, it can be determined that whether the area of current face's image, which is more than, has cached
The area of the facial image of tracking, if the area of current face's image is more than the face of the facial image tracked cached
Product can then use current face's image to replace the facial image tracked cached.
Wherein it is possible to the area of current face's image and the area of the facial image tracked cached are subtracted each other,
Obtain area difference；Judge whether the area difference is more than 0；If the area difference is more than 0, whether the area difference is judged
More than preset area threshold value, preset area threshold value is more than 0；If the area difference is more than preset area threshold value, can use
Current face's image replaces the facial image tracked cached.If the area difference is less than or equal to preset area threshold
Value then needs to calculate the human face characteristic point of the leftmost side in current face's image and first between the human face characteristic point of the rightmost side
Distance, and calculate the leftmost side in the facial image tracked that has cached human face characteristic point and the rightmost side face it is special
Second distance between sign point；Then judge whether the first distance is more than second distance；If the first distance is more than second distance,
Can then current face's image be used to replace the facial image tracked cached.
In still another embodiment of the process, if in the video image of the preset quantity after current video image not
Occur belonging to the facial image of same personage with current face's image, then illustrate personage corresponding to current face's image from
The acquisition range for opening monitoring camera, to stop using the tracking of particle filter track algorithm to belong to same with current face's image
The facial image of one personage obtains the geography information of the facial image for belonging to same personage with current face's image cached；
By cached belong to the facial image of same personage with current face's image and what is cached belongs to current face's image
Geography information storage where the facial image of same personage is in the database；What deletion had cached belongs to current face's image
The facial image of same personage and the position for deleting the facial image for belonging to same personage with current face's image cached.
It should be noted that for embodiment of the method, for simple description, therefore it is all expressed as a series of action group
It closes, but those skilled in the art should understand that, the embodiment of the present invention is not limited by the described action sequence, because according to
According to the embodiment of the present invention, certain steps can be performed in other orders or simultaneously.Secondly, those skilled in the art also should
Know, embodiment described in this description belongs to preferred embodiment, and the involved action not necessarily present invention is implemented
Necessary to example.
With reference to Fig. 2, shows a kind of structure diagram of face image processing device embodiment of the present invention, can specifically include
Following module：
Detection module 11, for detecting the current face in current video image by seetaface Face datection algorithms
Image；
Determining module 12, for determine current location of the current face's image in the current video image and
Current face's image is extracted from the current video image；
First judgment module 13, for judging whether to track using particle filter according to the current video image
The facial image that algorithm is tracking；
First acquisition module 14, for if there is the facial image for using particle filter track algorithm tracking,
The first direction histogram of gradients feature vector of the human face characteristic point of current face's image is obtained, and, acquisition is described just
In the second direction histogram of gradients feature vector of the human face characteristic point of the facial image of tracking；
Second judgment module 15, for according to the first direction histogram of gradients feature vector and second direction ladder
Whether the facial image and current face's image tracked described in degree histogram feature vector determination is same personage
Facial image；
First replacement module 16, if being same for the facial image tracked and current face's image
The facial image of personage replaces the position of the facial image tracked described in having cached using the current location.
In an optional realization method, described device further includes：
Third judgment module, for if there is no the facial image for using particle filter track algorithm tracking,
The facial image for same personage with current face's image is then judged whether in the face images cached；
First cache module, if for being not present and current face's image in the face images cached
For the facial image of same personage, then current face's image and the caching current location are cached；
First tracking module, it is described current for starting to track in the current location using particle filter track algorithm
Facial image.
In an optional realization method, described device further includes：
4th judgment module, if not being same for the facial image tracked and current face's image
The facial image of personage then judges whether with current face's image to be same in the face images cached
The facial image of personage；
Second cache module, if for being not present and current face's image in the face images cached
For the facial image of same personage, then current face's image and the caching current location are cached；
Second tracking module, it is described current for starting to track in the current location using particle filter track algorithm
Facial image.
In an optional realization method, described device further includes：
5th judgment module, for judge current face's image area whether be more than cached it is described with
The area of the facial image of track；
Second replacement module, if the area for current face's image is more than cached described and is tracking
The area of facial image replaces the facial image tracked described in having cached using current face's image.
In an optional realization method, described device further includes：
Stopping modular, if for not occurring in the video image of the preset quantity after the current video image
The facial image for belonging to same personage with current face's image, stop using particle filter track algorithm tracking with it is described
Current face's image belongs to the facial image of same personage；
Second acquisition module, for obtaining the facial image for belonging to same personage with current face's image cached
Geography information；
Memory module, for the facial image and for belonging to same personage with current face's image will to have been cached
The geography information storage of caching belonged to current face's image where the facial image of same personage is in the database；
Removing module, for delete cached with current face's image belong to same personage facial image and
Delete the position of the facial image for belonging to same personage with current face's image cached.
In embodiments of the present invention, the facial image detected in video image by seetaface Face datection algorithms can
Detected the face images occurred in video image, avoid omitting facial image.Histograms of oriented gradients is special
Sign vector has invariance to rotation, scaling and brightness, also has to factors such as visual angle change, light and noises steady
It is qualitative, therefore small, strong robustness is influenced by outside environmental elements, so, judged according to histograms of oriented gradients feature vector
The facial image that is tracking and current face's image whether be same personage facial image, judging nicety rate can be improved.
For device embodiments, since it is basically similar to the method embodiment, so fairly simple, the correlation of description
Place illustrates referring to the part of embodiment of the method.
Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with
The difference of other embodiment, the same or similar parts between the embodiments can be referred to each other.
It should be understood by those skilled in the art that, the embodiment of the embodiment of the present invention can be provided as method, apparatus or calculate
Machine program product.Therefore, the embodiment of the present invention can be used complete hardware embodiment, complete software embodiment or combine software and
The form of the embodiment of hardware aspect.Moreover, the embodiment of the present invention can be used one or more wherein include computer can
With in the computer-usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) of program code
The form of the computer program product of implementation.
The embodiment of the present invention be with reference to according to the method for the embodiment of the present invention, terminal device (system) and computer program
The flowchart and/or the block diagram of product describes.It should be understood that flowchart and/or the block diagram can be realized by computer program instructions
In each flow and/or block and flowchart and/or the block diagram in flow and/or box combination.These can be provided
Computer program instructions are set to all-purpose computer, special purpose computer, Embedded Processor or other programmable data processing terminals
Standby processor is to generate a machine so that is held by the processor of computer or other programmable data processing terminal equipments
Capable instruction generates for realizing in one flow of flow chart or multiple flows and/or one box of block diagram or multiple boxes
The device of specified function.
These computer program instructions, which can be also buffered in, can guide computer or other programmable data processing terminal equipments
In computer-readable buffer operate in a specific manner so that the instruction being buffered in the computer-readable buffer generates packet
The manufacture of command device is included, which realizes in one flow of flow chart or multiple flows and/or one side of block diagram
The function of being specified in frame or multiple boxes.
These computer program instructions can be also loaded into computer or other programmable data processing terminal equipments so that
Series of operation steps are executed on computer or other programmable terminal equipments to generate computer implemented processing, thus
The instruction executed on computer or other programmable terminal equipments is provided for realizing in one flow of flow chart or multiple flows
And/or in one box of block diagram or multiple boxes specify function the step of.
Although the preferred embodiment of the embodiment of the present invention has been described, once a person skilled in the art knows bases
This creative concept, then additional changes and modifications can be made to these embodiments.So the following claims are intended to be interpreted as
Including preferred embodiment and fall into all change and modification of range of embodiment of the invention.
Finally, it is to be noted that, herein, relational terms such as first and second and the like be used merely to by
One entity or operation are distinguished with another entity or operation, without necessarily requiring or implying these entities or operation
Between there are any actual relationship or orders.Moreover, the terms "include", "comprise" or its any other variant meaning
Covering non-exclusive inclusion, so that process, method, article or terminal device including a series of elements not only wrap
Those elements are included, but also include other elements that are not explicitly listed, or further include for this process, method, article
Or the element that terminal device is intrinsic.In the absence of more restrictions, being wanted by what sentence "including a ..." limited
Element, it is not excluded that there is also other identical elements in process, method, article or the terminal device including the element.
Above to a kind of face image processing process provided by the present invention and device, it is described in detail, herein
Applying specific case, principle and implementation of the present invention are described, and the explanation of above example is only intended to help
Understand the method and its core concept of the present invention；Meanwhile for those of ordinary skill in the art, according to the thought of the present invention,
There will be changes in the specific implementation manner and application range, in conclusion the content of the present specification should not be construed as to this
The limitation of invention.
Claims (10)
1. a kind of face image processing process, which is characterized in that the method includes：
Current face's image in current video image is detected by seetaface Face datection algorithms；
Determine current location of the current face's image in the current video image and from the current video image
In extract current face's image；
The facial image for judging whether tracking using particle filter track algorithm according to the current video image；
If there is the facial image for using particle filter track algorithm tracking, the people of current face's image is obtained
The first direction histogram of gradients feature vector of face characteristic point, and, the face of the facial image tracked described in acquisition is special
Levy the second direction histogram of gradients feature vector of point；
Judge institute according to the first direction histogram of gradients feature vector and the second direction histogram of gradients feature vector
State the facial image that is tracking and current face's image whether be same personage facial image；
If the facial image tracked and the facial image that current face's image is same personage, using described
Replace the position of the facial image tracked cached in current location.
2. according to the method described in claim 1, it is characterized in that, the method further includes：
If there is no the facial image for using particle filter track algorithm tracking, then in all face figures cached
The facial image for same personage with current face's image is judged whether as in；
If the facial image with current face's image for same personage is not present in the face images cached,
Then cache current face's image and the caching current location；
Start in the current location to track current face's image using particle filter track algorithm.
3. method according to claim 1 or 2, which is characterized in that the method further includes：
If the facial image tracked is not the facial image of same personage with current face's image,
The facial image for same personage with current face's image is judged whether in the face images of caching；
If the facial image with current face's image for same personage is not present in the face images cached,
Then cache current face's image and the caching current location；
Start in the current location to track current face's image using particle filter track algorithm.
4. according to the method described in claim 1, it is characterized in that, the method further includes：
Judge whether the area of current face's image is more than the area of the facial image tracked cached；
If the area of current face's image is more than the area of the facial image tracked cached, institute is used
It states current face's image and replaces the facial image tracked cached.
5. according to the method described in claim 1, it is characterized in that, the method further includes：
If not occurring scheming with the current face in the video image of the preset quantity after the current video image
Facial image as belonging to same personage stops using the tracking of particle filter track algorithm to belong to current face's image
The facial image of same personage；
Obtain the geography information of the facial image for belonging to same personage with current face's image cached；
By cached belong to the facial image of same personage with current face's image and cached with it is described current
Geography information where facial image belongs to the facial image of same personage stores in the database；
Delete cached with current face's image belong to same personage facial image and delete cached with institute
State the position that current face's image belongs to the facial image of same personage.
6. a kind of face image processing device, which is characterized in that described device includes：
Detection module, for detecting current face's image in current video image by seetaface Face datection algorithms；
Determining module, for determining current location of the current face's image in the current video image and from described
Current face's image is extracted in current video image；
First judgment module, for being judged whether using particle filter track algorithm just according to the current video image
In the facial image of tracking；
First acquisition module, for if there is the facial image for using particle filter track algorithm tracking, obtaining institute
The first direction histogram of gradients feature vector of the human face characteristic point of current face's image is stated, and, it is being tracked described in acquisition
Facial image human face characteristic point second direction histogram of gradients feature vector；
Second judgment module, for according to the first direction histogram of gradients feature vector and the second direction gradient histogram
Figure feature vector judge the facial image tracked and current face's image whether be same personage face figure
Picture；
First replacement module, if being same personage's for the facial image tracked and current face's image
Facial image replaces the position of the facial image tracked described in having cached using the current location.
7. device according to claim 6, which is characterized in that described device further includes：
Third judgment module, for if there is no the facial image for using particle filter track algorithm tracking, then existing
The facial image for same personage with current face's image is judged whether in the face images cached；
First cache module, if for there is no be same with current face's image in the face images cached
The facial image of one personage then caches current face's image and the caching current location；
First tracking module tracks the current face for starting in the current location using particle filter track algorithm
Image.
8. the device described according to claim 6 or 7, which is characterized in that described device further includes：
4th judgment module, if not being same personage for the facial image tracked and current face's image
Facial image, then judged whether in the face images cached with current face's image be same personage
Facial image；
Second cache module, if for there is no be same with current face's image in the face images cached
The facial image of one personage then caches current face's image and the caching current location；
Second tracking module tracks the current face for starting in the current location using particle filter track algorithm
Image.
9. device according to claim 6, which is characterized in that described device further includes：
5th judgment module is being tracked for judging whether the area of current face's image is more than cached described
The area of facial image；
Second replacement module, if the area for current face's image is more than the face tracked cached
The area of image replaces the facial image tracked described in having cached using current face's image.
10. device according to claim 6, which is characterized in that described device further includes：
Stopping modular, if for not occurring in the video image of the preset quantity after the current video image and institute
State the facial image that current face's image belongs to same personage, stop using the tracking of particle filter track algorithm with it is described current
Facial image belongs to the facial image of same personage；
Second acquisition module, the ground for obtaining the facial image for belonging to same personage with current face's image cached
Manage information；
Memory module belongs to the facial image of same personage with current face's image and has cached for will cache
The geography information storage belonged to current face's image where the facial image of same personage in the database；
Removing module, for deleting facial image and the deletion for belonging to same personage with current face's image cached
What is cached belongs to the position of the facial image of same personage with current face's image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711434909.8A CN108334811B (en) | 2017-12-26 | 2017-12-26 | Face image processing method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711434909.8A CN108334811B (en) | 2017-12-26 | 2017-12-26 | Face image processing method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108334811A true CN108334811A (en) | 2018-07-27 |
| CN108334811B CN108334811B (en) | 2021-06-04 |
Family
ID=62923707
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711434909.8A Active CN108334811B (en) | 2017-12-26 | 2017-12-26 | Face image processing method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108334811B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109977745A (en) * | 2018-12-25 | 2019-07-05 | 深圳云天励飞技术有限公司 | Face image processing process and relevant apparatus |
| CN111652070A (en) * | 2020-05-07 | 2020-09-11 | 南京航空航天大学 | Face sequence collaborative recognition method based on surveillance video |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101339608A (en) * | 2008-08-15 | 2009-01-07 | 北京中星微电子有限公司 | Object tracking method and system based on detection |
| US20120057751A1 (en) * | 2009-09-24 | 2012-03-08 | Liu ke-yan | Particle Tracking Methods |
| CN103116756A (en) * | 2013-01-23 | 2013-05-22 | 北京工商大学 | Face detecting and tracking method and device |
| CN103985137A (en) * | 2014-04-25 | 2014-08-13 | 北京大学深圳研究院 | Moving object tracking method and system applied to human-computer interaction |
| CN104036523A (en) * | 2014-06-18 | 2014-09-10 | 哈尔滨工程大学 | Improved mean shift target tracking method based on surf features |
| CN105354902A (en) * | 2015-11-10 | 2016-02-24 | 深圳市商汤科技有限公司 | Security management method and system based on face identification |
| WO2017080399A1 (en) * | 2015-11-12 | 2017-05-18 | 阿里巴巴集团控股有限公司 | Method and device for tracking location of human face, and electronic equipment |
| CN107066958A (en) * | 2017-03-29 | 2017-08-18 | 南京邮电大学 | A kind of face identification method based on HOG features and SVM multi-categorizers |
| CN107077738A (en) * | 2014-11-12 | 2017-08-18 | 高通股份有限公司 | System and method for tracking object |
-
2017
- 2017-12-26 CN CN201711434909.8A patent/CN108334811B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101339608A (en) * | 2008-08-15 | 2009-01-07 | 北京中星微电子有限公司 | Object tracking method and system based on detection |
| US20120057751A1 (en) * | 2009-09-24 | 2012-03-08 | Liu ke-yan | Particle Tracking Methods |
| CN103116756A (en) * | 2013-01-23 | 2013-05-22 | 北京工商大学 | Face detecting and tracking method and device |
| CN103985137A (en) * | 2014-04-25 | 2014-08-13 | 北京大学深圳研究院 | Moving object tracking method and system applied to human-computer interaction |
| CN104036523A (en) * | 2014-06-18 | 2014-09-10 | 哈尔滨工程大学 | Improved mean shift target tracking method based on surf features |
| CN107077738A (en) * | 2014-11-12 | 2017-08-18 | 高通股份有限公司 | System and method for tracking object |
| CN105354902A (en) * | 2015-11-10 | 2016-02-24 | 深圳市商汤科技有限公司 | Security management method and system based on face identification |
| WO2017080399A1 (en) * | 2015-11-12 | 2017-05-18 | 阿里巴巴集团控股有限公司 | Method and device for tracking location of human face, and electronic equipment |
| CN107066958A (en) * | 2017-03-29 | 2017-08-18 | 南京邮电大学 | A kind of face identification method based on HOG features and SVM multi-categorizers |
Non-Patent Citations (6)
| Title |
|---|
| YUNJI ZHAO ET AL.: "Model update particle filter for multiple objects detection and tracking", 《2011 INTERNATIONAL SYMPOSIUM ON INTELLIGENCE INFORMATION PROCESSING AND TRUSTED COMPUTING》 * |
| 李俊彦 等: "基于HOG和颜色特征的粒子滤波行人跟踪算法的研究", 《网络安全技术与应用》 * |
| 胡一帆 等: "基于视频监控的人脸检测跟踪识别***研究", 《计算机工程与应用》 * |
| 苏松志 等: "《行人检测：理论与实践》", 31 March 2016 * |
| 颜志国 等: "《多摄像机协同关注目标检测跟踪技术》", 30 June 2017 * |
| 魏武 等: "基于AdaBoost和RVM的实时多目标跟踪", 《计算机工程与设计》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109977745A (en) * | 2018-12-25 | 2019-07-05 | 深圳云天励飞技术有限公司 | Face image processing process and relevant apparatus |
| CN109977745B (en) * | 2018-12-25 | 2021-09-14 | 深圳云天励飞技术有限公司 | Face image processing method and related device |
| CN111652070A (en) * | 2020-05-07 | 2020-09-11 | 南京航空航天大学 | Face sequence collaborative recognition method based on surveillance video |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108334811B (en) | 2021-06-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR102122348B1 (en) | Method and device for face in-vivo detection | |
| US8208733B2 (en) | Method, medium, and apparatus with estimation of background change | |
| JP6525545B2 (en) | INFORMATION PROCESSING APPARATUS, INFORMATION PROCESSING METHOD, AND COMPUTER PROGRAM | |
| CN106937114B (en) | Method and device for detecting video scene switching | |
| US9953240B2 (en) | Image processing system, image processing method, and recording medium for detecting a static object | |
| US10373015B2 (en) | System and method of detecting moving objects | |
| EP1914994A1 (en) | Detection of gradual transitions in video sequences | |
| CN109598706A (en) | A kind of camera lens occlusion detection method and system | |
| CN112016353A (en) | Method and device for carrying out identity recognition on face image based on video | |
| US20110069155A1 (en) | Apparatus and method for detecting motion | |
| CN109308704B (en) | Background eliminating method, device, computer equipment and storage medium | |
| CN112598677B (en) | Method and apparatus for image analysis | |
| CN113012383A (en) | Fire detection alarm method, related system, related equipment and storage medium | |
| CN105574498A (en) | Face recognition system and recognition method based on customs security check | |
| CN113269127A (en) | Face recognition and pedestrian re-recognition monitoring method and system for real-time automatic database building | |
| CN108334811A (en) | A kind of face image processing process and device | |
| Lin et al. | Real-time active tampering detection of surveillance camera and implementation on digital signal processor | |
| CN109697417A (en) | A kind of production management system for pitch-controlled system cabinet | |
| Ilao et al. | Crowd estimation using region-specific HOG With SVM | |
| KR100920937B1 (en) | Apparatus and method for detecting motion, and storing video within security system | |
| Zhu et al. | Detection and recognition of abnormal running behavior in surveillance video | |
| US10147199B2 (en) | Method and apparatus for determining an orientation of a video | |
| KR101908938B1 (en) | Detection of foreground object using background model-based temporal minimum filtering and log histogram | |
| CN111191484A (en) | Method and device for recognizing human speaking in video image | |
| JP7118679B2 (en) | VIDEO RECORDING DEVICE, VIDEO RECORDING METHOD AND PROGRAM |
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 |