CN108229251A - A kind of action identification method and device - Google Patents
A kind of action identification method and device Download PDFInfo
- Publication number
- CN108229251A CN108229251A CN201611160942.1A CN201611160942A CN108229251A CN 108229251 A CN108229251 A CN 108229251A CN 201611160942 A CN201611160942 A CN 201611160942A CN 108229251 A CN108229251 A CN 108229251A
- Authority
- CN
- China
- Prior art keywords
- value
- match value
- distance
- apart
- area
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
- G06V40/23—Recognition of whole body movements, e.g. for sport training
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
Abstract
The present invention relates to smart home field more particularly to a kind of action identification methods and device, and in order to solve the problems, such as that in the prior art blocked because of camera sight cannot accurately carry out action recognition, this method is：First extract the video image in preset time period, determine the area match value of the face in each frame picture that video image includes and apart from match value, obtain area fitting value set and apart from fitting value set, it is fitted in value set from distance, according to preset rules, first is extracted apart from match value and second distance match value, value set is fitted based on area, determine extraction first apart from match value and second distance match value meet preset condition when, it determines that target action occurs in preset time period, in this way, only need the face of detection behavioral agent, can execution in real time identification, it effectively prevents blocking because of camera sight, and it can not determine behavioral agent four limbs and trunk, and then the problem of accurate action recognition can not be carried out.
Description
Technical field
The present invention relates to smart home field more particularly to a kind of action identification methods and device.
Background technology
With the development of intelligence science and technology, smart home industry is also constantly developing, and in smart home, camera is
It is essential, e.g., the action of the behavioral agent in family activity is identified by camera, to reach indoor security prison
The purpose of control, personal security's monitoring, personal entertainment etc..
Under the prior art, when carrying out action recognition by camera, the video image first provided based on camera is provided, is carried
The feature of the systemic features of behavioral agent, especially trunk and lower limb is taken, then, then by analyzing feature acquisition behavioral agent
Motor pattern resettles the mapping relations between video image content and behavior type, and based on the mapping relations, could carry out
Action recognition.
However, in smart home, action recognition is usually applied to real-time scene, has certain timeliness, moreover,
The action of behavioral agent has certain complexity and diversity under real-time scene, according in the action recognition under the prior art
A series of complex algorithm, carries out action recognition, and timeliness can substantially reduce, it is difficult to which meet smart home applies needs.
Moreover, under the prior art in action identification method, need to extract the systemic features of behavioral agent, especially trunk and
The feature of lower limb, however, in household scene, the scene being blocked for lower limb (e.g., sits down or stands before dining table, in sand
Sit down or stand on hair side), it can not be applicable in.
Therefore, it is necessary to design a kind of new action identification method, to overcome drawbacks described above.
Invention content
The embodiment of the present invention provides a kind of action identification method and device, in the prior art because of camera shooting to solve
Head sight is blocked the problem of cannot accurately carrying out action recognition.
Specific technical solution provided in an embodiment of the present invention is as follows：
A kind of action identification method, including：
Extract the video image in preset time period；
The area match value of the face in each frame picture for including of the video image is determined respectively and apart from match value,
Obtain area fitting value set and apart from fitting value set；
From the distance fitting value set, according to preset rules, extract first and intend apart from match value and second distance
Conjunction value；
Value set is fitted based on the area, determines that described first meets apart from match value and the second distance match value
During preset condition, determine that target action occurs in the preset time period.
Optionally, the area match value of the face in any one frame picture that the video image includes is determined and apart from plan
Conjunction value, including：
Face datection is carried out to any one frame picture that the video image includes, determines human face region；
Based on the human face region, coordinate position of the face in the picture is determined；
Based on the coordinate position, the area value and distance value of face are calculated respectively, and the distance value is face and place
The distance between lower horizontal line of picture；
Based on the area value and the distance value, the corresponding area match value of the area value, Yi Jisuo are determined respectively
It is corresponding apart from match value to state distance value.
Optionally, it is described from the distance fitting value set, according to preset rules, extract first apart from match value and
Second distance match value, including：
Each in the distance fitting value set is recorded apart from match value, is formed discrete apart from match value curve；
Determine a pair of of the wave crest and trough nearest with the starting point of the preset time period；
The wave crest and the trough are extracted, and determines the wave crest for described first apart from match value, Yi Jisuo respectively
Trough is stated as the second distance match value.
Optionally, value set is fitted based on the area, determines that described first intends apart from match value and the second distance
When conjunction value meets preset condition, determine that target action occurs in the preset time period, including：
Determine that described first is corresponding apart from match value corresponding first frame moment and the second distance match value respectively
Second frame moment；
It is fitted in value set in the area, extracts the second frame moment corresponding first area match value；
The absolute difference between the second frame moment and the first frame moment is judged, positioned at the preset first frame moment
Between threshold value and preset second frame moment threshold value；Also,
The second distance match value and the described first absolute difference between match value are judged, positioned at twice of institute
When stating between the first area match value and the first area match value of three times；
It determines that target action occurs in the preset time period.
Optionally, it determines after target action occurs in the preset time period, further includes：
If the second distance match value is higher than described first apart from match value, it is determined that the target action is station
It rises；
If the second distance match value is less than described first apart from match value, it is determined that the target action is sits
Under.
A kind of action recognition device, including：
First extraction unit, for extracting the video image in preset time period；
First determination unit, the area for determining the face in each frame picture that the video image includes respectively are intended
Conjunction value and apart from match value, obtains area fitting value set and apart from fitting value set；
Second extraction unit, for from the distance fitting value set, according to preset rules, extracting the first distance and intending
Conjunction value and second distance match value；
Second determination unit for being based on area fitting value set, determines described first apart from match value and described
When second distance match value meets preset condition, determine that target action occurs in the preset time period.
Optionally, the area match value of the face in any one frame picture that the video image includes is determined and apart from plan
During conjunction value, first determination unit is used for：
Face datection is carried out to any one frame picture that the video image includes, determines human face region；
Based on the human face region, coordinate position of the face in the picture is determined；
Based on the coordinate position, the area value and distance value of face are calculated respectively, and the distance value is face and place
The distance between lower horizontal line of picture；
Based on the area value and the distance value, the corresponding area match value of the area value, Yi Jisuo are determined respectively
It is corresponding apart from match value to state distance value.
Optionally, it is described from the distance fitting value set, according to preset rules, extract first apart from match value and
During second distance match value, second extraction unit is used for：
Each in the distance fitting value set is recorded apart from match value, is formed discrete apart from match value curve；
Determine a pair of of the wave crest and trough nearest with the starting point of the preset time period；
The wave crest and the trough are extracted, and determines the wave crest for described first apart from match value, Yi Jisuo respectively
Trough is stated as the second distance match value.
Optionally, value set is fitted based on the area, determines that described first intends apart from match value and the second distance
When conjunction value meets preset condition, when determining to occur in the preset time period target action, second determination unit is used for：
Determine that described first is corresponding apart from match value corresponding first frame moment and the second distance match value respectively
Second frame moment；
It is fitted in value set in the area, extracts the second frame moment corresponding first area match value；
The absolute difference between the second frame moment and the first frame moment is judged, positioned at the preset first frame moment
Between threshold value and preset second frame moment threshold value；Also,
The second distance match value and the described first absolute difference between match value are judged, positioned at twice of institute
When stating between the first area match value and the first area match value of three times；
It determines that target action occurs in the preset time period.
Optionally, it determines after target action occurs in the preset time period, second determination unit is additionally operable to：
If the second distance match value is higher than described first apart from match value, it is determined that the target action is station
It rises；
If the second distance match value is less than described first apart from match value, it is determined that the target action is sits
Under.
In the embodiment of the present invention, the video image in preset time period is first extracted, determines each frame that video image includes
The area match value of face in picture and apart from match value obtains area fitting value set and apart from fitting value set, from away from
From in fitting value set, according to preset rules, first is extracted apart from match value and second distance match value, is fitted based on area
Value set, determine extraction first apart from match value and second distance match value meet preset condition when, determine in preset time
Target action occurs in section, in this way, only needing the face of detection behavioral agent, so that it may which real-time execution identification has
Effect is avoided because camera sight is blocked, and can not determine behavioral agent four limbs and trunk, and then can not accurately be acted
The problem of identification.
Description of the drawings
Fig. 1 is motion recognition system structure chart in the embodiment of the present invention；
Fig. 2 is action identification method flow chart in the embodiment of the present invention；
Fig. 3 is example coordinate position figure in the embodiment of the present invention
Fig. 4 is exemplary graph in the embodiment of the present invention
Action recognition device figure in the embodiment of the present invention of Fig. 5 positions.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present invention, the technical solution in the embodiment of the present invention is carried out clear, complete
Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, is not whole embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art are obtained every other without making creative work
Embodiment shall fall within the protection scope of the present invention.
In order to solve under real-time scene, the timeliness that the prior art carries out behavioral agent action recognition is low, Yi Ji
In household scene, the scene (e.g., sit down or stand before dining table, sit down or stand on sofa side) that is blocked for lower limb,
The problem of action recognition can not accurately be carried out, in the embodiment of the present invention, devises a kind of method of action recognition, and this method is,
The video image in preset time period is first extracted, determines the area match value of the face in each frame picture that video image includes
With apart from match value, obtain area fitting value set and apart from fitting value set, be fitted in value set from distance, according to default rule
Then, first is extracted apart from match value and second distance match value, and value set is fitted based on area, determines the first distance of extraction
When match value and second distance match value meet preset condition, determine that target action occurs in preset time period.
As shown in fig.1, in the embodiment of the present invention, area to be monitored is monitored in real time using intelligent video camera head, and
Video image in area to be monitored is transferred to human motion analysis platform in real time, human motion analysis platform is to receiving
Video image analyzed and processed in real time, with identify be located at area to be monitored in behavioral agent occur target move
Make, wherein, video image can be transferred to human motion analysis by intelligent video camera head by the form of wireless network or cable network
At platform.
Specifically, as shown in fig.2, human motion analysis platform carries out action recognition according to the video image received
Detailed process it is as follows：
Step 200：Video image in human motion analysis platform extraction preset time period.
Specifically, human motion analysis platform, according to preset time period, extraction is located at the video shadow in the preset time period
Picture.
Preset time period can be adjusted according to actual demand, if for example, in the video image in half an hour
The action of behavioral agent is identified, then, human motion analysis platform can extract the video image in half an hour, if right
The action of the behavioral agent in video image in 10 minutes is identified, then, human motion analysis platform can extract 10
Video image in minute.
Preset time period can also be divided according to the ability of human motion analysis platform processes video image, for example,
If the capacity that human motion analysis platform disposably handles video image is 1T, according to the resolution for the video image that will be extracted
Rate calculates the period S that can be accommodated, and using period S as preset time period.
In the embodiment of the present invention, the setting means of preset time period is not limited.
Step 210：Human motion analysis platform carries out face to each frame picture included in the video image that extracts
Detection, determines human face region.
Specifically, the video image that human motion analysis platform will extract, according to frame format, splits into one by one
Picture, and Face datection algorithm is used, Face datection is carried out, and in each frame picture to detecting face to each frame picture
Face be labeled, determine human face region, preferably, in the embodiment of the present invention, face be labeled using box.
Step 220：Human motion analysis platform determines face in respective picture based on the human face region in each frame picture
Coordinate position.
Specifically, human motion analysis platform is in each frame picture of the face detected, it has been determined that human face region, base
The human face region determined in each frame picture determines the coordinate position of the face in each frame picture, referring particularly to Fig. 3 institutes
Show.
Such as, it has been determined that face is detected in T frame pictures, and has determined that human face region, that is, small box is drawn with T frames
The left upper apex in face is coordinate points, establishes coordinate system, determines the coordinate position on each vertex comprising human face region (small box),
The set of coordinates of the small box is { X1T, X2T, Y1T, Y2TAnd determine the picture maximum width XfWith maximum height Yf。
Step 230：Coordinate position of the human motion analysis platform based on face in each frame picture calculates face respectively
Area value and distance value, wherein, distance value is the distance between face and the lower horizontal line of place picture.
Specifically, human motion analysis platform has determined that the coordinate position of face in each frame picture for detect face,
Coordinate position based on face calculates the area value of face and the distance value of face respectively.
Continue to continue to use above-mentioned example, as shown in fig.3, calculating the area value and distance value of face in T frame pictures, preferably
, in the embodiment of the present invention, the area value S for obtaining face in T frame pictures is calculated using equation belowT：
ST=Y2T-Y1T
Wherein, Y2TAnd Y1TIt is coordinate value of the human face region on Y coordinate in T frame pictures.
Preferably, in the embodiment of the present invention, the distance value H for obtaining face is calculated using equation belowT：
HT=Yf-(Y1T+Y2T)/2
Wherein, Y2TAnd Y1TIt is coordinate value of the human face region on Y coordinate, Y in T frame picturesfFor T frame pictures
Maximum height value.
Step 240：Human motion analysis platform is based on the area value and distance value of face obtained in each frame picture, really
Determine corresponding area match value and apart from match value, and obtain area and be fitted value set and apart from fitting value set.
Specifically, in step 230, in each frame picture that human motion analysis platform obtains the area value of face and away from
It from value, is determined when the previous frame moment based on corresponding, however, behavioral agent is performing an action, e.g., squatted
Under, it is typically successional, therefore, when the area value and distance value of face in the corresponding picture at previous frame moment, usually
The area value of face and distance value or even picture corresponding with the first two frames moment in picture also corresponding with the previous frame moment
The area value of middle face is related with distance value.
Further, after the area value and distance value of face of a certain frame picture is obtained, with reference to the previous of the picture
The area value and distance value of face in frame moment corresponding picture, in first two frames moment corresponding picture the area value of face and
The area value of face and distance value calculate in distance value and first three frame moment corresponding picture, obtain face in the picture
Area match value and apart from match value.
Preferably, in the embodiment of the present invention, still by taking T frame pictures as an example, corresponded to using the equation below acquisition T frame moment
Picture in face apart from match value HCT：
Wherein, HTFor the distance value of face in T frame moment corresponding picture, HT-1For (T-1) frame moment corresponding picture
The distance value of face, H in the corresponding picture of previous frame moment of the distance value of face in face, i.e. T frames momentT-2For (T-
2) in frame moment corresponding picture face distance value, i.e., face in the first two frames moment corresponding picture at T frames moment
Distance value, HT-3First three frame moment pair for the distance value of face in (T-3) frame moment corresponding picture, i.e. T frames moment
The distance value of face in the picture answered.
Preferably, in the embodiment of the present invention, still by taking T frame pictures as an example, corresponded to using the equation below acquisition T frame moment
Picture in face area match value SCT：
Wherein, STFor the area value of the face in T frame moment corresponding picture, SCT-1It is corresponded to for (T-1) frame moment
Picture in face apart from match value, i.e., the area value of face in the corresponding picture of previous frame moment at T frames moment.
So far, by above-mentioned steps, being fitted apart from match value and area for face in T frame moment corresponding picture is obtained
Value, due to it is above-mentioned be by taking T frame pictures as an example, only obtain intending apart from match value and area for face in T frame pictures
Conjunction value, for having determined that each frame picture there are face, using the respective face of aforesaid way acquisition apart from match value
With area match value, and the several of acquisition are formed into distance fitting value set apart from match value, several areas of acquisition are fitted
Value composition area fitting value set.
For example, it is assumed that currently determine to detect face in 8-10 frame pictures, if the face in the 8th frame picture away from
It is respectively HC from match value and area match value8And SC8If face in the 9th frame picture apart from match value and area match value
Respectively HC9And SC9If face in the 10th frame picture apart from match value and area match value be respectively HC10And SC10, that
, the distance fitting value set of acquisition is { HC8、HC9、HC10, the area fitting value set of acquisition is { SC8、SC9、SC10}。
Step 250：Distance fitting value set of the human motion analysis platform based on acquisition, according to preset rules, extracts
First apart from match value and second distance match value.
Specifically, each in the distance fitting value set that human motion analysis platform record obtains is apart from match value,
Form a pair of of wave crest and wave discrete apart from match value curve, and that determine nearest with the starting point of preset time period in the curve
Paddy extracts the wave crest and trough, and the wave crest is determined as first apart from match value, which is determined as second distance and is intended
Conjunction value.
For example, it is assumed that in step 210, video image is split as m frame pictures, the corresponding distance of m frame pictures is intended
Conjunction value is respectively：HCT-m+1, HCT-m+2..., HCT-2, HCT-1, HCT, discrete distance is formed apart from match value be fitted above-mentioned m
It is worth curve, wherein, HCTThe corresponding frame moment is the starting point of preset time period, referring particularly to shown in Fig. 4.
Further, it is determined that at this in match value curve, nearest a pair of of wave crest and trough with the T frame moment are corresponding
In Fig. 4, it may be determined that (T-1) frame moment and (T-2) frame moment it is corresponding apart from match value be nearest a pair of of wave crest and wave
Paddy, that is, determine that (T-1) frame moment is corresponding apart from match value HCT-1For wave crest, the corresponding distance fitting of (T-2) frame moment is determined
Value HCT-2For trough.
Step 260：Area fitting value set of the human motion analysis platform based on acquisition, determines that the first distance of extraction is intended
When conjunction value and second distance match value meet preset condition, determine that target action occurs in preset time period.
Specifically, first is determined respectively apart from the match value corresponding first frame moment and second distance match value corresponding the
Two frame moment, and value set is fitted in the area of acquisition, extract the second frame moment corresponding first area match value, judgement the
Absolute difference between two frame moment and first frame moment, positioned at preset first frame moment threshold value and preset second frame moment
Between threshold value, also, second distance match value and the first absolute difference between match value are judged, positioned at described in twice
When between the first area match value and the first area match value of three times, determine that target occurs in preset time period to be moved
Make.
In the embodiment of the present invention, preset first frame moment threshold value and the second frame moment threshold value are to choose human body to perform mesh
Empirical value when mark acts, have been trained for a long time acquisition, equally, is judging second distance match value and the first distance fitting
During absolute difference between value, " twice " of selection and " three times " are also what is obtained when prolonged exercise performance objective acts, are
It is dynamically adapted.
Preferably, in the embodiment of the present invention, determine whether target action occurs in preset time period using equation below：
tmin≤|L-K|≤tmax；Also,
2×SCL≤|HCL-HCK|≤3×SCL
Wherein, in formula one, tminFor preset first frame moment threshold value, tmaxFor preset second frame moment threshold value, K
For wave crest (first apart from match value) the corresponding frame moment, L is trough (second distance match value) the corresponding frame moment.
Wherein, in formula two, HCKFor first apart from match value, HCLFor second distance match value, SCLFor the first area
Match value.
Further, it is determined that after target action occurs for behavioral agent, second distance match value and the first distance can be passed through
Relationship between match value further judges that target action stands or to sit down, if second distance match value is higher than first
During apart from match value, determine target action to stand；If second distance match value is less than first apart from match value, target is determined
It acts to sit down.
Preferably, in the embodiment of the present invention, determine after target action occurs, further judge that target is moved using equation below
As standing or sit down：
Wherein, HCKFor first apart from match value, HCLFor second distance match value.
As shown in fig.5, in the embodiment of the present application, action recognition device includes at least the first extraction unit 50, first really
Order member 51, the second extraction unit 52 and the second determination unit 53, wherein,
First extraction unit 50, for extracting the video image in preset time period；
First determination unit 51, for determining the area of the face in each frame picture that the video image includes respectively
Match value and apart from match value, obtains area fitting value set and apart from fitting value set；
Second extraction unit 52, for from the distance fitting value set, according to preset rules, extracting the first distance
Match value and second distance match value；
Second determination unit 53 for being based on the area fitting value set, determines described first apart from match value and institute
When stating second distance match value and meeting preset condition, determine that target action occurs in the preset time period.
Optionally, the area match value of the face in any one frame picture that the video image includes is determined and apart from plan
During conjunction value, first determination unit 51 is used for：
Face datection is carried out to any one frame picture that the video image includes, determines human face region；
Based on the human face region, coordinate position of the face in the picture is determined；
Based on the coordinate position, the area value and distance value of face are calculated respectively, and the distance value is face and place
The distance between lower horizontal line of picture；
Based on the area value and the distance value, the corresponding area match value of the area value, Yi Jisuo are determined respectively
It is corresponding apart from match value to state distance value.
Optionally, it is described from the distance fitting value set, according to preset rules, extract first apart from match value and
During second distance match value, second extraction unit 52 is used for：
Each in the distance fitting value set is recorded apart from match value, is formed discrete apart from match value curve；
Determine a pair of of the wave crest and trough nearest with the starting point of the preset time period；
The wave crest and the trough are extracted, and determines the wave crest for described first apart from match value, Yi Jisuo respectively
Trough is stated as the second distance match value.
Optionally, value set is fitted based on the area, determines that described first intends apart from match value and the second distance
When conjunction value meets preset condition, when determining to occur in the preset time period target action, second determination unit 53 is used
In：
Determine that described first is corresponding apart from match value corresponding first frame moment and the second distance match value respectively
Second frame moment；
It is fitted in value set in the area, extracts the second frame moment corresponding first area match value；
The absolute difference between the second frame moment and the first frame moment is judged, positioned at the preset first frame moment
Between threshold value and preset second frame moment threshold value；Also,
The second distance match value and the described first absolute difference between match value are judged, positioned at twice of institute
When stating between the first area match value and the first area match value of three times；
It determines that target action occurs in the preset time period.
Optionally, it determines after target action occurs in the preset time period, second determination unit 53 is also used
In：
If the second distance match value is higher than described first apart from match value, it is determined that the target action is station
It rises；
If the second distance match value is less than described first apart from match value, it is determined that the target action is sits
Under.
In the embodiment of the present invention, the video image in preset time period is first extracted, determines each frame that video image includes
The area match value of face in picture and apart from match value obtains area fitting value set and apart from fitting value set, from away from
From in fitting value set, according to preset rules, first is extracted apart from match value and second distance match value, is fitted based on area
Value set, determine extraction first apart from match value and second distance match value meet preset condition when, determine in preset time
Target action occurs in section, in this way, only needing the face of detection behavioral agent, so that it may which real-time execution identification has
Effect is avoided because camera sight is blocked, and can not determine behavioral agent four limbs and trunk, and then can not accurately be acted
The problem of identification, meanwhile, it also effectively prevents because the height difference and camera of behavioral agent are far and near, and then standard can not be carried out
The problem of true action recognition.
It should be understood by those skilled in the art that, the embodiment of the present invention can be provided as method, system or computer program
Product.Therefore, the reality in terms of complete hardware embodiment, complete software embodiment or combination software and hardware can be used in the present invention
Apply the form of example.Moreover, the computer for wherein including computer usable program code in one or more can be used in the present invention
The computer program production that usable storage medium is implemented on (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.)
The form of product.
The present invention be with reference to according to the method for the embodiment of the present invention, the flow of equipment (system) and computer program product
Figure and/or block diagram describe.It should be understood that it can be realized by computer program instructions every first-class in flowchart and/or the block diagram
The combination of flow and/or box in journey and/or box and flowchart and/or the block diagram.These computer programs can be provided
The processor of all-purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices is instructed to produce
A raw machine so that the instruction performed by computer or the processor of other programmable data processing devices is generated for real
The device of function specified in present one flow of flow chart or one box of multiple flows and/or block diagram or multiple boxes.
These computer program instructions, which may also be stored in, can guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works so that the instruction generation being stored in the computer-readable memory includes referring to
Enable the manufacture of device, the command device realize in one flow of flow chart or multiple flows and/or one box of block diagram or
The function of being specified in multiple boxes.
These computer program instructions can be also loaded into computer or other programmable data processing devices so that counted
Series of operation steps are performed on calculation machine or other programmable devices to generate computer implemented processing, so as in computer or
The instruction offer performed on other programmable devices is used to implement in one flow of flow chart or multiple flows and/or block diagram one
The step of function of being specified in a box or multiple boxes.
Although preferred embodiments of the present invention have been described, but those skilled in the art once know basic creation
Property concept, then additional changes and modifications may be made to these embodiments.So appended claims be intended to be construed to include it is excellent
It selects embodiment and falls into all change and modification of the scope of the invention.
Obviously, those skilled in the art can carry out the embodiment of the present invention various modification and variations without departing from this hair
The spirit and scope of bright embodiment.In this way, if these modifications and variations of the embodiment of the present invention belong to the claims in the present invention
And its within the scope of equivalent technologies, then the present invention is also intended to include these modifications and variations.
Claims (10)
1. a kind of action identification method, which is characterized in that including：
Extract the video image in preset time period；
The area match value of the face in each frame picture that the video image includes is determined respectively and apart from match value, obtain
Area is fitted value set and apart from fitting value set；
From the distance fitting value set, according to preset rules, first is extracted apart from match value and second distance match value；
Value set is fitted based on the area, determines that described first is default apart from match value and second distance match value satisfaction
During condition, determine that target action occurs in the preset time period.
2. the method as described in claim 1, which is characterized in that determine in any one frame picture that the video image includes
The area match value of face and apart from match value, including：
Face datection is carried out to any one frame picture that the video image includes, determines human face region；
Based on the human face region, coordinate position of the face in the picture is determined；
Based on the coordinate position, the area value and distance value of face are calculated respectively, and the distance value is face and place picture
The distance between lower horizontal line；
Based on the area value and the distance value, determine respectively the corresponding area match value of the area value and it is described away from
From be worth it is corresponding with a distance from match value.
3. the method as described in claim 1, which is characterized in that it is described from the distance fitting value set, according to default rule
Then, first is extracted apart from match value and second distance match value, including：
Each in the distance fitting value set is recorded apart from match value, is formed discrete apart from match value curve；
Determine a pair of of the wave crest and trough nearest with the starting point of the preset time period；
The wave crest and the trough are extracted, and determines the wave crest for described first apart from match value and the wave respectively
Paddy is the second distance match value.
4. such as claim 1-3 any one of them methods, which is characterized in that be fitted value set based on the area, determine institute
State first apart from match value and the second distance match value meet preset condition when, determine to occur in the preset time period
Target action, including：
Described first is determined respectively apart from match value corresponding first frame moment and the second distance match value corresponding second
The frame moment；
It is fitted in value set in the area, extracts the second frame moment corresponding first area match value；
The absolute difference between the second frame moment and the first frame moment is judged, positioned at preset first frame moment threshold value
Between preset second frame moment threshold value；Also,
Judge the second distance match value and the described first absolute difference between match value, positioned at twice described
When between one area match value and the first area match value of three times；
It determines that target action occurs in the preset time period.
5. method as claimed in claim 4, which is characterized in that determine in the preset time period occur target action it
Afterwards, it further includes：
If the second distance match value is higher than described first apart from match value, it is determined that the target action is stands；
If the second distance match value is less than described first apart from match value, it is determined that the target action is sits down.
6. a kind of action recognition device, which is characterized in that including：
First extraction unit, for extracting the video image in preset time period；
First determination unit, for determining the area match value of the face in each frame picture that the video image includes respectively
With apart from match value, area fitting value set is obtained and apart from fitting value set；
Second extraction unit, for from the distance fitting value set, according to preset rules, extracting first apart from match value
With second distance match value；
Second determination unit for being based on the area fitting value set, determines described first apart from match value and described second
When meeting preset condition apart from match value, determine that target action occurs in the preset time period.
7. device as claimed in claim 6, which is characterized in that determine in any one frame picture that the video image includes
The area match value of face and during apart from match value, first determination unit is used for：
Face datection is carried out to any one frame picture that the video image includes, determines human face region；
Based on the human face region, coordinate position of the face in the picture is determined；
Based on the coordinate position, the area value and distance value of face are calculated respectively, and the distance value is face and place picture
The distance between lower horizontal line；
Based on the area value and the distance value, determine respectively the corresponding area match value of the area value and it is described away from
From be worth it is corresponding with a distance from match value.
8. device as claimed in claim 6, which is characterized in that it is described from the distance fitting value set, according to default rule
Then, when extracting first apart from match value and second distance match value, second extraction unit is used for：
Each in the distance fitting value set is recorded apart from match value, is formed discrete apart from match value curve；
Determine a pair of of the wave crest and trough nearest with the starting point of the preset time period；
The wave crest and the trough are extracted, and determines the wave crest for described first apart from match value and the wave respectively
Paddy is the second distance match value.
9. such as claim 6-8 any one of them devices, which is characterized in that be fitted value set based on the area, determine institute
State first apart from match value and the second distance match value meet preset condition when, determine to occur in the preset time period
During target action, second determination unit is used for：
Described first is determined respectively apart from match value corresponding first frame moment and the second distance match value corresponding second
The frame moment；
It is fitted in value set in the area, extracts the second frame moment corresponding first area match value；
The absolute difference between the second frame moment and the first frame moment is judged, positioned at preset first frame moment threshold value
Between preset second frame moment threshold value；Also,
Judge the second distance match value and the described first absolute difference between match value, positioned at twice described
When between one area match value and the first area match value of three times；
It determines that target action occurs in the preset time period.
10. device as claimed in claim 9, which is characterized in that determine in the preset time period occur target action it
Afterwards, second determination unit is additionally operable to：
If the second distance match value is higher than described first apart from match value, it is determined that the target action is stands；
If the second distance match value is less than described first apart from match value, it is determined that the target action is sits down.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611160942.1A CN108229251A (en) | 2016-12-15 | 2016-12-15 | A kind of action identification method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611160942.1A CN108229251A (en) | 2016-12-15 | 2016-12-15 | A kind of action identification method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108229251A true CN108229251A (en) | 2018-06-29 |
Family
ID=62651501
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611160942.1A Pending CN108229251A (en) | 2016-12-15 | 2016-12-15 | A kind of action identification method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108229251A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109117757A (en) * | 2018-07-27 | 2019-01-01 | 四川大学 | A kind of method of drag-line in extraction Aerial Images |
| CN109165578A (en) * | 2018-08-08 | 2019-01-08 | 盎锐（上海）信息科技有限公司 | Expression detection device and data processing method based on filming apparatus |
| CN112614214A (en) * | 2020-12-18 | 2021-04-06 | 北京达佳互联信息技术有限公司 | Motion capture method, motion capture device, electronic device and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103310192A (en) * | 2013-06-06 | 2013-09-18 | 南京邮电大学 | Movement behavior recognition method based on axial acceleration sensor |
| CN103327250A (en) * | 2013-06-24 | 2013-09-25 | 深圳锐取信息技术股份有限公司 | Method for controlling camera lens based on pattern recognition |
| CN103780837A (en) * | 2014-01-02 | 2014-05-07 | 中安消技术有限公司 | Motion detection and positioning photography method and device thereof |
| JP2014147813A (en) * | 2014-04-07 | 2014-08-21 | Copcom Co Ltd | Game apparatus, and game program for realizing the game apparatus |
| US20150154449A1 (en) * | 2013-11-29 | 2015-06-04 | Fujitsu Limited | Method and apparatus for recognizing actions |
-
2016
- 2016-12-15 CN CN201611160942.1A patent/CN108229251A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103310192A (en) * | 2013-06-06 | 2013-09-18 | 南京邮电大学 | Movement behavior recognition method based on axial acceleration sensor |
| CN103327250A (en) * | 2013-06-24 | 2013-09-25 | 深圳锐取信息技术股份有限公司 | Method for controlling camera lens based on pattern recognition |
| US20150154449A1 (en) * | 2013-11-29 | 2015-06-04 | Fujitsu Limited | Method and apparatus for recognizing actions |
| CN103780837A (en) * | 2014-01-02 | 2014-05-07 | 中安消技术有限公司 | Motion detection and positioning photography method and device thereof |
| JP2014147813A (en) * | 2014-04-07 | 2014-08-21 | Copcom Co Ltd | Game apparatus, and game program for realizing the game apparatus |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109117757A (en) * | 2018-07-27 | 2019-01-01 | 四川大学 | A kind of method of drag-line in extraction Aerial Images |
| CN109117757B (en) * | 2018-07-27 | 2022-02-22 | 四川大学 | Method for extracting guy cable in aerial image |
| CN109165578A (en) * | 2018-08-08 | 2019-01-08 | 盎锐（上海）信息科技有限公司 | Expression detection device and data processing method based on filming apparatus |
| CN112614214A (en) * | 2020-12-18 | 2021-04-06 | 北京达佳互联信息技术有限公司 | Motion capture method, motion capture device, electronic device and storage medium |
| CN112614214B (en) * | 2020-12-18 | 2023-10-27 | 北京达佳互联信息技术有限公司 | Motion capture method, motion capture device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11789545B2 (en) | Information processing device and method, program and recording medium for identifying a gesture of a person from captured image data | |
| CN109076198B (en) | Video-based object tracking occlusion detection system, method and equipment | |
| US9311534B2 (en) | Method and apparatus for tracking object | |
| US11398049B2 (en) | Object tracking device, object tracking method, and object tracking program | |
| JP6392756B2 (en) | System and method for obtaining accurate body size measurements from a two-dimensional image sequence | |
| US9224037B2 (en) | Apparatus and method for controlling presentation of information toward human object | |
| Jalal et al. | Depth map-based human activity tracking and recognition using body joints features and self-organized map | |
| US8965050B2 (en) | Behavior analysis device | |
| CN104346816B (en) | Depth determining method and device and electronic equipment | |
| US10659670B2 (en) | Monitoring system and control method thereof | |
| CN103870824B (en) | A kind of face method for catching and device during Face datection tracking | |
| US20110019920A1 (en) | Method, apparatus, and program for detecting object | |
| CN105426827A (en) | Living body verification method, device and system | |
| US10474876B2 (en) | Image processing apparatus and image processing method | |
| CN110298218B (en) | Interactive fitness device and interactive fitness system | |
| CN109308437B (en) | Motion recognition error correction method, electronic device, and storage medium | |
| JPH06266840A (en) | Status detector for moving object | |
| CN108229251A (en) | A kind of action identification method and device | |
| CN109308438A (en) | Method for building up, electronic equipment, the storage medium in action recognition library | |
| JP5940862B2 (en) | Image processing device | |
| Iwasawa et al. | Real-time, 3D estimation of human body postures from trinocular images | |
| CN113449570A (en) | Image processing method and device | |
| Kang et al. | Using depth and skin color for hand gesture classification | |
| CN110111368B (en) | Human body posture recognition-based similar moving target detection and tracking method | |
| Iwasawa et al. | Human body postures from trinocular camera images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180629 |