CN104252231A - Camera based motion sensing recognition system and method - Google Patents
Camera based motion sensing recognition system and method Download PDFInfo
- Publication number
- CN104252231A CN104252231A CN201410492045.5A CN201410492045A CN104252231A CN 104252231 A CN104252231 A CN 104252231A CN 201410492045 A CN201410492045 A CN 201410492045A CN 104252231 A CN104252231 A CN 104252231A
- Authority
- CN
- China
- Prior art keywords
- body sense
- interest
- area
- skin
- angle
- 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
Abstract
The invention discloses a camera based motion sensing recognition system which comprises a camera and a recognition module. The recognition module is used for recognizing types of motions captured by the camera. According to the technical scheme, the problem about how to realize effective fusion of the motion sensing system and android and IOS platforms can be solved effectively.
Description
Technical field
The present invention relates to the other system and method for a kind of body perception, be specifically related to the other system and method for a kind of body perception based on camera.
Background technology
Existing body sense recognition technology has following several:
(1) the AIWI mobile phone body sense recognition technology of game is applied to: the AIWI mobile phone body sense recognition technology being applied to game goes the motion of induction mobile phone in physical space based on hardware devices such as acceleration sensor (induction gravity), distance-sensor and three-axis gyroscopes, comprise the physical quantitys such as movement velocity, acceleration, track, carry out the Large Amplitude Motion of analogue mobile phone holder, and based on some SPGs of these design data.
(2) the body sense recognition technology of Kinect and the Xbox equipment of Microsoft is applied to: the body sense recognition technology being applied to the equipment such as kinect and Xbox360 of Microsoft, that the camera carried by equipment carrys out Gather and input data, and by analyzing the view data of input, carry out body sense identification activity.
The shortcoming of existing body sense recognition technology:
(1) be applied to the AIWI mobile phone body sense recognition technology of game, owing to being based on the identification to mobile phone movement locus, therefore it cannot move the motion of mobile phone and body the change done and be separated, such as imitating shell piano; Its accuracy identified is not high, trickle motion None-identified.
(2) be applied to the body sense recognition technology of the equipment such as kinect and Xbox360 of Microsoft, shortcoming is high cost, and carries underaction, and excessive being not suitable for of calculated amount transplants the platform in mobile terminal.
Summary of the invention
The object of this invention is to provide a kind of body sense recognition system based on camera.
The present invention also aims to provide a kind of body sense recognition methods based on camera.
First aspect present invention provides a kind of body sense recognition system based on camera: comprise camera, and the described body sense recognition system based on camera also comprises identification module, and the body that described identification module is used for identification camera picked-up moves the classification done.
Body sense recognition system based on camera also comprises the display module of the recognition result transmitted for Identification display module.
Identification module comprises area-of-interest setup unit, body sense recognition unit, and area-of-interest setup unit is connected to display module through body sense recognition unit.
Identification module also comprises result synthesis unit, and area-of-interest setup unit is connected to display module through body sense recognition unit, synthesis unit.
Another aspect of the present invention provides a kind of body sense recognition methods based on camera, comprises the following steps:
A, startup camera;
B, camera receptor move makes original image, and passes to identification module by moved for body as original image;
Moved for body original image of doing mates as library file with body is moved by C, identification module, draws recognition result, and recognition result is presented at display module.
Described step B is:
Camera receptor moves makes original image, and passes to area-of-interest setup unit by moved for the body of picked-up as original image.
Step C comprises following step:
Body is moved the part cutting being positioned at area-of-interest as original image by C1, area-of-interest setup unit, and by the image transfer of cutting to body sense recognition unit;
The image received from area-of-interest setup unit is carried out skin colorimetric detection and local feature point extraction by C2, body sense recognition unit;
The local feature region extracted is converted into proper vector by descriptor by C3, body sense recognition unit;
Described proper vector and body are moved the moved proper vector done of each individuality done in library file and mate by C4, body sense recognition unit, draw recognition result;
The position of the area-of-interest at described recognition result place and size are fed back to area-of-interest setup unit by C5, body sense recognition unit;
Recognition result is sent to display module and shows by C6, body sense recognition unit.
The recognition result wherein drawn in step C4 moves for certain is individual to be done.
Step C4 comprises following step:
Described proper vector and body emotion are made each individuality in library file and are moved the calculating that the proper vector done carries out colour of skin angle number distance by C41, body sense recognition unit, calculate formula as follows
Dis
angleNum=∑
i ∈ [1, n]shu c
i-fc
ishu
2
Wherein, n moves the characteristic number doing to have for certain in body sense maneuver library is individual, c
iand fc
ibe respectively the colour of skin angle number of i-th feature that the image that collects specifically has and this standard body and move the colour of skin angle number that work i-th feature have, if dis
angleNumbe less than given threshold value, threshold value is greater than 3, is less than 15, then think that it may move with this body and coincide, and moved for this body coincide work is carried out record;
The moved proper vector done of the standard body that proper vector and step C41 filter out by C42, body sense recognition unit is carried out distance and is calculated, and formula is as follows:
Dis
angle=∑
i ∈ [1, n]∑
j ∈ [1, m]shu a
ij-fa
ijshu
2
Wherein n makes for this body in body sense maneuver library moves the characteristic number that has, and m is then its colour of skin angle number, a
ijand fa
ijthen be respectively that the image that collects specifically has and standard body moves the value of a jth colour of skin angle under i-th feature done.If dis
anglebe less than given threshold value, threshold value is greater than 3, is less than 15, then think that it may move with this body and coincide, and moved for this body coincide work is carried out record;
Proper vector is moved with first two standard bodies filtered out the proper vector done and carries out mating of non-colour of skin angle by C43, body sense recognition unit, and its formula calculating distance is as follows:
Dis
nonAngle=∑
i ∈ [1, n]∑
j ∈ [1, m]shu na
ij-fna
ijshu
2
Wherein n makes for this body in body sense maneuver library moves the characteristic number that has, and m is then its non-colour of skin angle number, na
ijand fna
ijbe respectively that image has and standard body moves the value of jth non-colour of skin angle under i-th feature done.After the distance of all non-colour of skin angles all calculates, choose the shortest standard body of its middle distance and move work, be set to current matching result and sent to result synthesis unit.Step C6 comprises following step:
Recognition result is sent to result synthesis unit by C61, body sense recognition unit;
The area-of-interest at the recognition result place received moves at body by C62, result synthesis unit to be done original image indicates, and marks out the code of recognition result on the moved area-of-interest doing recognition result place in original image of body;
The display screen that body emotion is sent to mobile device as original image shows by C63, result synthesis unit.The method to set up of described area-of-interest is: when first time arranges area-of-interest, acquiescence area-of-interest covers the whole image collected, when arranging afterwards, can adjust according to the position of the area-of-interest at the described recognition result place of body sense recognition unit feedback and size at every turn.
The method that described local feature region extracts is:
A, by after Face Detection, the image received can be carried out masking-out process by body sense feature identification module;
B, with the center of source images for n concentric circles is done in the center of circle, n is characteristic number, and n is greater than 3, is less than 7, and maximum concentric circles radius is the distance r that source images border is arrived in the center of circle
m, n concentrically ringed radius is followed successively by, r
1, r
2..., r
n, and meet
now with r
1for on the circle of radius, equidistant searching 360 points, are labeled as p
1, p
2..., p
360, its value is the gray-scale value of the pixel of this position, arranges ordered series of numbers f simultaneously
1, f
2..., f
360represent the feature of these 360 points, wherein
s.t.i∈[2,360]
Separately,
For arbitrfary point P
xif, its f
xvalue is 1, represents that this x is the transition point from background to health, if f
xvalue is 2, represents that this x is the transition point from health to background;
C, the some arrangement being 1 by f value corresponding in 360 points are a sequence l
1, l
2..., l
m1, wherein m
1be the number of the point of 1 for f value, the point numbering in p sequence of the value of element corresponding to it in this sequence, the some arrangement being 2 by f value corresponding in 360 points is a sequence
D, ordered series of numbers angle is set
1, angle
2..., angle
m, wherein
i ∈ [1, m], wherein angle
ibe i-th colour of skin angle, m is the number of colour of skin angle, in like manner arranges ordered series of numbers
wherein
i ∈ [2, m], wherein
be i-th non-colour of skin angle.Then angle sequence and angle` sequence being together as first eigenwert, in like manner, is r at radius successively
2, r
3..., r
ncircle on do identical operation, second, the 3rd can be obtained until the n-th eigenwert;
E, finally this n eigenwert is together as the image that this time collects body move make eigenwert, i.e. local feature region.
The method of described skin colorimetric detection is:
A, be first YCbCr form by following formula from rgb format conversion by image,
Y＝(0.299*R+0.587*G+0.114*B)
Cb＝((B-Y)*0.564+128)
Cr=((R-y) * 0.713+128), wherein R, G, B represents the value of red, green, blue three colourities of pixel respectively, to some pixels, if it belongs to set { (Y, Cb, Cr) Shu Cb ∈ [129,78], Cr ∈ [134,172] }, then think that it is skin pixels point;
B, the skin pixels of image point has been extracted after, noise reduction process is carried out to image;
C, for arbitrfary point (X, Y), judge whether this point is skin according to following formula:
t(x,y)＝(Not?t(x-1,Y)AND?Not?t(X+1,y))OR(Not?t(x,y-1)AND(x,y+1))。
In technical scheme of the present invention, simple body moves that to make recognizer calculated amount little, is suitable for the mobile terminal device being difficult to provide intensive.
Compared with " being applied to the AIWI mobile phone body sense recognition technology of game ", it is wider that the present invention can identify, the motion of human body and the motion of mobile phone can be realized to be separated, the identification with gesture can be applied, and " being applied to the AIWI mobile phone body sense recognition technology of game " cannot to accomplish.
Compared with " being applied to the body sense recognition technology of the equipment such as kinect and Xbox360 of Microsoft ", the present invention is without the need to buying extra equipment, and cost is lower, as long as and mobile phone with oneself can realize the application of " body perception is other ", very convenient to carry.
Accompanying drawing explanation
The block diagram of a kind of body sense recognition system based on camera that Fig. 1 provides for the embodiment of the present invention.
Embodiment
Below by specific embodiment, technical scheme of the present invention is described in detail.
Embodiment 1
As shown in Figure 1, the present embodiment first aspect provides a kind of body sense recognition system based on camera: comprise camera, the described body sense recognition system based on camera also comprises identification module, and the body that described identification module is used for identification camera picked-up moves the classification done.
Body sense recognition system based on camera also comprises the display module of the recognition result transmitted for Identification display module.
Identification module comprises area-of-interest setup unit, body sense recognition unit, and area-of-interest setup unit is connected to display module through body sense recognition unit.
Identification module also comprises result synthesis unit, and area-of-interest setup unit is connected to display module through body sense recognition unit, synthesis unit.
The present embodiment provides a kind of body sense recognition methods based on camera on the other hand, comprises the following steps:
A, startup camera;
B, camera receptor move makes original image, and passes to identification module by moved for body as original image;
Moved for body original image of doing mates as library file with body is moved by C, identification module, draws recognition result, and recognition result is presented at display module.
Described step B is:
Camera receptor moves makes original image, and passes to area-of-interest setup unit by moved for the body of picked-up as original image.
Step C comprises following step:
Body is moved the part cutting being positioned at area-of-interest as original image by C1, area-of-interest setup unit, and by the image transfer of cutting to body sense recognition unit;
The image received from area-of-interest setup unit is carried out skin colorimetric detection and local feature point extraction by C2, body sense recognition unit;
The local feature region extracted is converted into proper vector by descriptor by C3, body sense recognition unit;
Described proper vector and body are moved the moved proper vector done of each individuality done in library file and mate by C4, body sense recognition unit, draw recognition result;
The position of the area-of-interest at described recognition result place and size are fed back to area-of-interest setup unit by C5, body sense recognition unit;
Recognition result is sent to display module and shows by C6, body sense recognition unit.
The recognition result wherein drawn in step C4 moves for certain is individual to be done.
Step C4 comprises following step:
Described proper vector and body emotion are made each individuality in library file and are moved the calculating that the proper vector done carries out colour of skin angle number distance by C41, body sense recognition unit, calculate formula as follows
Dis
angleNum=∑
i ∈ [1, n]shu c
i-fc
ishu
2
Wherein, n moves the characteristic number doing to have for certain in body sense maneuver library is individual, c
iand fc
ibe respectively the colour of skin angle number of i-th feature that the image that collects specifically has and this standard body and move the colour of skin angle number that work i-th feature have, if dis
angleNumbe less than given threshold value, threshold value is greater than 3, is less than 15, then think that it may move with this body and coincide, and moved for this body coincide work is carried out record;
The moved proper vector done of the standard body that proper vector and step C41 filter out by C42, body sense recognition unit is carried out distance and is calculated, and formula is as follows:
Dis
angle=∑
i ∈ [1, n]∑
j ∈ [1, m]shu a
ij-fa
ijshu
2
Wherein n makes for this body in body sense maneuver library moves the characteristic number that has, and m is then its colour of skin angle number, a
ijand fa
ijthen be respectively that the image that collects specifically has and standard body moves the value of a jth colour of skin angle under i-th feature done.If dis
anglebe less than given threshold value, threshold value is greater than 3, is less than 15, then think that it may move with this body and coincide, and moved for this body coincide work is carried out record;
Proper vector is moved with first two standard bodies filtered out the proper vector done and carries out mating of non-colour of skin angle by C43, body sense recognition unit, and its formula calculating distance is as follows:
Dis
nonAngle=∑
i ∈ [1, n]∑
j ∈ [1, m]shu na
ij-fna
ijshu
2
Wherein n makes for this body in body sense maneuver library moves the characteristic number that has, and m is then its non-colour of skin angle number, na
ijand fna
ijbe respectively that image has and standard body moves the value of jth non-colour of skin angle under i-th feature done.After the distance of all non-colour of skin angles all calculates, choose the shortest standard body of its middle distance and move work, be set to current matching result and sent to result synthesis unit.Step C6 comprises following step:
Recognition result is sent to result synthesis unit by C61, body sense recognition unit;
The area-of-interest at the recognition result place received moves at body by C62, result synthesis unit to be done original image indicates, and marks out the code of recognition result on the moved area-of-interest doing recognition result place in original image of body;
The display screen that body emotion is sent to mobile device as original image shows by C63, result synthesis unit.The method to set up of described area-of-interest is: when first time arranges area-of-interest, acquiescence area-of-interest covers the whole image collected, when arranging afterwards, can adjust according to the position of the area-of-interest at the described recognition result place of body sense recognition unit feedback and size at every turn.
The method that described local feature region extracts is:
A, by after Face Detection, the image received can be carried out masking-out process by body sense feature identification module;
B, with the center of source images for n concentric circles is done in the center of circle, n is characteristic number, and n is greater than 3, is less than 7, and maximum concentric circles radius is the distance r that source images border is arrived in the center of circle
m, n concentrically ringed radius is followed successively by, r
1, r
2..., r
n, and meet
now with r
1for on the circle of radius, equidistant searching 360 points, are labeled as p
1, p
2..., p
360, its value is the gray-scale value of the pixel of this position, arranges ordered series of numbers f simultaneously
1, f
2..., f
360represent the feature of these 360 points, wherein
s.t.i∈[2,360]
Separately,
For arbitrfary point P
xif, its f
xvalue is 1, represents that this x is the transition point from background to health, if f
xvalue is 2, represents that this x is the transition point from health to background;
C, the some arrangement being 1 by f value corresponding in 360 points are a sequence l
1, l
2..., l
m1, wherein m
1be the number of the point of 1 for f value, the point numbering in p sequence of the value of element corresponding to it in this sequence, the some arrangement being 2 by f value corresponding in 360 points is a sequence
D, ordered series of numbers angle is set
1, angle
2..., angle
m, wherein
i ∈ [1, m], wherein angle
ibe i-th colour of skin angle, m is the number of colour of skin angle, in like manner arranges ordered series of numbers
wherein
i ∈ [2, m], wherein
be i-th non-colour of skin angle.Then angle sequence and angle` sequence being together as first eigenwert, in like manner, is r at radius successively
2, r
3..., r
ncircle on do identical operation, second, the 3rd can be obtained until the n-th eigenwert;
E, finally this n eigenwert is together as the image that this time collects body move make eigenwert, i.e. local feature region.
The method of described skin colorimetric detection is:
A, be first YCbCr form by following formula from rgb format conversion by image,
Y＝(0.299*R+0.587*G+0.114*B)
Cb＝((B-Y)*0.564+128)
Cr=((R-y) * 0.713+128), wherein R, G, B represents the value of red, green, blue three colourities of pixel respectively, to some pixels, if it belongs to set { (Y, Cb, Cr) Shu Cb ∈ [129,78], Cr ∈ [134,172] }, then think that it is skin pixels point;
B, the skin pixels of image point has been extracted after, noise reduction process is carried out to image;
C, for arbitrfary point (X, Y), judge whether this point is skin according to following formula:
t(x,y)＝(Not?t(x-1,Y)AND?Not?t(X+1,y))OR(Not?t(x,y-1)AND(x,y+1))。
The described body sense recognition system based on camera also comprises timer, and described timer is for setting the shooting interval time of camera.System starts timer, and transmits beginning timing signal to it.0.3 second (adjustable parameter, depending on the arithmetic capability of hardware device, 0.3 second is set in this test) after, timer can to camera transmission " shooting " signal of mobile device, and reset the time of oneself timer inside simultaneously, be triggered after the time of wait set by next time terminates.
After camera executes shoot function (shooting be a pictures), collected original image frame can be passed to area-of-interest (ROI, Region Of Interesting) setup unit (Main Function is screening area-of-interest).In the present system, first time is when arranging region-of-interest, give tacit consent to it and cover the whole image collected, when arranging afterwards, the feedback information (body moves the area-of-interest doing original image place) that all can transmit according to body sense recognition unit adjusts at every turn.Image in area-of-interest can be carried out cutting and pass to body sense recognition unit by area-of-interest setup unit.The image received can be carried out local feature region extraction by body sense recognition unit, and be translated into proper vector by Feature Descriptor (descriptor), and mate with body sense maneuver library, after moving by calculating and each body the vector distance done between original image, moved result as this identification of body (recognition result is that certain individuality moves work) that selected distance is the shortest.After the identification of body sense recognition unit, the body that this can be identified moves the area-of-interest doing place and feeds back to area-of-interest setting module, facilitate it to the correction of area-of-interest, also result (certain individual moved work) can be sent to result synthesis module simultaneously.The body received emotion can be made place area-of-interest red rectangle frame and iris out on original image frame by result synthesis module, and marking out the current result identified on the region of interest for the moved work of which kind of body, the display screen finally the synthesis result processed being sent to mobile device shows.
When module first time is called, can be loaded into the body sense maneuver library from server, after analyzing it, the number and each body that therefrom read the moved work of body move the feature (high dimension vector) done.Subsequently, when receiving the image transmitted from area-of-interest setup unit at every turn, skin colorimetric detection and posture feature extraction can be carried out to this image, position and the size information that also occlusion body can be moved the area-of-interest done detect simultaneously, and feed back to area-of-interest setup unit.
The present invention effectively can solve the how more effective problem merged with android and ios platform of body sensing system.
Claims (13)
1. based on a body sense recognition system for camera, comprise camera, it is characterized in that: the described body sense recognition system based on camera also comprises identification module, the body that described identification module is used for identification camera picked-up moves the classification done.
2. a kind of body sense recognition system based on camera according to claim 1, it is characterized in that, the described body sense recognition system based on camera also comprises the display module of the recognition result transmitted for Identification display module.
3. a kind of body sense recognition system based on camera according to claim 2, it is characterized in that, described identification module comprises area-of-interest setup unit, body sense recognition unit, and area-of-interest setup unit is connected to display module through body sense recognition unit.
4. a kind of body sense recognition system based on camera according to claim 3, it is characterized in that, described identification module also comprises result synthesis unit, and area-of-interest setup unit is connected to display module through body sense recognition unit, synthesis unit.
5., based on a body sense recognition methods for camera, comprise the following steps:
A, startup camera;
B, camera receptor move makes original image, and passes to identification module by moved for body as original image;
Moved for body original image of doing mates as library file with body is moved by C, identification module, draws recognition result, and recognition result is presented at display module.
6. a kind of body sense recognition methods based on camera according to claim 5, described step B is:
Camera receptor moves makes original image, and passes to area-of-interest setup unit by moved for the body of picked-up as original image.
7. a kind of body sense recognition methods based on camera according to claim 6, step C comprises following step:
Body is moved the part cutting being positioned at area-of-interest as original image by C1, area-of-interest setup unit, and by the image transfer of cutting to body sense recognition unit;
The image received from area-of-interest setup unit is carried out skin colorimetric detection and local feature point extraction by C2, body sense recognition unit;
The local feature region extracted is converted into proper vector by descriptor by C3, body sense recognition unit;
Described proper vector and body are moved the moved proper vector done of each individuality done in library file and mate by C4, body sense recognition unit, draw recognition result;
The position of the area-of-interest at described recognition result place and size are fed back to area-of-interest setup unit by C5, body sense recognition unit;
Recognition result is sent to display module and shows by C6, body sense recognition unit.
8. a kind of body sense recognition methods based on camera according to claim 7, the recognition result wherein drawn in step C4 moves for certain is individual to be done.
9. a kind of body sense recognition methods based on camera according to claim 7, step C4 comprises following step:
Described proper vector and body emotion are made each individuality in library file and are moved the calculating that the proper vector done carries out colour of skin angle number distance by C41, body sense recognition unit, calculate formula as follows
Dis
angleNum=∑
i ∈ [1, n]shu c
i-fc
ishu
2
Wherein, n moves the characteristic number doing to have for certain in body sense maneuver library is individual, c
iand fc
ibe respectively the colour of skin angle number of i-th feature that the image that collects specifically has and this standard body and move the colour of skin angle number that work i-th feature have, if dis
angleNumbe less than given threshold value, threshold value is greater than 3, is less than 15, then think that it may move with this body and coincide, and moved for this body coincide work is carried out record;
The moved proper vector done of the standard body that proper vector and step C41 filter out by C42, body sense recognition unit is carried out distance and is calculated, and formula is as follows:
Dis
angle=∑
i ∈ [1, n]∑
j ∈ [1, m]shu a
ij-fa
ijshu
2
Wherein n makes for this body in body sense maneuver library moves the characteristic number that has, and m is then its colour of skin angle number, a
ijand fa
ijthen be respectively that the image that collects specifically has and standard body moves the value of a jth colour of skin angle under i-th feature done.If dis
anglebe less than given threshold value, threshold value is greater than 3, is less than 15, then think that it may move with this body and coincide, and moved for this body coincide work is carried out record;
Proper vector is moved with first two standard bodies filtered out the proper vector done and carries out mating of non-colour of skin angle by C43, body sense recognition unit, and its formula calculating distance is as follows:
Dis
nonAngle=∑
i ∈ [1, n]∑
j ∈ [1, m]shu na
ij-fna
ijshu
2
Wherein n makes for this body in body sense maneuver library moves the characteristic number that has, and m is then its non-colour of skin angle number, na
ijand fna
ijbe respectively that image has and standard body moves the value of jth non-colour of skin angle under i-th feature done.After the distance of all non-colour of skin angles all calculates, choose the shortest standard body of its middle distance and move work, be set to current matching result and sent to result synthesis unit.
10. a kind of body sense recognition methods based on camera according to claim 7, step C6 comprises following step:
Recognition result is sent to result synthesis unit by C61, body sense recognition unit;
The area-of-interest at the recognition result place received moves at body by C62, result synthesis unit to be done original image indicates, and marks out the code of recognition result on the moved area-of-interest doing recognition result place in original image of body;
The display screen that body emotion is sent to mobile device as original image shows by C63, result synthesis unit.
11. a kind of body sense recognition methodss based on camera according to claim 7, it is characterized in that, the method to set up of described area-of-interest is: when first time arranges area-of-interest, acquiescence area-of-interest covers the whole image collected, when arranging afterwards, can adjust according to the position of the area-of-interest at the described recognition result place of body sense recognition unit feedback and size at every turn.
12. a kind of body sense recognition methodss based on camera according to claim 7, is characterized in that, the method that described local feature region extracts is:
A, by after Face Detection, the image received can be carried out masking-out process by body sense feature identification module;
B, with the center of source images for n concentric circles is done in the center of circle, n is characteristic number, and n is greater than 3, is less than 7, and maximum concentric circles radius is the distance r that source images border is arrived in the center of circle
m, n concentrically ringed radius is followed successively by, r
1, r
2..., r
n, and meet
now with r
1for on the circle of radius, equidistant searching 360 points, are labeled as p
1, p
2..., p
360, its value is the gray-scale value of the pixel of this position, arranges ordered series of numbers f simultaneously
1, f
2..., f
360represent the feature of these 360 points, wherein
s.t.i∈[2,360]
Separately,
For arbitrfary point P
xif, its f
xvalue is 1, represents that this x is the transition point from background to health, if f
xvalue is 2, represents that this x is the transition point from health to background;
C, the some arrangement being 1 by f value corresponding in 360 points are a sequence l
1, l
2..., l
m1, wherein m
1be the number of the point of 1 for f value, the point numbering in p sequence of the value of element corresponding to it in this sequence, the some arrangement being 2 by f value corresponding in 360 points is a sequence
D, ordered series of numbers angle is set
1, angle
2..., angle
m, wherein
i ∈ [1, m], wherein angle
ibe i-th colour of skin angle, m is the number of colour of skin angle, in like manner arranges ordered series of numbers
wherein
i ∈ [2, m], wherein
be i-th non-colour of skin angle.Then angle sequence and angle` sequence being together as first eigenwert, in like manner, is r at radius successively
2, r
3..., r
ncircle on do identical operation, second, the 3rd can be obtained until the n-th eigenwert;
E, finally this n eigenwert is together as the image that this time collects body move make eigenwert, i.e. local feature region.
13. a kind of body sense recognition methodss based on camera according to claim 7, it is characterized in that, the method for described skin colorimetric detection is:
A, be first YCbCr form by following formula from rgb format conversion by image,
Y＝(0.299*R+0.587*G+0.114*B)
Cb＝((B-Y)*0.564+128)
Cr=((R-y) * 0.713+128), wherein R, G, B represents the value of red, green, blue three colourities of pixel respectively, to some pixels, if it belongs to set { (Y, Cb, Cr) Shu Cb ∈ [129,78], Cr ∈ [134,172] }, then think that it is skin pixels point;
B, the skin pixels of image point has been extracted after, noise reduction process is carried out to image;
C, for arbitrfary point (X, Y), judge whether this point is skin according to following formula:
t(x,y)＝(Not?t(x-1,Y)AND?Not?t(X+1,y))OR(Not?t(x,y-1)AND(x,y+1))。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410492045.5A CN104252231B (en) | 2014-09-23 | 2014-09-23 | A kind of somatosensory recognition system and method based on camera |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410492045.5A CN104252231B (en) | 2014-09-23 | 2014-09-23 | A kind of somatosensory recognition system and method based on camera |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104252231A true CN104252231A (en) | 2014-12-31 |
| CN104252231B CN104252231B (en) | 2017-10-10 |
Family
ID=52187222
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410492045.5A Expired - Fee Related CN104252231B (en) | 2014-09-23 | 2014-09-23 | A kind of somatosensory recognition system and method based on camera |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104252231B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018006481A1 (en) * | 2016-07-04 | 2018-01-11 | 中兴通讯股份有限公司 | Motion-sensing operation method and device for mobile terminal |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101076125A (en) * | 2007-06-18 | 2007-11-21 | 山东经济学院 | Algorithm for optimizing RGB and YCbCr conversion computing in image compression |
| US20090087088A1 (en) * | 2007-09-28 | 2009-04-02 | Samsung Electronics Co., Ltd. | Image forming system, apparatus and method of discriminative color features extraction thereof |
| CN102426480A (en) * | 2011-11-03 | 2012-04-25 | 康佳集团股份有限公司 | Man-machine interactive system and real-time gesture tracking processing method for same |
| CN102508246A (en) * | 2011-10-13 | 2012-06-20 | 吉林大学 | Method for detecting and tracking obstacles in front of vehicle |
| CN103558911A (en) * | 2013-10-24 | 2014-02-05 | 广东欧珀移动通信有限公司 | Mute achievement method and system of mobile terminal |
| CN103646425A (en) * | 2013-11-20 | 2014-03-19 | 深圳先进技术研究院 | A method and a system for body feeling interaction |
-
2014
- 2014-09-23 CN CN201410492045.5A patent/CN104252231B/en not_active Expired - Fee Related
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101076125A (en) * | 2007-06-18 | 2007-11-21 | 山东经济学院 | Algorithm for optimizing RGB and YCbCr conversion computing in image compression |
| US20090087088A1 (en) * | 2007-09-28 | 2009-04-02 | Samsung Electronics Co., Ltd. | Image forming system, apparatus and method of discriminative color features extraction thereof |
| CN102508246A (en) * | 2011-10-13 | 2012-06-20 | 吉林大学 | Method for detecting and tracking obstacles in front of vehicle |
| CN102426480A (en) * | 2011-11-03 | 2012-04-25 | 康佳集团股份有限公司 | Man-machine interactive system and real-time gesture tracking processing method for same |
| CN103558911A (en) * | 2013-10-24 | 2014-02-05 | 广东欧珀移动通信有限公司 | Mute achievement method and system of mobile terminal |
| CN103646425A (en) * | 2013-11-20 | 2014-03-19 | 深圳先进技术研究院 | A method and a system for body feeling interaction |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018006481A1 (en) * | 2016-07-04 | 2018-01-11 | 中兴通讯股份有限公司 | Motion-sensing operation method and device for mobile terminal |
| CN107577334A (en) * | 2016-07-04 | 2018-01-12 | 中兴通讯股份有限公司 | A kind of somatosensory operation method and device of mobile terminal |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104252231B (en) | 2017-10-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10372226B2 (en) | Visual language for human computer interfaces | |
| US8754934B2 (en) | Dual-camera face recognition device and method | |
| CN104395856B (en) | For recognizing the computer implemented method and system of dumb show | |
| US9025024B2 (en) | System and method for object identification and tracking | |
| CN110463176A (en) | Image quality measure | |
| WO2019132589A1 (en) | Image processing device and method for detecting multiple objects | |
| CN107018316B (en) | Image processing apparatus, image processing method, and storage medium | |
| KR20150039252A (en) | Apparatus and method for providing application service by using action recognition | |
| US20150379333A1 (en) | Three-Dimensional Motion Analysis System | |
| WO2009142098A1 (en) | Image processing device, camera, image processing method, and program | |
| CN109493310A (en) | Biological information analytical equipment and its hand skin analysis method | |
| CN105678817A (en) | Method for extracting central point of circular image with high speed | |
| KR101344851B1 (en) | Device and Method for Processing Image | |
| CN108837478B (en) | Foul automatic discrimination system for hurdling competition based on machine vision | |
| CN110266953A (en) | Image processing method, device, server and storage medium | |
| CN110175553A (en) | The method and device of feature database is established based on Gait Recognition and recognition of face | |
| CN107665328A (en) | Face face recognition method | |
| CN104349197B (en) | A kind of data processing method and device | |
| KR101895813B1 (en) | Apparatus and method for object creation augmented reality | |
| CN104252231A (en) | Camera based motion sensing recognition system and method | |
| CN107493431A (en) | A kind of image taking synthetic method, terminal and computer-readable recording medium | |
| CN108521594B (en) | Free viewpoint video playing method based on motion sensing camera gesture recognition | |
| CN115393962A (en) | Motion recognition method, head-mounted display device, and storage medium | |
| JP5539565B2 (en) | Imaging apparatus and subject tracking method | |
| KR101738589B1 (en) | A coaching system for stretching using image recognition |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20171010 Termination date: 20190923 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |