CN109146913A - A kind of face tracking method and device - Google Patents
A kind of face tracking method and device Download PDFInfo
- Publication number
- CN109146913A CN109146913A CN201810871087.8A CN201810871087A CN109146913A CN 109146913 A CN109146913 A CN 109146913A CN 201810871087 A CN201810871087 A CN 201810871087A CN 109146913 A CN109146913 A CN 109146913A
- Authority
- CN
- China
- Prior art keywords
- facial
- face
- facial characteristics
- size
- recognized
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/207—Analysis of motion for motion estimation over a hierarchy of resolutions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
-
- 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
- G06V40/171—Local features and components; Facial parts ; Occluding parts, e.g. glasses; Geometrical relationships
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30196—Human being; Person
- G06T2207/30201—Face
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention provides a kind of face tracking method and devices, it include: that at least one facial characteristics of tracked object is stored in advance corresponding to prestore size, at least two images to be recognized are acquired, for each images to be recognized, identify at least one face facial area in images to be recognized;It is directed to everyone face facial area, at least one facial characteristics in face facial area is determined, determines the coordinate value of the corresponding characteristic point of each facial characteristics respectively;The corresponding characteristic size of facial characteristics is determined according to corresponding coordinate value for each facial characteristics determined;According to characteristic size and size is prestored, target person face facial area corresponding with tracked object is determined from the face facial area that images to be recognized carries;Determine the location of target person face facial area in images to be recognized;According to the location of target person face facial area, the motion profile of tracked object is determined.This programme can realize face tracking based on face at any angle.
Description
Technical field
The present invention relates to technical field of face recognition, in particular to a kind of face tracking method and device.
Background technique
With the fast development of information technology and the extensive use of Video Supervision Technique, face tracking technology is supervised in security protection
The fields such as control, precision marketing, intelligent entrance guard, intelligent building, human-computer interaction are widely used.
Currently, face tracking technology needs to detect the front face of tracked object, it could be by collected positive dough figurine
Face alignment in face and each image of storage identifies the motion profile for being tracked object, then determining tracked object.
As can be seen from the above description, the face tracking of the prior art is realized based on front face, lead to face tracking
The scope of application it is small.
Summary of the invention
The embodiment of the invention provides a kind of face tracking method and devices, can realize people based on face at any angle
Face tracking.
In a first aspect, tracked object is stored in advance at least the embodiment of the invention provides a kind of face tracking method
One facial characteristics is corresponding to prestore size, further includes:
The images to be recognized of at least two carrier's face facial areas is acquired, and is directed to each described images to be recognized,
It executes:
C1: at least one carried in the images to be recognized face facial area is identified;
C2: it is directed to each described face facial area, determines at least one for including in the face facial area
The facial characteristics, and the coordinate of at least one corresponding characteristic point of each determined facial characteristics is obtained respectively
Value;
C3: institute is determined according to the corresponding coordinate value of acquisition for each the described facial characteristics determined
State the corresponding characteristic size of facial characteristics;
C4: according to the corresponding characteristic size of each described facial characteristics determined and it is described prestore size,
Target corresponding with the tracked object is determined from each face facial area that the images to be recognized carries
Face facial area;
C5: the location of target person face facial area described in the images to be recognized is determined;
According to the location of target person face facial area described at least two images to be recognized, the quilt is determined
The motion profile of tracking object.
Preferably, the C4, comprising:
The corresponding object to be identified of each face facial area and the tracked object are determined according to following formula
Matching value:
Wherein, s characterizes the matching value, αiCharacterize preset threshold value, βiCharacterize the corresponding power of i-th of facial characteristics
Weight values, yiCharacterize that i-th of facial characteristics is corresponding described to prestore size, tiCharacterize the corresponding institute of i-th of facial characteristics
State characteristic size, wherein prestore size described in the threshold value is corresponding with i-th of facial characteristics and the characteristic size is poor
It is worth corresponding;
From each matching value determined, it is preset to determine whether the smallest matching value of numerical value is less than or equal to
Error amount;
If so, the corresponding face facial area of the determining matching value the smallest with numerical value is target person face
Portion region.
Preferably, before the C2, further comprise:
The corresponding characteristic sequence of each facial characteristics is set;
At least one the described facial characteristics for including in the determination face facial area, and determined respectively
At least one corresponding characteristic point of each described facial characteristics coordinate value, comprising:
Using first facial characteristics in the characteristic sequence as current face feature, execute:
D0: determining with the presence or absence of the current face feature in the face facial area, if so, D1 is executed, otherwise,
Execute D2;
D1: acquiring the coordinate value of the corresponding each characteristic point of the current face feature, executes D3;
D2: the coordinate value of the corresponding each characteristic point of the current face feature is recorded, wherein the seat of record
The abscissa and ordinate of scale value are 0, execute D3;
D3: using next facial characteristics of the current face feature as current face feature, D0 is returned to.
Preferably, before the C3, further comprise:
It is put in order according to preset, arranges each described facial characteristics and each described facial characteristics is corresponding
Each characteristic point the coordinate value;
Using the corresponding coordinate value of the every two adjacent facial characteristics after blank character partitioned arrangement, and record
The corresponding relationship of the blank character and two adjacent facial characteristics;
For facial characteristics described in each, execute:
S0: according to the corresponding relationship, determine whether the quantity of the corresponding blank character of the facial characteristics is 2, such as
Fruit is to execute S1, otherwise, executes S2;
S1: each coordinate value between two blank characters corresponding with the facial characteristics is extracted;
S2: determining whether the blank character corresponding with the facial characteristics is first blank character, if so, executing
Otherwise S3 executes S4;
S3: each coordinate value before the corresponding blank character of the facial characteristics is extracted;
S4: each coordinate value after the corresponding blank character of the facial characteristics is extracted;
The C3, comprising:
For facial characteristics described in each, the Euclidean between the corresponding each coordinate value of the facial characteristics is calculated
Distance, and using the Euclidean distance as the corresponding characteristic size of the facial characteristics.
Preferably, the facial characteristics, comprising: at least one of facial contour, eyes, mouth and nose are multiple,
Wherein, the eyes, comprising: left eye and right eye.
Preferably, when the facial characteristics includes the facial contour, the face wheel for prestoring characterization of size and prestoring
Wide perimeter, the characteristic size characterize the facial contour perimeter in the face facial area;
When the facial characteristics includes the eyes, the eye widths for prestoring characterization of size and prestoring, the feature
Eye widths in face facial area described in characterization of size;
When the facial characteristics includes the mouth, the mouth width for prestoring characterization of size and prestoring, the feature
Mouth width in face facial area described in characterization of size;
When the facial characteristics includes the nose, the nose width for prestoring characterization of size and prestoring, the feature
Nose width in face facial area described in characterization of size;
It is described to prestore the nose and two that characterization of size prestores when the facial characteristics includes the nose and the mouth
The perimeter for the geometric figure that the side corners of the mouth is formed, the characteristic size characterize nose and the two sides corners of the mouth in the face facial area
The perimeter of the geometric figure of formation;
When the facial characteristics includes the eyes, the nose and the mouth, the characterization of size that prestores is prestored
Eye center and the perimeter of geometric figure that is centrally formed of nose and mouth, the characteristic size characterize the face facial area
The perimeter for the geometric figure that eye center and nose and mouth in domain are centrally formed.
Preferably, the position locating for the target person face facial area according at least two images to be recognized
It sets, before the motion profile for determining the tracked object, further comprises
Determine the corresponding acquisition time of each described images to be recognized;
Described the location of target person face facial area according at least two images to be recognized, determines institute
State the motion profile of tracked object, comprising:
According to the corresponding acquisition time of the images to be recognized where the target person face facial area, institute is arranged
Target facial image is stated in the location of described images to be recognized, forms the motion profile of the tracked object.
Second aspect, the embodiment of the invention provides a kind of face tracking devices, comprising:
Administrative unit, at least one facial characteristics for tracked object to be stored in advance is corresponding to prestore size;
Acquisition unit for acquiring the images to be recognized of at least two carrier's face facial areas, and is directed to each institute
Images to be recognized is stated, executes: identifying at least one carried in the images to be recognized face facial area;It is directed to every
One face facial area determines at least one the described facial characteristics for including in the face facial area, and respectively
Obtain the coordinate value of at least one corresponding characteristic point of each determined facial characteristics;
Processing unit, each described facial characteristics for being determined for the acquisition unit, according to the phase of acquisition
The corresponding coordinate value determines the corresponding characteristic size of the facial characteristics;It is special according to each the described face determined
It levies the corresponding characteristic size and the described of administrative unit storage prestores size, carried from the images to be recognized each
Target person face facial area corresponding with the tracked object is determined in the face facial area;It determines described to be identified
The location of facial area of target person face described in image;
Tracking cell, target person described at least two images to be recognized for being determined according to the processing unit
The location of face facial area determines the motion profile of the tracked object.
Preferably, the processing unit, for according to following formula determine each face facial area it is corresponding to
Identify the matching value of object and the tracked object;It determines in the matching value, whether the smallest matching value of numerical value is small
In equal to preset matching value;
If so, the corresponding face facial area of the determining matching value the smallest with numerical value is target person face
Portion region;
Wherein, s characterizes the matching value, αiCharacterize preset threshold value, βiCharacterize the corresponding power of i-th of facial characteristics
Weight values, yiCharacterize that i-th of facial characteristics is corresponding described to prestore size, tiCharacterize the corresponding institute of i-th of facial characteristics
State characteristic size, wherein prestore size described in the threshold value is corresponding with i-th of facial characteristics and the characteristic size is poor
It is worth corresponding.
Preferably, administrative unit is further used for that the corresponding characteristic sequence of each facial characteristics is arranged;
The acquisition unit, for holding using first facial characteristics in the characteristic sequence as current face feature
Row:
D0: determining with the presence or absence of the current face feature in the face facial area, if so, D1 is executed, otherwise,
Execute D2;
D1: acquiring the coordinate value of the corresponding each characteristic point of the current face feature, executes D3;
D2: the coordinate value of the corresponding each characteristic point of the current face feature is recorded, wherein the seat of record
The abscissa and ordinate of scale value are 0, execute D3;
D3: using next facial characteristics of the current face feature as current face feature, D0 is returned to.
Preferably, the acquisition unit is further used for being put in order according to preset, and it is special to arrange each described face
The coordinate value of sign and the corresponding each characteristic point of each described facial characteristics;Utilize blank character partitioned arrangement
The corresponding coordinate value of the every two adjacent facial characteristics afterwards, and record the blank character with it is two adjacent described
The corresponding relationship of facial characteristics;For facial characteristics described in each, execute:
S0: according to the corresponding relationship, determine whether the quantity of the corresponding blank character of the facial characteristics is 2, such as
Fruit is to execute S1, otherwise, executes S2;
S1: each coordinate value between two blank characters corresponding with the facial characteristics is extracted;
S2: determining whether the blank character corresponding with the facial characteristics is first blank character, if so, executing
Otherwise S3 executes S4;
S3: each coordinate value before the corresponding blank character of the facial characteristics is extracted;
S4: each coordinate value after the corresponding blank character of the facial characteristics is extracted;
The processing unit calculates the corresponding each institute of the facial characteristics for being directed to each described facial characteristics
The Euclidean distance between coordinate value is stated, and using the Euclidean distance as the corresponding characteristic size of the facial characteristics.
Preferably, the facial characteristics, comprising: at least one of facial contour, eyes, mouth and nose are multiple,
Wherein, the eyes, comprising: left eye and right eye.
Preferably, when the facial characteristics includes the facial contour, the face wheel for prestoring characterization of size and prestoring
Wide perimeter, the characteristic size characterize the facial contour perimeter in the face facial area;
When the facial characteristics includes the eyes, the eye widths for prestoring characterization of size and prestoring, the feature
Eye widths in face facial area described in characterization of size;
When the facial characteristics includes the mouth, the mouth width for prestoring characterization of size and prestoring, the feature
Mouth width in face facial area described in characterization of size;
When the facial characteristics includes the nose, the nose width for prestoring characterization of size and prestoring, the feature
Nose width in face facial area described in characterization of size;
It is described to prestore the nose and two that characterization of size prestores when the facial characteristics includes the nose and the mouth
The perimeter for the geometric figure that the side corners of the mouth is formed, the characteristic size characterize nose and the two sides corners of the mouth in the face facial area
The perimeter of the geometric figure of formation;
When the facial characteristics includes the eyes, the nose and the mouth, the characterization of size that prestores is prestored
Eye center and the perimeter of geometric figure that is centrally formed of nose and mouth, the characteristic size characterize the face facial area
The perimeter for the geometric figure that eye center and nose and mouth in domain are centrally formed.
Preferably, the administrative unit is further used for determining the corresponding acquisition time of each described images to be recognized;
The tracking cell, for according to the corresponding institute of the images to be recognized where the target person face facial area
Acquisition time is stated, the target facial image is arranged in the location of described images to be recognized, forms the tracked object
Motion profile.
In embodiments of the present invention, at least one face facial area in collected images to be recognized is identified, and
The face facial area identified can be face at any angle, special by the face for determining that everyone face facial area includes
The coordinate value of corresponding characteristic point is levied, that is, can determine the corresponding characteristic size of each facial characteristics, according to each facial characteristics pair
The characteristic size answered and it is corresponding prestore size, can determine to be tracked the corresponding target person of object from images to be recognized
Face facial area can determine tracked object according to position of the target person face facial area in corresponding images to be recognized
Motion profile.To sum up, based on face facial area corresponding to face at any angle, that is, it can determine tracked object
Motion profile, realize face tracking, without based on front face realize, so as to increase the applicable model of face tracking
It encloses.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is the present invention
Some embodiments for those of ordinary skill in the art without creative efforts, can also basis
These attached drawings obtain other attached drawings.
Fig. 1 is a kind of flow chart for face tracking method that one embodiment of the invention provides;
Fig. 2 is the flow chart for another face tracking method that one embodiment of the invention provides;
Fig. 3 is a kind of structural schematic diagram for face tracking device that one embodiment of the invention provides.
Specific embodiment
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
A part of the embodiment of the present invention, instead of all the embodiments, based on the embodiments of the present invention, those of ordinary skill in the art
Every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
As shown in Figure 1, the embodiment of the invention provides a kind of face tracking methods, comprising:
Step 101: at least one facial characteristics that tracked object is stored in advance corresponding prestores size;
Step 102: the images to be recognized of at least two carrier's face facial areas of acquisition, and it is described wait know to be directed to each
Other image executes:
Step 103:C1: at least one carried in the images to be recognized face facial area is identified;
Step 104:C2: being directed to each described face facial area, determines in the face facial area and includes
At least one described facial characteristics, and at least one corresponding characteristic point of each determined facial characteristics is obtained respectively
Coordinate value;
Step 105:C3: for each the described facial characteristics determined, according to the corresponding coordinate of acquisition
Value, determines the corresponding characteristic size of the facial characteristics;
Step 106:C4: according to the corresponding characteristic size of each described facial characteristics determined and described pre-
Size is deposited, is determined from each face facial area that the images to be recognized carries corresponding with the tracked object
Target person face facial area;
Step 107:C5: the location of target person face facial area described in the images to be recognized is determined;
Step 108: according to the location of target person face facial area described at least two images to be recognized, really
The motion profile of the fixed tracked object.
In embodiments of the present invention, at least one face facial area in collected images to be recognized is identified, and
The face facial area identified can be face at any angle, special by the face for determining that everyone face facial area includes
The coordinate value of corresponding characteristic point is levied, that is, can determine the corresponding characteristic size of each facial characteristics, according to each facial characteristics pair
The characteristic size answered and it is corresponding prestore size, can determine to be tracked the corresponding target person of object from images to be recognized
Face facial area can determine tracked object according to position of the target person face facial area in corresponding images to be recognized
Motion profile.To sum up, based on face facial area corresponding to face at any angle, that is, it can determine tracked object
Motion profile, realize face tracking, without based on front face realize, so as to increase the applicable model of face tracking
It encloses.
In an embodiment of the present invention, the C4, comprising:
The corresponding object to be identified of each face facial area and the tracked object are determined according to following formula
Matching value:
Wherein, s characterizes the matching value, αiCharacterize preset threshold value, βiCharacterize the corresponding power of i-th of facial characteristics
Weight values, yiCharacterize that i-th of facial characteristics is corresponding described to prestore size, tiCharacterize the corresponding institute of i-th of facial characteristics
State characteristic size, wherein prestore size described in the threshold value is corresponding with i-th of facial characteristics and the characteristic size is poor
It is worth corresponding;
From each matching value determined, it is preset to determine whether the smallest matching value of numerical value is less than or equal to
Error amount;
If so, the corresponding face facial area of the determining matching value the smallest with numerical value is target person face
Portion region.
In embodiments of the present invention, by each facial characteristics it is corresponding prestore size and corresponding characteristic size it
Between difference absolute value, weighted value corresponding with the absolute value of difference and each facial characteristics calculating, can obtain every
Difference value between each facial characteristics that personal face facial area includes facial characteristics corresponding with tracked object, by right
Difference value summation is obtained, that is, can determine the matching value of face facial area corresponding object to be identified and tracked object.Work as people
When the corresponding object to be identified of face facial area and tracked object are same people, calculated matching value is equal to 0 or approach
In 0, therefore after determining each matching value, it is thus necessary to determine that whether the smallest matching value of numerical value is less than error amount, if it is
Determination identifies that successfully the corresponding face facial area of the smallest matching value of the numerical value is the face for being tracked object.
Whether weighted value can be varied to determine according to facial characteristics, for example, facial characteristics is that Female Eyes may
By makeup, the accuracy for determining the corresponding characteristic point of eyes is influenced, therefore, the corresponding weight of eyes is lower, therefore weighted value
It can be set relatively low, when facial characteristics is nose, nose not easily passs through makeup and changes, therefore the weight of nose is higher, nose
The weighted value of son can be set higher.
In an embodiment of the present invention, before the C2, further comprise:
The corresponding characteristic sequence of each facial characteristics is set;
At least one the described facial characteristics for including in the determination face facial area, and determined respectively
At least one corresponding characteristic point of each described facial characteristics coordinate value, comprising:
Using first facial characteristics in the characteristic sequence as current face feature, execute:
D0: determining with the presence or absence of the current face feature in the face facial area, if so, D1 is executed, otherwise,
Execute D2;
D1: acquiring the coordinate value of the corresponding each characteristic point of the current face feature, executes D3;
D2: the coordinate value of the corresponding each characteristic point of the current face feature is recorded, wherein the seat of record
The abscissa and ordinate of scale value are 0, execute D3;
D3: using next facial characteristics of the current face feature as current face feature, D0 is returned to.
In embodiments of the present invention, in the coordinate for acquiring the corresponding characteristic point of each facial characteristics from face facial area
It when value, needs successively to acquire according to preset characteristic sequence, when not collecting facial characteristics, needs to record not collected face
The corresponding abscissa of portion's feature and ordinate are 0, in order to calculate the characteristic size of facial characteristics according to the coordinate value of record.
For example, when not collecting facial characteristics left eye, the coordinate value for needing to record the corresponding characteristic point of left eye is
(0,0)。
In an embodiment of the present invention, before the C3, further comprise:
It is put in order according to preset, arranges each described facial characteristics and each described facial characteristics is corresponding
Each characteristic point the coordinate value;
Using the corresponding coordinate value of the every two adjacent facial characteristics after blank character partitioned arrangement, and record
The corresponding relationship of the blank character and two adjacent facial characteristics;
For facial characteristics described in each, execute:
S0: according to the corresponding relationship, determine whether the quantity of the corresponding blank character of the facial characteristics is 2, such as
Fruit is to execute S1, otherwise, executes S2;
S1: each coordinate value between two blank characters corresponding with the facial characteristics is extracted;
S2: determining whether the blank character corresponding with the facial characteristics is first blank character, if so, executing
Otherwise S3 executes S4;
S3: each coordinate value before the corresponding blank character of the facial characteristics is extracted;
S4: each coordinate value after the corresponding blank character of the facial characteristics is extracted;
The C3, comprising:
For facial characteristics described in each, the Euclidean between the corresponding each coordinate value of the facial characteristics is calculated
Distance, and using the Euclidean distance as the corresponding characteristic size of the facial characteristics.
In embodiments of the present invention, it after collecting each coordinate value, needs to the corresponding coordinate value of each facial characteristics
Sequence, and blank character is set in the every two adjacent facial characteristics after sequence, the face that intra-record slack byte accords with and separated is special
Corresponding relationship between sign can quickly position correspondence according to corresponding relationship when the corresponding characteristic size of calculating facial characteristics
Blank character where position, and then determine the corresponding each coordinate value of facial characteristics, calculate European between each coordinate value
Distance can be obtained the corresponding characteristic size of facial characteristics.
In an embodiment of the present invention, the facial characteristics, comprising: in facial contour, eyes, mouth and nose at least
It is one or more, wherein the eyes, comprising: left eye and right eye.
In embodiments of the present invention, determine that the facial characteristics for including in everyone face facial area can be, facial contour,
At least one of left eye, right eye, mouth and nose are multiple, corresponding according to facial contour, left eye, right eye, mouth and nose
The coordinate value of characteristic point recognition of face can be realized.
In an embodiment of the present invention, described to prestore characterization of size when the facial characteristics includes the facial contour
The facial contour perimeter prestored, the characteristic size characterize the facial contour perimeter in the face facial area;
When the facial characteristics includes the eyes, the eye widths for prestoring characterization of size and prestoring, the feature
Eye widths in face facial area described in characterization of size;
When the facial characteristics includes the mouth, the mouth width for prestoring characterization of size and prestoring, the feature
Mouth width in face facial area described in characterization of size;
When the facial characteristics includes the nose, the nose width for prestoring characterization of size and prestoring, the feature
Nose width in face facial area described in characterization of size;
It is described to prestore the nose and two that characterization of size prestores when the facial characteristics includes the nose and the mouth
The perimeter for the geometric figure that the side corners of the mouth is formed, the characteristic size characterize nose and the two sides corners of the mouth in the face facial area
The perimeter of the geometric figure of formation;
When the facial characteristics includes the eyes, the nose and the mouth, the characterization of size that prestores is prestored
Eye center and the perimeter of geometric figure that is centrally formed of nose and mouth, the characteristic size characterize the face facial area
The perimeter for the geometric figure that eye center and nose and mouth in domain are centrally formed;
In embodiments of the present invention, prestoring size can be the left eye width, right eye width, mouth width, nose prestored
The geometric graph that the perimeter for the geometric figure that width, nose and the two sides corners of the mouth are formed, left eye center are centrally formed with nose and mouth
The perimeter for the geometric figure that the perimeter of shape and right eye center and nose and mouth are centrally formed.Similarly, characteristic size can be with
It is that left eye width, right eye width, mouth width, nose width, nose and the two sides corners of the mouth in face facial area is formed
The perimeter for the geometric figure that the perimeter of geometric figure, left eye center are centrally formed with nose and mouth and right eye center and nose
The perimeter for the geometric figure that point and mouth are centrally formed.
In an embodiment of the present invention, the target person face portion according at least two images to be recognized described
The location of region before the motion profile for determining the tracked object, further comprises
Determine the corresponding acquisition time of each described images to be recognized;
Described the location of target person face facial area according at least two images to be recognized, determines institute
State the motion profile of tracked object, comprising:
According to the corresponding acquisition time of the images to be recognized where the target person face facial area, institute is arranged
Target facial image is stated in the location of described images to be recognized, forms the motion profile of the tracked object.
In embodiments of the present invention, it by the corresponding acquisition time of each images to be recognized, can determine each wait know
The acquisition order of other image sorts the location of in images to be recognized to target person face facial area according to acquisition order,
It can be formed and be tracked corresponding motion profile, realize the face tracking based on any angle face.
For example, the acquisition time of images to be recognized x is 9:00, and the acquisition time of images to be recognized n is 9:01;
Target person face facial area the location of images to be recognized x be bank outdoors;
Target person face facial area is Bank Hall in the location of images to be recognized n;
Sequentially in time, it arranges the location of target person face portion and is followed successively by bank outdoors and Bank Hall, therefore
It can determine that being tracked corresponding motion profile is from being moved to hall outdoors.
In order to more clearly illustrate technical solution of the present invention and advantage, to a kind of face provided in an embodiment of the present invention
Tracking is described in detail, as shown in Fig. 2, can specifically include following steps:
Step 201, at least one facial characteristics that tracked object is stored in advance corresponding prestores size.
Specifically, the camera for carrying out face acquisition is often that resolution ratio is higher, the image file obtained
Can be larger, Image Compression is only capable of by single file fixing fabric structure in a zone of reasonableness, to keep image data
Clarity.The picture number magnitude acquired with face tracking is significantly increased, and conventional method is in image file storage, read-write, biography
Defeated and etc. every cost also will increase.
For example, facial characteristics be left eye, right eye and nose, it is corresponding prestore size be followed successively by left eye width be 2.5,
Right eye width is 2.6 and nose width is 3.
Step 202, the corresponding characteristic sequence of each facial characteristics is set.
It specifically, can be to the seat of the corresponding characteristic point of facial characteristics according to the corresponding characteristic sequence of each facial characteristics
Acquisition or record when scale value according to this feature sequence, avoid the coordinate value for omitting some characteristic point.
For example, the characteristic sequence of setting is followed successively by left eye, right eye and nose.
Step 203, the images to be recognized of at least two carrier's face facial areas is acquired, and determines each figure to be identified
As corresponding acquisition time.
Specifically, when acquiring image, need to acquire the images to be recognized of carrier's face facial area, so as to be identified
Face at any angle is identified in image.The acquisition time for determining each images to be recognized can aid in determining be tracked
The motion profile of object.
For example, the images to be recognized A and images to be recognized B of carrier's face facial area are acquired.
Step 204, it for each images to be recognized, executes: at least one face carried in identification images to be recognized
Facial area.
Specifically, each face facial area can be therefrom identified after collecting images to be recognized, so that according to people
Face facial area determines the motion profile of tracked object.
For example, face facial area a and face facial area b are identified from images to be recognized A,
Face facial area c, face facial area d and face facial area e are identified from images to be recognized B.
Step 205, it is directed to everyone face facial area, using first facial characteristics in characteristic sequence as current
Facial characteristics executes step 206.
Specifically, successively to obtain each facial characteristics in everyone face facial area according to the characteristic sequence of setting corresponding
The coordinate value of characteristic point can acquire or record the coordinate value of the corresponding characteristic point of some facial characteristics to avoid omission.
For example, it is corresponding each that facial characteristics left eye, right eye and nose are successively acquired or recorded according to characteristic sequence
The coordinate value of characteristic point.
Step 206, it determines with the presence or absence of current face feature in face facial area, if so, step 207 is executed, it is no
Then, step 208 is executed.
Specifically, since the Partial Feature in face facial area is there may be situation is obscured by an object, for example, wearing mouth
Cover, facial characteristics nose and mouth are blocked, and therefore, when acquiring the coordinate value of the corresponding characteristic point of facial characteristics, need really
It is scheduled in face facial area with the presence or absence of current face feature, if it is, determining the corresponding each spy of current face feature
The coordinate value of point is levied, the abscissa and ordinate that otherwise record the corresponding each characteristic point of current face feature are 0, to account for
Position, determines the corresponding characteristic size of facial characteristics convenient for the coordinate value according to record.
For example, in determining the face facial area a in images to be recognized A, it is determined whether there are current face spies
Left eye is levied,
If it is, the coordinate value of the corresponding characteristic point z1 of acquisition left eye is (1,1), the corresponding characteristic point z2 of acquisition left eye
Coordinate value be (1,3);
Otherwise, the coordinate value for needing to record the corresponding characteristic point z1 of left eye is (0,0), the corresponding characteristic point z2 of acquisition left eye
Coordinate value be (0,0).
The acquisition methods of the coordinate value for the corresponding characteristic point of each facial characteristics that face facial area b includes are the same as face face
Portion region a, repeats no more.
Step 207, the coordinate value of the corresponding each characteristic point of acquisition current face feature, executes step 209.
Step 208, the coordinate value of the corresponding each characteristic point of record current face feature, wherein the coordinate value of record
Abscissa and ordinate are 0, execute step 209.
Step 209, using next facial characteristics of current face feature as current face feature, return step 206.
Step 210, when the acquisition of the coordinate value of the corresponding characteristic point of facial characteristics or record finish, according to preset arrangement
Sequentially, each facial characteristics and the corresponding each coordinate value of each facial characteristics are arranged.
Specifically, the coordinate value of each facial characteristics and the corresponding each characteristic point of each facial characteristics is ranked up,
Facilitate in the corresponding characteristic size of calculating facial characteristics, according to the quickly corresponding each spy of positioning facial characteristics that puts in order
Levy the coordinate value of point.
For example, the corresponding characteristic point z1 of facial characteristics left eye of face facial area a in images to be recognized A is acquired
Coordinate value is (1,1), and the coordinate value of the corresponding characteristic point z2 of acquisition left eye is (1,3);
The coordinate value for acquiring the corresponding characteristic point y1 of facial characteristics right eye of face facial area a is (1,6), acquires left eye
The coordinate value of corresponding characteristic point y2 is (1,8);
The coordinate value for acquiring the corresponding characteristic point b1 of facial characteristics nose of face facial area a is (4,4), acquires left eye
The coordinate value of corresponding characteristic point b2 is (4,6);
It is preset to put in order as the corresponding characteristic point z1 of left eye, the corresponding characteristic point z2 of left eye, the corresponding feature of right eye
The corresponding characteristic point y2 of point y1, right eye, the corresponding characteristic point b1 of nose and the corresponding characteristic point b2 of nose;
It puts in order according to this and coordinate value of corresponding characteristic point of each facial characteristics is arranged:
(1,1)(1,3)(1,6)(1,8)(4,4)(4,6)。
Step 211, it using the corresponding coordinate value of every two adjacent facial characteristics after blank character partitioned arrangement, and records
The corresponding relationship of blank character and two adjacent facial characteristics.
Specifically, by the way that blank character is arranged between every two adjacent facial characteristics, face facial area can be formed
Corresponding condition code, and the corresponding relationship of intra-record slack byte symbol and every two adjacent facial characteristics can calculate facial characteristics
When corresponding characteristic size, the coordinate of the corresponding each characteristic point of facial characteristics is quickly positioned from condition code according to blank character
Value.
For example, the coordinate value after blank character is arranged between every two adjacent facial characteristics is formed by condition code
It is as follows:
(1,1)(1,3)┃(1,6)(1,8)┃(4,4)(4,6)。
It is corresponding with facial characteristics left eye and right eye to record first blank character;
It is corresponding with facial characteristics right eye and nose to record second blank character.
Step 212, the number of the corresponding blank character of facial characteristics is determined according to corresponding relationship for each facial characteristics
Whether amount is 2, if so, executing step 213, otherwise, executes step 214.
Specifically, it when the coordinate value to the corresponding each characteristic point of each facial characteristics calculates, needs first to determine
Whether facial characteristics corresponds to two blank characters, if so, determining that the coordinate value between corresponding two blank characters is facial characteristics
The coordinate value of corresponding characteristic point, otherwise, the coordinate value of the corresponding characteristic point of facial characteristics be likely located at condition code stem or
Tail portion.
For example, when extracting the coordinate value of the corresponding each characteristic point of facial characteristics right eye, facial characteristics right eye with
There are corresponding relationships for first blank character and the second blank character, so extracting the coordinate value between the first blank character and the second blank character
(1,6)(1,8)。
Step 213, each coordinate value between two blank characters corresponding with facial characteristics is extracted, step 217 is executed.
Step 214, determine whether blank character corresponding with facial characteristics is first blank character, if so, executing step
215, otherwise, execute step 216.
Specifically, in varying numbers in 2 when the corresponding blank character of facial characteristics, it is thus necessary to determine that blank character is in condition code
Position determines whether the corresponding blank character of facial characteristics is characterized first blank character in code, if it is, can determine
Coordinate value before the corresponding blank character of facial characteristics code is the coordinate value of the corresponding characteristic point of facial characteristics, otherwise, can be with
Determine that the coordinate value after the corresponding blank character of facial characteristics is the coordinate value of the corresponding characteristic point of facial characteristics.
For example, when extracting the coordinate value of the corresponding each characteristic point of facial characteristics left eye, facial characteristics left eye is only
With the first blank character there are corresponding relationship, blank character it is in varying numbers in 2, and corresponding first blank character of facial characteristics left eye is
First blank character in condition code, so, extract the coordinate value (1,1) (1,3) before the first blank character.
Step 215, each coordinate value before blank character corresponding with facial characteristics is extracted, step 217 is executed.
Step 216, each coordinate value after blank character corresponding with facial characteristics is extracted, step 217 is executed.
Step 217, for each facial characteristics, calculate Euclidean between the corresponding each coordinate value of facial characteristics away from
From, and using Euclidean distance as the corresponding characteristic size of facial characteristics.
Specifically, by calculating the Euclidean distance between each characteristic point, that is, it can determine the corresponding feature ruler of facial characteristics
It is very little.
It should be noted that when facial characteristics is there are when multiple characteristic sizes, can also according to preset computation rule, from
The coordinate value to be calculated is selected in the coordinate value extracted, then calculates the Euclidean distance before the coordinate value of selection, for example, face
Portion's feature is the perimeter for the geometric figure that left eye center is centrally formed with nose and mouth.The corresponding characteristic point of left eye can be extracted
Coordinate value, the corresponding characteristic point of mouth the corresponding characteristic point of coordinate value, nose coordinate value, from the coordinate value extracted
The corresponding coordinate value in selection left eye center respectively, the coordinate value of mouth center corresponding coordinate value and nose.
For example, calculate the coordinate value (1,1) (1,3) of the corresponding each characteristic point of facial characteristics left eye it is European away from
From, can determine the corresponding characteristic size of facial characteristics left eye be 2.
The Euclidean distance for calculating the coordinate value (1,6) (1,8) of the corresponding each characteristic point of facial characteristics right eye, can determine
The corresponding characteristic size of facial characteristics left eye is 2.
The Euclidean distance for calculating the coordinate value (4,4) (4,6) of the corresponding each characteristic point of facial characteristics left eye, can determine
The corresponding characteristic size of facial characteristics left eye is 2.
Step 218, the matching value of everyone face facial area corresponding object to be identified and tracked object is determined.
Specifically, by prestoring difference between size and corresponding characteristic size to each facial characteristics is corresponding
The calculating of absolute value, weighted value corresponding with the absolute value of difference and each facial characteristics, can obtain everyone face portion
Difference value between each facial characteristics that region includes facial characteristics corresponding with tracked object, by acquisition difference value
Summation, that is, can determine the matching value of face facial area corresponding object to be identified and tracked object.
For example, when the difference for prestoring size and characteristic size is 0, corresponding threshold value is 0.
Prestore the difference of size and characteristic size section be (0,1] when, corresponding threshold value be 1.
Prestore the difference of size and characteristic size section be (1,2] when, corresponding threshold value be 2.
The corresponding weighted value of facial characteristics left eye is 5, and the corresponding weighted value of right eye is 5, and the corresponding weighted value of nose is 10.
According to following formula, of face facial area a corresponding object to be identified and tracked object can be calculated
It is 15.5 with value.
Wherein, s characterizes matching value, αiCharacterize preset threshold value, βiCharacterize the corresponding weighted value of i-th of facial characteristics, yiTable
Levy that i-th of facial characteristics is corresponding to prestore size, tiCharacterize the corresponding characteristic size of i-th of facial characteristics, wherein threshold value and the
I facial characteristics be corresponding, and to prestore size corresponding with characteristic size difference.
Step 219, from each matching value determined, it is preset to determine whether the smallest matching value of numerical value is less than or equal to
Error amount.
Specifically, after determining the smallest matching value of numerical value, it is also necessary to determine whether the smallest matching value of numerical value is less than
Equal to preset error amount, so as to improve the accuracy of identification.
For example, the matching value for determining the corresponding object to be identified of face facial area a and tracked object is 15.5
Whether preset error amount 3 is greater than, accordingly, it is determined that the corresponding object to be identified of face facial area a is not with tracked object
Same people is not target person face facial area.
Step 220, if so, determining that face facial area corresponding with the smallest matching value of numerical value is target person face
Portion region.
Step 221, the location of target person face facial area in images to be recognized is determined.
Specifically, the location of target person face facial area in determining images to be recognized, can be target person face
The geometric center in portion region can also be the upper left corner, the right side of target person face facial area the location of in images to be recognized
Upper angle, the lower left corner or the lower right corner are the location of in images to be recognized.
Step 222, according to the corresponding acquisition time of images to be recognized where target person face facial area, target person is arranged
Face image forms the motion profile for being tracked object in the location of images to be recognized.
Specifically, the acquisition order that images to be recognized can be determined according to acquisition time, can be to mesh according to acquisition order
It marks face facial area to be ranked up the location of in images to be recognized, generates motion profile.
For example, the acquisition time of images to be recognized a is 11:00, and the acquisition time of images to be recognized n is 11:01;
Target person face facial area the location of images to be recognized a be dining room outdoors;
Target person face facial area is in the door of dining room in the location of images to be recognized b;
According to acquisition time, arranges the location of target person face portion and be followed successively by dining room outdoors and in the door of dining room, therefore
It can determine that being tracked corresponding motion profile is moved to outdoors in the door of dining room from dining room.
As shown in figure 3, the embodiment of the invention provides a kind of face tracking devices, comprising:
Administrative unit 301, at least one facial characteristics for tracked object to be stored in advance is corresponding to prestore size;
Acquisition unit 302 for acquiring the images to be recognized of at least two carrier's face facial areas, and is directed to each
The images to be recognized executes: identifying at least one carried in the images to be recognized face facial area;It is directed to
Each described face facial area determines at least one the described facial characteristics for including in the face facial area, and divides
The coordinate value of at least one corresponding characteristic point of each determined facial characteristics is not obtained;
Processing unit 303, each described facial characteristics for determining for the acquisition unit 302, according to obtaining
The corresponding coordinate value obtained, determines the corresponding characteristic size of the facial characteristics;According to each determined
The corresponding characteristic size of facial characteristics and the described of the storage of administrative unit 301 prestore size, from the images to be recognized
Target person face facial area corresponding with the tracked object is determined in each face facial area carried;It determines
The location of target person face facial area described in the images to be recognized;
Tracking cell 304, described at least two images to be recognized for being determined according to the processing unit 303
The location of target person face facial area determines the motion profile of the tracked object.
In an embodiment of the present invention, the processing unit, for determining each face face according to following formula
The matching value of the corresponding object to be identified in region and the tracked object;It determines in the matching value, numerical value is the smallest described
Whether matching value is less than or equal to preset matching value;
If so, the corresponding face facial area of the determining matching value the smallest with numerical value is target person face
Portion region;
Wherein, s characterizes the matching value, αiCharacterize preset threshold value, βiCharacterize the corresponding power of i-th of facial characteristics
Weight values, yiCharacterize that i-th of facial characteristics is corresponding described to prestore size, tiCharacterize the corresponding institute of i-th of facial characteristics
State characteristic size, wherein prestore size described in the threshold value is corresponding with i-th of facial characteristics and the characteristic size is poor
It is worth corresponding.
In an embodiment of the present invention, administrative unit is further used for that the corresponding feature of each facial characteristics is arranged
Sequentially;
The acquisition unit, for holding using first facial characteristics in the characteristic sequence as current face feature
Row:
D0: determining with the presence or absence of the current face feature in the face facial area, if so, D1 is executed, otherwise,
Execute D2;
D1: acquiring the coordinate value of the corresponding each characteristic point of the current face feature, executes D3;
D2: the coordinate value of the corresponding each characteristic point of the current face feature is recorded, wherein the seat of record
The abscissa and ordinate of scale value are 0, execute D3;
D3: using next facial characteristics of the current face feature as current face feature, D0 is returned to.
In an embodiment of the present invention, the acquisition unit is further used for being put in order according to preset, arranges each
The coordinate value of a facial characteristics and the corresponding each characteristic point of each described facial characteristics;Between utilization
The corresponding coordinate value of the every two adjacent facial characteristics after according with partitioned arrangement, and record the blank character and two
The corresponding relationship of a adjacent facial characteristics;For facial characteristics described in each, execute:
S0: according to the corresponding relationship, determine whether the quantity of the corresponding blank character of the facial characteristics is 2, such as
Fruit is to execute S1, otherwise, executes S2;
S1: each coordinate value between two blank characters corresponding with the facial characteristics is extracted;
S2: determining whether the blank character corresponding with the facial characteristics is first blank character, if so, executing
Otherwise S3 executes S4;
S3: each coordinate value before the corresponding blank character of the facial characteristics is extracted;
S4: each coordinate value after the corresponding blank character of the facial characteristics is extracted;
The processing unit calculates the corresponding each institute of the facial characteristics for being directed to each described facial characteristics
The Euclidean distance between coordinate value is stated, and using the Euclidean distance as the corresponding characteristic size of the facial characteristics.
In an embodiment of the present invention, the facial characteristics, comprising: in facial contour, eyes, mouth and nose at least
It is one or more, wherein the eyes, comprising: left eye and right eye.
In an embodiment of the present invention, described to prestore characterization of size when the facial characteristics includes the facial contour
The facial contour perimeter prestored, the characteristic size characterize the facial contour perimeter in the face facial area;
When the facial characteristics includes the eyes, the eye widths for prestoring characterization of size and prestoring, the feature
Eye widths in face facial area described in characterization of size;
When the facial characteristics includes the mouth, the mouth width for prestoring characterization of size and prestoring, the feature
Mouth width in face facial area described in characterization of size;
When the facial characteristics includes the nose, the nose width for prestoring characterization of size and prestoring, the feature
Nose width in face facial area described in characterization of size;
It is described to prestore the nose and two that characterization of size prestores when the facial characteristics includes the nose and the mouth
The perimeter for the geometric figure that the side corners of the mouth is formed, the characteristic size characterize nose and the two sides corners of the mouth in the face facial area
The perimeter of the geometric figure of formation;
When the facial characteristics includes the eyes, the nose and the mouth, the characterization of size that prestores is prestored
Eye center and the perimeter of geometric figure that is centrally formed of nose and mouth, the characteristic size characterize the face facial area
The perimeter for the geometric figure that eye center and nose and mouth in domain are centrally formed.
In an embodiment of the present invention, the administrative unit is further used for determining each described images to be recognized pair
The acquisition time answered;
The tracking cell, for according to the corresponding institute of the images to be recognized where the target person face facial area
Acquisition time is stated, the target facial image is arranged in the location of described images to be recognized, forms the tracked object
Motion profile.
The each embodiment of the present invention at least has the following beneficial effects:
1, at least one face facial area in collected images to be recognized in an embodiment of the present invention, is identified
Domain, and the face facial area identified can be face at any angle, include by everyone determining face facial area
The coordinate value of the corresponding characteristic point of facial characteristics can determine the corresponding characteristic size of each facial characteristics, according to each face
The corresponding characteristic size of feature and it is corresponding prestore size, can be determined from images to be recognized be tracked object it is corresponding
Target person face facial area, according to position of the target person face facial area in corresponding images to be recognized, that is, can determine by with
The motion profile of track object.To sum up, based on face facial area corresponding to face at any angle, that is, can determine by with
The motion profile of track object realizes face tracking, without being realized based on front face, so as to increase the suitable of face tracking
Use range.
2, in an embodiment of the present invention, by prestoring size and corresponding feature ruler to each facial characteristics is corresponding
The calculating of the absolute value of difference between very little, weighted value corresponding with the absolute value of difference and each facial characteristics, can obtain
The difference value between each facial characteristics facial characteristics corresponding with tracked object that everyone face facial area includes is obtained, is led to
It crosses to difference value summation is obtained, that is, can determine the matching value of face facial area corresponding object to be identified and tracked object.
When the corresponding object to be identified of face facial area and tracked object are same people, calculated matching value be equal to 0 or
Level off to 0, therefore after determining each matching value, it is thus necessary to determine that whether the smallest matching value of numerical value is less than error amount, if
Be it is then determining identify successfully, the corresponding face facial area of the smallest matching value of the numerical value be tracked object face.
3, the corresponding characteristic point of each facial characteristics in an embodiment of the present invention, is being acquired from face facial area
It when coordinate value, needs successively to acquire according to preset characteristic sequence, when not collecting facial characteristics, record is needed not collect
The corresponding abscissa of facial characteristics and ordinate be 0, in order to according to the coordinate value of record calculate facial characteristics feature
Size.
4, it in an embodiment of the present invention, after collecting each coordinate value, needs to the corresponding seat of each facial characteristics
Scale value sequence, and blank character is set in the every two adjacent facial characteristics after sequence, the face that intra-record slack byte accords with and separated
Corresponding relationship between portion's feature can quickly be positioned when the corresponding characteristic size of calculating facial characteristics according to corresponding relationship
Position where corresponding blank character, and then determine the corresponding each coordinate value of facial characteristics, it calculates between each coordinate value
Euclidean distance can be obtained the corresponding characteristic size of facial characteristics.
It should be noted that, in this document, such as first and second etc relational terms are used merely to an entity
Or operation is distinguished with another entity or operation, is existed without necessarily requiring or implying between these entities or operation
Any actual relationship or order.Moreover, the terms "include", "comprise" or its any other variant be intended to it is non-
It is exclusive to include, so that the process, method, article or equipment for including a series of elements not only includes those elements,
It but also including other elements that are not explicitly listed, or further include solid by this process, method, article or equipment
Some elements.In the absence of more restrictions, the element limited by sentence " including one ", is not arranged
Except there is also other identical factors in the process, method, article or apparatus that includes the element.
Finally, it should be noted that the foregoing is merely presently preferred embodiments of the present invention, it is merely to illustrate skill of the invention
Art scheme, is not intended to limit the scope of the present invention.Any modification for being made all within the spirits and principles of the present invention,
Equivalent replacement, improvement etc., are included within the scope of protection of the present invention.
Claims (10)
1. a kind of face tracking method, which is characterized in that at least one facial characteristics that tracked object is stored in advance is corresponding
Prestore size, further includes:
The images to be recognized of at least two carrier's face facial areas is acquired, and is directed to each described images to be recognized, is executed:
C1: at least one carried in the images to be recognized face facial area is identified;
C2: it is directed to each described face facial area, is determined described at least one for including in the face facial area
Facial characteristics, and the coordinate value of at least one corresponding characteristic point of each determined facial characteristics is obtained respectively;
C3: the face is determined according to the corresponding coordinate value of acquisition for each the described facial characteristics determined
The corresponding characteristic size of portion's feature;
C4: according to the corresponding characteristic size of each described facial characteristics determined and it is described prestore size, from institute
It states and determines target face corresponding with the tracked object in each face facial area of images to be recognized carrying
Facial area;
C5: the location of target person face facial area described in the images to be recognized is determined;
According to the location of target person face facial area described at least two images to be recognized, described be tracked is determined
The motion profile of object.
2. the method according to claim 1, wherein
The C4, comprising:
Of each face facial area corresponding object to be identified and the tracked object is determined according to following formula
With value:
Wherein, s characterizes the matching value, αiCharacterize preset threshold value, βiThe corresponding weighted value of i-th of facial characteristics is characterized,
yiCharacterize that i-th of facial characteristics is corresponding described to prestore size, tiCharacterize the corresponding spy of i-th of facial characteristics
Levy size, wherein prestore size and the characteristic size difference phase described in the threshold value is corresponding with i-th of facial characteristics
It is corresponding;
From each matching value determined, determine whether the smallest matching value of numerical value is less than or equal to preset error
Value;
If so, the corresponding face facial area of the determining matching value the smallest with numerical value is target person face facial area
Domain.
3. the method according to claim 1, wherein
Before the C2, further comprise:
The corresponding characteristic sequence of each facial characteristics is set;
At least one the described facial characteristics for including in the determination face facial area, and obtain respectively determine it is every
The coordinate value of at least one corresponding characteristic point of one facial characteristics, comprising:
Using first facial characteristics in the characteristic sequence as current face feature, execute:
D0: determining with the presence or absence of the current face feature in the face facial area, if so, executing D1, otherwise, executes
D2;
D1: acquiring the coordinate value of the corresponding each characteristic point of the current face feature, executes D3;
D2: the coordinate value of the corresponding each characteristic point of the current face feature is recorded, wherein the coordinate value of record
Abscissa and ordinate be 0, execute D3;
D3: using next facial characteristics of the current face feature as current face feature, D0 is returned to.
4. the method according to claim 1, wherein
Before the C3, further comprise:
It is put in order according to preset, arranges each described facial characteristics and each described facial characteristics is corresponding each
The coordinate value of a characteristic point;
Using the corresponding coordinate value of the every two adjacent facial characteristics after blank character partitioned arrangement, and described in recording
The corresponding relationship of blank character and two adjacent facial characteristics;
For facial characteristics described in each, execute:
S0: according to the corresponding relationship, determining whether the quantity of the corresponding blank character of the facial characteristics is 2, if so,
S1 is executed, otherwise, executes S2;
S1: each coordinate value between two blank characters corresponding with the facial characteristics is extracted;
S2: determining whether the blank character corresponding with the facial characteristics is first blank character, if so, S3 is executed, it is no
Then, S4 is executed;
S3: each coordinate value before the corresponding blank character of the facial characteristics is extracted;
S4: each coordinate value after the corresponding blank character of the facial characteristics is extracted;
The C3, comprising:
For facial characteristics described in each, calculate Euclidean between the corresponding each coordinate value of the facial characteristics away from
From, and using the Euclidean distance as the corresponding characteristic size of the facial characteristics.
5. according to claim 1 to method described in 4, which is characterized in that
The facial characteristics, comprising: at least one of facial contour, eyes, mouth and nose are multiple, wherein the eye
Eyeball, comprising: left eye and right eye;
And/or
When the facial characteristics includes the facial contour, the facial contour perimeter for prestoring characterization of size and prestoring is described
Characteristic size characterizes the facial contour perimeter in the face facial area;
When the facial characteristics includes the eyes, the eye widths for prestoring characterization of size and prestoring, the characteristic size
Characterize the eye widths in the face facial area;
When the facial characteristics includes the mouth, the mouth width for prestoring characterization of size and prestoring, the characteristic size
Characterize the mouth width in the face facial area;
When the facial characteristics includes the nose, the nose width for prestoring characterization of size and prestoring, the characteristic size
Characterize the nose width in the face facial area;
It is described to prestore the nose and two sides mouth that characterization of size prestores when the facial characteristics includes the nose and the mouth
The perimeter for the geometric figure that angle is formed, the characteristic size characterize the nose in the face facial area and are formed with the two sides corners of the mouth
Geometric figure perimeter;
When the facial characteristics includes the eyes, the nose and the mouth, the eye for prestoring characterization of size and prestoring
The perimeter for the geometric figure that eyeball center and nose and mouth are centrally formed, the characteristic size characterize in the face facial area
Eye center and the perimeter of geometric figure that is centrally formed of nose and mouth;
And/or
In described the location of target person face facial area according at least two images to be recognized, determine described in
It is tracked before the motion profile of object, further comprises
Determine the corresponding acquisition time of each described images to be recognized;
Described the location of target person face facial area according at least two images to be recognized, determines the quilt
The motion profile of tracking object, comprising:
According to the corresponding acquisition time of the images to be recognized where the target person face facial area, the mesh is arranged
Facial image is marked in the location of described images to be recognized, forms the motion profile of the tracked object.
6. a kind of face tracking device characterized by comprising
Administrative unit, at least one facial characteristics for tracked object to be stored in advance is corresponding to prestore size;
Acquisition unit, for acquiring the images to be recognized of at least two carrier's face facial areas, and for described in each to
It identifies image, executes: identifying at least one carried in the images to be recognized face facial area;It is directed to each
The face facial area determines at least one the described facial characteristics for including in the face facial area, and obtains respectively
The coordinate value of at least one corresponding characteristic point of each determined facial characteristics;
Processing unit, each described facial characteristics for being determined for the acquisition unit, according to the corresponding of acquisition
The coordinate value, determine the corresponding characteristic size of the facial characteristics;According to each the described facial characteristics pair determined
The described of the characteristic size and administrative unit storage answered prestores size, and what is carried from the images to be recognized is each described
Target person face facial area corresponding with the tracked object is determined in face facial area;Determine the images to be recognized
Described in the location of target person face facial area;
Tracking cell, target person face described at least two images to be recognized for being determined according to the processing unit
The location of portion region determines the motion profile of the tracked object.
7. device according to claim 6, which is characterized in that
The processing unit, for determining the corresponding object to be identified of each face facial area and institute according to following formula
State the matching value of tracked object;It determines in the matching value, it is preset whether the smallest matching value of numerical value is less than or equal to
Matching value;
If so, the corresponding face facial area of the determining matching value the smallest with numerical value is target person face facial area
Domain;
Wherein, s characterizes the matching value, αiCharacterize preset threshold value, βiThe corresponding weighted value of i-th of facial characteristics is characterized,
yiCharacterize that i-th of facial characteristics is corresponding described to prestore size, tiCharacterize the corresponding spy of i-th of facial characteristics
Levy size, wherein prestore size and the characteristic size difference phase described in the threshold value is corresponding with i-th of facial characteristics
It is corresponding.
8. device according to claim 6, which is characterized in that
Administrative unit is further used for that the corresponding characteristic sequence of each facial characteristics is arranged;
The acquisition unit, for executing using first facial characteristics in the characteristic sequence as current face feature:
D0: determining with the presence or absence of the current face feature in the face facial area, if so, executing D1, otherwise, executes
D2;
D1: acquiring the coordinate value of the corresponding each characteristic point of the current face feature, executes D3;
D2: the coordinate value of the corresponding each characteristic point of the current face feature is recorded, wherein the coordinate value of record
Abscissa and ordinate be 0, execute D3;
D3: using next facial characteristics of the current face feature as current face feature, D0 is returned to.
9. device according to claim 6, which is characterized in that
The acquisition unit is further used for being put in order according to preset, arranges each described facial characteristics and each
The coordinate value of the corresponding each characteristic point of a facial characteristics;Utilize the every two phase after blank character partitioned arrangement
The adjacent corresponding coordinate value of the facial characteristics, and record pair of the blank character Yu two adjacent facial characteristics
It should be related to;For facial characteristics described in each, execute:
S0: according to the corresponding relationship, determining whether the quantity of the corresponding blank character of the facial characteristics is 2, if so,
S1 is executed, otherwise, executes S2;
S1: each coordinate value between two blank characters corresponding with the facial characteristics is extracted;
S2: determining whether the blank character corresponding with the facial characteristics is first blank character, if so, S3 is executed, it is no
Then, S4 is executed;
S3: each coordinate value before the corresponding blank character of the facial characteristics is extracted;
S4: each coordinate value after the corresponding blank character of the facial characteristics is extracted;
The processing unit calculates the corresponding each seat of the facial characteristics for being directed to each described facial characteristics
Euclidean distance between scale value, and using the Euclidean distance as the corresponding characteristic size of the facial characteristics.
10. according to the device any in claim 6 to 9, which is characterized in that
The facial characteristics, comprising: at least one of facial contour, eyes, mouth and nose are multiple, wherein the eye
Eyeball, comprising: left eye and right eye;
And/or
When the facial characteristics includes the facial contour, the facial contour perimeter for prestoring characterization of size and prestoring is described
Characteristic size characterizes the facial contour perimeter in the face facial area;
When the facial characteristics includes the eyes, the eye widths for prestoring characterization of size and prestoring, the characteristic size
Characterize the eye widths in the face facial area;
When the facial characteristics includes the mouth, the mouth width for prestoring characterization of size and prestoring, the characteristic size
Characterize the mouth width in the face facial area;
When the facial characteristics includes the nose, the nose width for prestoring characterization of size and prestoring, the characteristic size
Characterize the nose width in the face facial area;
It is described to prestore the nose and two sides mouth that characterization of size prestores when the facial characteristics includes the nose and the mouth
The perimeter for the geometric figure that angle is formed, the characteristic size characterize the nose in the face facial area and are formed with the two sides corners of the mouth
Geometric figure perimeter;
When the facial characteristics includes the eyes, the nose and the mouth, the eye for prestoring characterization of size and prestoring
The perimeter for the geometric figure that eyeball center and nose and mouth are centrally formed, the characteristic size characterize in the face facial area
Eye center and the perimeter of geometric figure that is centrally formed of nose and mouth;
And/or
The administrative unit is further used for determining the corresponding acquisition time of each described images to be recognized;
The tracking cell, for described being adopted according to the images to be recognized where the target person face facial area is corresponding
Collect the time, arranges the target facial image in the location of described images to be recognized, form the fortune of the tracked object
Dynamic rail mark.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810871087.8A CN109146913B (en) | 2018-08-02 | 2018-08-02 | Face tracking method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810871087.8A CN109146913B (en) | 2018-08-02 | 2018-08-02 | Face tracking method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109146913A true CN109146913A (en) | 2019-01-04 |
CN109146913B CN109146913B (en) | 2021-05-18 |
Family
ID=64799515
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810871087.8A Active CN109146913B (en) | 2018-08-02 | 2018-08-02 | Face tracking method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109146913B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110266947A (en) * | 2019-06-26 | 2019-09-20 | Oppo广东移动通信有限公司 | Photographic method and relevant apparatus |
TWI719409B (en) * | 2019-02-23 | 2021-02-21 | 和碩聯合科技股份有限公司 | Tracking system and tracking method thereof |
CN112614168A (en) * | 2020-12-21 | 2021-04-06 | 浙江大华技术股份有限公司 | Target face tracking method and device, electronic equipment and storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101142584A (en) * | 2005-02-21 | 2008-03-12 | 三菱电机株式会社 | Method for facial features detection |
CN102214291A (en) * | 2010-04-12 | 2011-10-12 | 云南清眸科技有限公司 | Method for quickly and accurately detecting and tracking human face based on video sequence |
CN102332095A (en) * | 2011-10-28 | 2012-01-25 | 中国科学院计算技术研究所 | Face motion tracking method, face motion tracking system and method for enhancing reality |
CN103632126A (en) * | 2012-08-20 | 2014-03-12 | 华为技术有限公司 | Human face tracking method and device |
-
2018
- 2018-08-02 CN CN201810871087.8A patent/CN109146913B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101142584A (en) * | 2005-02-21 | 2008-03-12 | 三菱电机株式会社 | Method for facial features detection |
CN102214291A (en) * | 2010-04-12 | 2011-10-12 | 云南清眸科技有限公司 | Method for quickly and accurately detecting and tracking human face based on video sequence |
CN102332095A (en) * | 2011-10-28 | 2012-01-25 | 中国科学院计算技术研究所 | Face motion tracking method, face motion tracking system and method for enhancing reality |
CN103632126A (en) * | 2012-08-20 | 2014-03-12 | 华为技术有限公司 | Human face tracking method and device |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWI719409B (en) * | 2019-02-23 | 2021-02-21 | 和碩聯合科技股份有限公司 | Tracking system and tracking method thereof |
CN110266947A (en) * | 2019-06-26 | 2019-09-20 | Oppo广东移动通信有限公司 | Photographic method and relevant apparatus |
CN112614168A (en) * | 2020-12-21 | 2021-04-06 | 浙江大华技术股份有限公司 | Target face tracking method and device, electronic equipment and storage medium |
CN112614168B (en) * | 2020-12-21 | 2023-08-29 | 浙江大华技术股份有限公司 | Target face tracking method and device, electronic equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN109146913B (en) | 2021-05-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106647742B (en) | Movement routine method and device for planning | |
CN104881637B (en) | Multimodal information system and its fusion method based on heat transfer agent and target tracking | |
CN103577815B (en) | A kind of face alignment method and system | |
Kölsch et al. | Robust Hand Detection. | |
CN109657609A (en) | Face identification method and system | |
CN106530310B (en) | A kind of pedestrian count method and device based on the identification of human body overhead | |
CN110210276A (en) | A kind of motion track acquisition methods and its equipment, storage medium, terminal | |
CN108986064A (en) | A kind of people flow rate statistical method, equipment and system | |
CN102063607B (en) | Method and system for acquiring human face image | |
CN103902958A (en) | Method for face recognition | |
EP1631933A1 (en) | Human figure contour outlining in images | |
CN109146913A (en) | A kind of face tracking method and device | |
CN110084258A (en) | Face preferred method, equipment and storage medium based on video human face identification | |
CN103020607A (en) | Face recognition method and face recognition device | |
CN108268814A (en) | A kind of face identification method and device based on the fusion of global and local feature Fuzzy | |
Olague et al. | Evolving head tracking routines with brain programming | |
CN109297489A (en) | A kind of indoor navigation method based on user characteristics, electronic equipment and storage medium | |
CN103902992B (en) | Human face recognition method | |
JP2016181159A (en) | System, retrieval method and program | |
CN106599785A (en) | Method and device for building human body 3D feature identity information database | |
CN110032932A (en) | A kind of human posture recognition method based on video processing and decision tree given threshold | |
CN106650628B (en) | Fingertip detection method based on three-dimensional K curvature | |
CN109063643A (en) | A kind of facial expression pain degree recognition methods under the hidden conditional for facial information part | |
CN106611158A (en) | Method and equipment for obtaining human body 3D characteristic information | |
CN106156739B (en) | A kind of certificate photo ear detection and extracting method based on face mask analysis |
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 | ||
TA01 | Transfer of patent application right |
Effective date of registration: 20210420 Address after: Wusong Industrial Park, Wuzhong Development District of Suzhou City, Jiangsu province 215100 Wusong Road No. 818 Applicant after: INSPUR FINANCIAL INFORMATION TECHNOLOGY Co.,Ltd. Address before: 215100 No. 178 Tayun Road, Yuexi Street, Wuzhong District, Suzhou City, Jiangsu Province Applicant before: SUZHOU INSPUR INTELLIGENT SOFTWARE Co.,Ltd. |
|
TA01 | Transfer of patent application right | ||
GR01 | Patent grant | ||
GR01 | Patent grant |