CN105825170B - Toward the detection method and device of renaturation gesture - Google Patents
Toward the detection method and device of renaturation gesture Download PDFInfo
- Publication number
- CN105825170B CN105825170B CN201610137640.6A CN201610137640A CN105825170B CN 105825170 B CN105825170 B CN 105825170B CN 201610137640 A CN201610137640 A CN 201610137640A CN 105825170 B CN105825170 B CN 105825170B
- Authority
- CN
- China
- Prior art keywords
- image
- value
- sequence
- mask
- frame video
- 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.)
- Active
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
- G06V40/28—Recognition of hand or arm movements, e.g. recognition of deaf sign language
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides the detection method and device of a kind of past renaturation gesture.This method comprises: obtaining the corresponding two-value mask image of every frame video image according to the corresponding two-value movement mask image of frame video image every in video sequence and the corresponding two-value colour of skin mask image of every frame video image；According to the corresponding two-value mask image of every frame video image, exposure mask sequence of the video sequence in preset duration is obtained；The corresponding response image of the video sequence is determined according to the exposure mask sequence；It is determined in the video sequence according to the response image with the presence or absence of the past renaturation gesture.Method provided by the invention, detection algorithm is simple, and detection efficiency is high.
Description
Technical field
The present invention relates to Video Supervision Technique more particularly to a kind of detection methods and device of past renaturation gesture.
Background technique
With the development of computer technology, network communication technology, people's lives level has huge raising, inhabitation item
Part is greatly improved, and it is Modern Residence that the development of smart home system, which is the inexorable trend of China's residence technology innovation,
Inevitable choice.Gesture Recognition is using oneself however it is more harmonious, natural to be supplied to user as input mode for intuitive manpower
Interactive mode, be the hot spot of human-computer interaction, gesture be applied in smart home and undoubtedly brings huge convenience to life.It is logical
It crosses and the target gesture of user is tracked and identified, can be triggered or be controlled accordingly according to the gesture recognized.
For example, waving movement by detection to trigger corresponding event.
The mode that prior art detection waves to act generally comprises three detection, tracking, identification key components, it is necessary first to
The position of manpower is positioned, then opponent tracks, and analyzes the motion profile of hand to be identified.
The scheme of movement but prior art detection is waved, often has excessive restrictive condition, such as the speed waved,
The complexity etc. of background, algorithm comparison is complicated, and detection efficiency is low.
Summary of the invention
The present invention provides the detection method and device of a kind of past renaturation gesture, in the prior art waves to act to solve
Detection scheme algorithm comparison it is complicated, the low technical problem of detection efficiency.
In a first aspect, the present invention provides a kind of detection method of past renaturation gesture, comprising:
It is corresponding according to the corresponding two-value movement mask image of frame video image every in video sequence and every frame video image
Two-value colour of skin mask image obtains the corresponding two-value mask image of every frame video image；
According to the corresponding two-value mask image of every frame video image, the video sequence is obtained in preset duration
Exposure mask sequence；The exposure mask sequence includes the part two-value exposure mask figure in the corresponding all two-value mask images of the video sequence
Picture；
The corresponding response image of the video sequence is determined according to the exposure mask sequence, the pixel in the response image
Value be used to characterize the probability of past the renaturation gesture appearance；
It is determined in the video sequence according to the response image with the presence or absence of the past renaturation gesture.
Further, described to be regarded according to the corresponding two-value movement mask image of frame video image every in video sequence and every frame
The corresponding two-value colour of skin mask image of frequency image obtains the corresponding two-value mask image of every frame video image, specifically includes:
By each first pixel of the corresponding two-value movement mask image of every frame video image and every frame view
Each second pixel of the corresponding two-value colour of skin mask image of frequency image carries out and operation, obtains every frame video image pair
The two-value mask image answered；Wherein, it carries out identical as the coordinate of the first pixel of operation and the second pixel.
Further, described according to the corresponding two-value mask image of every frame video image, obtain the video sequence
Exposure mask sequence within a preset time interval, specifically includes:
Successively judge whether be greater than the preset duration at the time of every frame video image corresponds to；
If being less than or equal to the preset duration at the time of current frame video image corresponds to, by the present frame video
Element of the corresponding two-value mask image of image as the front end of the exposure mask sequence；
If being greater than the preset duration at the time of current frame video image corresponds to, by the current frame video image
Element of the corresponding two-value mask image as the front end of the exposure mask sequence, and delete the most end in the exposure mask sequence
Element.
Further, described that the corresponding response image of the video sequence is determined according to the exposure mask sequence, it specifically includes:
According to formula:Obtain the response image R；Wherein, the MiFor two in the exposure mask sequence
It is worth mask image, the N is the length of the exposure mask sequence.
Further, described to be determined in the video sequence according to the response image with the presence or absence of the past renaturation hand
Gesture specifically includes:
According to formula:To in the response image each third pixel R (x,
Y) thresholding processing is carried out, the corresponding bianry image X of the response image is obtained；Wherein, the x, the y are the third picture
The coordinate of vegetarian refreshments, the TlFor preset threshold range；
Connected domain analysis is carried out to the bianry image X, obtains rectangular area set O={ o1,...om-1,om}；The m
For the number of rectangle in the set of the rectangular area；
If there are width to meet the first preset condition in the set of the rectangular area, height meets the of the second preset condition
One rectangle, it is determined that there are the past renaturation gestures in the video sequence.
Second aspect, the present invention provide a kind of detection device of past renaturation gesture, comprising:
First obtains module, for according to the corresponding two-value movement mask image of frame video image every in video sequence and often
The corresponding two-value colour of skin mask image of frame video image obtains the corresponding two-value mask image of every frame video image；
Second obtains module, for obtaining the video according to the corresponding two-value mask image of every frame video image
Exposure mask sequence of the sequence in preset duration；The exposure mask sequence includes the corresponding all two-value mask images of the video sequence
In part two-value mask image；
First determining module, it is described for determining the corresponding response image of the video sequence according to the exposure mask sequence
The value of pixel in response image is used to characterize the probability that the past renaturation gesture occurs；
Second determining module, for being determined in the video sequence according to the response image with the presence or absence of the past renaturation
Gesture.
Further, described first module is obtained, specifically for covering the corresponding two-value movement of every frame video image
Each first pixel of film image and each second pixel of the corresponding two-value colour of skin mask image of every frame video image
Point carries out and operation, obtains the corresponding two-value mask image of every frame video image；Wherein, the first pixel with operation is carried out
Point is identical with the coordinate of the second pixel.
Further, described second module is obtained, comprising:
Whether judging unit is greater than the preset duration at the time of for successively judging that every frame video image corresponds to；
Processing unit, for being less than or equal to institute at the time of the judging unit judges that current frame video image corresponds to
When stating preset duration, using the corresponding two-value mask image of the current frame video image as the front end of the exposure mask sequence
Element；And it is greater than the preset duration at the time of judging unit judges that current frame video image corresponds to, work as by described in
The element of front end of the corresponding two-value mask image of preceding frame video image as the exposure mask sequence, and delete the exposure mask sequence
The element of most end in column.
Further, first determining module is specifically used for according to formula:Obtain the response diagram
As R；Wherein, the MiFor the two-value mask image in the exposure mask sequence, the N is the length of the exposure mask sequence.
Further, second determining module, specifically includes:
Thresholding unit, for according to formula:To each in the response image
Third pixel R (x, y) carries out thresholding processing, obtains the corresponding bianry image X of the response image；Wherein, the x, institute
State the coordinate that y is the third pixel, the TlFor preset threshold range；
Analytical unit obtains rectangular area set O={ o for carrying out connected domain analysis to the bianry image X1,
...om-1,om}；The m is the number of rectangle in the set of the rectangular area；
Determination unit, for there are width to meet the first preset condition, height completely in judging the rectangular area set
When the first rectangle of the second preset condition of foot, determine that there are the past renaturation gestures in the video sequence.
The detection method and device of past renaturation gesture provided by the invention, it is only necessary to which common camera can be detected view
With the presence or absence of toward renaturation gesture in frequency sequence, testing cost is low；Also, present invention only requires obtained by way of the prior art
The two-value movement mask image and two-value colour of skin mask image of every frame image, and then by simply obtaining every frame image with operation
Two-value mask image and then obtain response image to obtain calculating the exposure mask sequence of response image by simply judgement,
And determine that the algorithm being related to is simple, substantially increases with the presence or absence of toward renaturation gesture in video sequence according to the response image
Detection efficiency.
Detailed description of the invention
Technical solution in the prior art in order to illustrate the embodiments of the present invention more clearly, below will be to embodiment or existing
Attached drawing needed in technical description is briefly described, it should be apparent that, the accompanying drawings in the following description is the present invention
Some embodiments for those of ordinary skill in the art without any creative labor, can be with root
Other attached drawings are obtained according to these attached drawings.
Fig. 1 is the flow diagram of the detection method embodiment one of past renaturation gesture provided by the invention；
Fig. 2 is the flow diagram of the detection method embodiment two of past renaturation gesture provided by the invention；
Fig. 3 is current frame video image schematic diagram provided by the invention；
Fig. 4 is that the two-value of current frame video image provided by the invention moves mask image schematic diagram；
Fig. 5 is the two-value colour of skin mask image schematic diagram of current frame video image provided by the invention；
Fig. 6 is the two-value mask image schematic diagram of current frame video image provided by the invention；
Fig. 7 is response image schematic diagram provided by the invention；
Fig. 8 is the first rectangle schematic diagram provided by the invention；
Fig. 9 is the structural schematic diagram of the detection device embodiment one of past renaturation gesture provided by the invention；
Figure 10 is the structural schematic diagram of the detection device embodiment two of past renaturation gesture provided by the invention.
Specific embodiment
To make the object, technical solutions and advantages of the present invention clearer, below in conjunction with the attached drawing in the present invention, to this
Technical solution in invention is clearly and completely described, it is clear that and described embodiments are some of the embodiments of the present invention,
Instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not making creative labor
Every other embodiment obtained under the premise of dynamic, shall fall within the protection scope of the present invention.
The detection method and device of past renaturation gesture according to the present invention, can be adapted for any Target Tracking System
Or in video monitoring system, such as it can be adapted for single goal gesture tracking system, when single goal gesture tracking system occurs
Tracking mistake perhaps tracking failure when can using method of the invention trigger in time monotrack system stop tracking or
Restart to track.That is, method and apparatus according to the present invention, it can be by judging whether to detect toward renaturation hand
Gesture triggers a certain event or a certain function in a certain system.This waves to move toward the left and right that renaturation gesture can be manpower
Make, can be the up and down reciprocatingly property movement of manpower, can also be that manpower waves to act up and down, the present invention does not do this
It limits.
The executing subject of following methods embodiment can be the detection device toward renaturation gesture, which can be collection
At the device in video monitoring system perhaps Target Tracking System can also be individual video monitoring system or target with
Track system.The detection device can be realized by way of software, hardware or soft or hard combination.Following embodiments are with executing subject
For video monitoring system.
The detection method and device of past renaturation gesture of the present invention, it is intended to solve detection in the prior art and wave to move
The Scheme algorithm of work is complicated, the low technical problem of detection efficiency.
Technical solution of the present invention is described in detail with specifically embodiment below.These specific implementations below
Example can be combined with each other, and the same or similar concept or process may be repeated no more in certain embodiments.
Fig. 1 is the flow diagram of the detection method embodiment one of past renaturation gesture provided by the invention.The present embodiment relates to
And be by determine video sequence in response image come detect in video sequence with the presence or absence of toward renaturation gesture specific mistake
Journey.As shown in Figure 1, this method may include steps of:
S101: according to the corresponding two-value movement mask image of frame video image every in video sequence and every frame video image pair
The two-value colour of skin mask image answered obtains the corresponding two-value mask image of every frame video image.
Specifically, video monitoring system arrives video sequence by the way that camera is available, which can be common
WEB camera, the video sequence may include multi-frame video image.After getting video sequence, in video monitoring system
Processor every frame video image can be performed corresponding processing, to obtain every frame video image corresponding two-value movement exposure mask
Image and two-value colour of skin mask image.Wherein, the value of the partial pixel point in two-value movement mask image then shows this if 1
Partial pixel point is moving region, if the value of some pixel in two-value movement mask image is 0, shows that the pixel is
Stagnant zone；The value of partial pixel point in two-value colour of skin mask image then shows that this partial pixel point is colour of skin area if 1
Domain shows that the pixel is non-area of skin color if the value of some pixel in two-value colour of skin mask image is 0.Optionally,
Processor carries out operation detection to every frame video image in video sequence and obtains the two-value movement exposure mask figure of every frame video image
The method of picture may include a variety of, for example, can pass through the side such as background subtraction, frame differential method, mixed Gaussian background modeling
Method carries out motion detection to every frame video image in video sequence, obtains two-value movement mask image；Optionally, processor pair
It also includes a variety of that every frame video image in video sequence, which carries out the method that Face Detection obtains two-value colour of skin mask image, such as
It can be the skin color detection method based on YCbCr color space.It is following can be empty in inter-frame difference method and based on YCbCr color
Between skin color detection method for come simply introduce video monitoring system obtain every frame video image two-value movement mask image
With two-value colour of skin mask image.
(1) the two-value movement mask image M of every frame video image is obtained using inter-frame difference methodd(wherein, worked as with obtaining
The corresponding video frame images I of preceding moment ttFor, other video frame images as reference), be broadly divided into two step a and
B:
A: I is calculatedtAnd its previous frame image It-1Corresponding gray level image Gt、Gt-1, specific calculation may refer to
The prior art, details are not described herein.
B: G is calculatedtPixel and Gt-1(i.e. two pixels of calculating difference are respective for the difference of upper corresponding pixel points
Coordinate on image is identical), if the difference of corresponding pixel points is greater than T, otherwise it is 0 that the value of the pixel, which is 1,.Calculation formula
It is as follows:
Wherein x, y MdThe coordinate of upper pixel, T are threshold value, and optionally, T's may range from 5 < T < 35.
A and b through the above steps, so that it may obtain two-value movement mask image MdOn each pixel value, in turn
Obtain two-value movement mask image.
(2) skin color detection method based on YCbCr color space is utilized, the two-value colour of skin exposure mask of every frame video image is obtained
Image Ms(wherein, equally to obtain the corresponding video frame images I of current time ttFor, other video frame images are as ginseng
Examine), it is broadly divided into two steps c and d:
C: by image ItIt is transformed into YCbCr color space, 3 channels are respectively Y, Cb、Cr.Image ItIt is transformed into YCbCr face
The concrete mode of the colour space may refer to the prior art, and details are not described herein.
D: to Cb、Cr2 Color Channels carry out threshold process, obtain colour of skin mask image Ms.The formula of threshold process is such as
Under:
Wherein x, y MsThe coordinate of upper pixel.
C and d through the above steps, so that it may obtain two-value colour of skin mask image MsOn each pixel value, in turn
Obtain two-value colour of skin mask image Ms。
The corresponding video frame images I of current time t is obtained in video monitoring systemtCorresponding two-value moves mask image Md
With two-value colour of skin mask image MsLater, video monitoring system can obtain I according to bothtCorresponding two-value mask image,
Optionally, two-value can be moved mask image M by video monitoring systemdWith two-value colour of skin mask image MsProgress is obtained with operation
ItCorresponding two-value mask image can also obtain I by other logical operationstCorresponding two-value mask image, the present invention couple
This and with no restrictions.
So all doing this processing to each frame video image according to above-mentioned steps a and b and step c and d, obtain every
The two-value movement mask image and two-value colour of skin mask image of frame video image, and then video monitoring system can be obtained by every frame
The corresponding two-value mask image of video image.
S102: according to the corresponding two-value mask image of every frame video image, the video sequence is obtained when default
Exposure mask sequence in length；The exposure mask sequence includes the part two-value in the corresponding all two-value mask images of the video sequence
Mask image.
Specifically, not having started work in video monitoring system or just starting to start, the value of exposure mask sequence is sky, is being regarded
Frequency monitoring system gets the corresponding two-value mask image of every frame video image, so that it may detect video sequence according to default rule
T obtains the exposure mask sequence Γ in preset duration Δ t at the time of every frame video image in column corresponds to, it is assumed that and Γ=
{M1···Mn-1,MnIt is wherein M1Indicate present frame (the two-value mask image that the former frame of moment t) detects, M2Refer to
Be present frame the two-value mask image that detects of preceding 2 frame, MnThe two-value mask image that the preceding n frame of present frame detects is (i.e.
The two-value mask image that t- time Δt detects), the time interval between preceding n frame and present frame is that (i.e. preset duration is Δ t
Δt)。
S103: the corresponding response image of the video sequence is determined according to the exposure mask sequence, in the response image
The value of pixel is used to characterize the probability that the past renaturation gesture occurs.
It optionally, can be according to multiple in exposure mask sequence after video monitoring system has determined above-mentioned exposure mask sequence
Two-value mask image determines the corresponding response image of video sequence, and the value characterization of each pixel in the response image is reciprocal
Property gesture occur probability.Optionally, it can be and be averaged multiple two-value mask images summation in exposure mask sequence again
To the corresponding response image of video sequence, such as it can be and weighted average is taken to obtain response image, can also be and take its other party
The average value of formula obtains response image.
S104: it is determined in the video sequence according to the response image with the presence or absence of the past renaturation gesture.
Optionally, after video monitoring system obtains video sequence corresponding response image, the response can be passed through
Image judges optionally to can be by judging the pixel in response image in video sequence with the presence or absence of toward renaturation gesture
Value be greater than the number of a certain threshold value and whether reach certain preset condition to determine whether there is toward renaturation gesture, work as response image
In pixel value number when meeting preset condition that is greater than a certain threshold value, just illustrate exist in video sequence toward renaturation hand
Gesture；Optionally, whether can also be greater than a certain threshold by judging the average value of the value of pixel in some region in response image
Value determines in video sequence with the presence or absence of response image, when the value of the pixel in some region in response image is averaged
When value is greater than a certain threshold value, just illustrate that there may be past renaturation gestures in the region.The present invention is determined to by response image
With the presence or absence of the mode and without limitation toward renaturation gesture in video sequence.
Seen from the above description, the detection method of past renaturation gesture of the present invention, it is only necessary to which common camera is
It can be detected with the presence or absence of toward renaturation gesture in video sequence, testing cost is low；Also, present invention only requires pass through the prior art
Mode obtain the two-value movement mask image and two-value colour of skin mask image of every frame image, and then by simply being obtained with operation
To the two-value mask image of every frame image, thus obtain calculating the exposure mask sequence of response image by simply judgement, and then
Determine that the algorithm being related to is simple with the presence or absence of toward renaturation gesture in video sequence to response image, and according to the response image,
Substantially increase detection efficiency.
Fig. 2 is the flow diagram of the detection method embodiment two of past renaturation gesture provided by the invention.In above-mentioned implementation
On the basis of example, the present embodiment is related to being two-value mask image that video monitoring system obtains every frame video image, and then obtains
Response image, and image determines that video sequence whether there is the detailed process toward renaturation gesture according to response.As shown in Fig. 2,
This method may include steps of:
S201: video sequence is obtained, and every frame video image in the video sequence is obtained according to the video sequence
Two-value moves mask image and the corresponding two-value colour of skin mask image of every frame video image.
Specifically, the detailed process of the S201 may refer to the description of above-mentioned S101, details are not described herein.
S202: by each first pixel of the corresponding two-value movement mask image of every frame video image and described every
Each second pixel of the corresponding two-value colour of skin mask image of frame video image carries out and operation, obtains every frame video figure
As corresponding two-value mask image；Wherein, it carries out identical as the coordinate of the first pixel of operation and the second pixel.
Specifically, each first pixel and corresponding each second pixel are carried out can be with the detailed process of operation
Are as follows: for each pixel in two-value mask image, if the value of corresponding first pixel and the second pixel is 1,
The pixel of the two-value mask image of the coordinate position is just 1, is otherwise 0.That is, for the same coordinate position, this
The first pixel and the second pixel on a position are 1, then the value of the pixel of this coordinate is just in two-value mask image
Be in 1, otherwise the first pixel and the second pixel one be 0 or both for 0, then this seat in two-value mask image
The value of target pixel is just 0.
S203: successively judge whether be greater than the preset duration at the time of every frame video image corresponds to.
S204:, will be described current if being less than or equal to the preset duration at the time of current frame video image corresponds to
Element of the corresponding two-value mask image of frame video image as the front end of the exposure mask sequence.
S205: if being greater than the preset duration at the time of current frame video image corresponds to, the present frame is regarded
Element of the corresponding two-value mask image of frequency image as the front end of the exposure mask sequence, and delete in the exposure mask sequence
The element of most end.
Judge at the time of specifically, successively being corresponded to every frame video image in video sequence, that is, judges every frame view
Whether it is greater than preset duration at the time of frequency image corresponds to, if the corresponding of current frame video image is constantly for t, when the value of t
When carving less than or equal to above-mentioned preset duration Δ t, then using the two-value mask image of present frame as the front end of exposure mask sequence
Element；If the value of t is carved greater than above-mentioned preset duration Δ t, using the two-value mask image of present frame as exposure mask sequence
The element of front end, and the element for being located at most end in exposure mask sequence is deleted, the element in exposure mask sequence refers to meet item
The two-value mask image of part.According to this mode, the exposure mask sequence in preset time Δ t has just been obtained, if the exposure mask sequence
Length is N, i.e., the number of the two-value mask image in the exposure mask sequence is N.
Referring to shown in Fig. 3-Fig. 8, Fig. 3 is current frame video image (original image, i.e. color image), and Fig. 4 is present frame
The two-value of video image moves mask image, and Fig. 5 is the two-value colour of skin mask image of current frame video image, and Fig. 6 is present frame
The two-value mask image of video image.
S206: according to formula:Obtain the response image R.
Wherein, the MiFor the two-value mask image in the exposure mask sequence.Specifically, may refer to sound shown in Fig. 7
Answer image.
S207: according to formula:To each third pixel in the response image
R (x, y) carries out thresholding processing, obtains the corresponding bianry image X of the response image.
Wherein, the x, the y are the coordinate of the third pixel, the TlFor preset threshold range.Optionally, should
TlValue range can be 30 < Tl<100。
S208: connected domain analysis is carried out to the bianry image X, obtains rectangular area set O={ o1,...om-1,om}；
The m is the number of rectangle in the set of the rectangular area.
S209: if there are width to meet the first preset condition in the set of the rectangular area, height meets the second default item
First rectangle of part, it is determined that there are the past renaturation gestures in the video sequence.
Optionally, which can be w > 6, which can be h > 6；Wherein, w is rectangle
Width, h are the length of rectangle.It may refer to the first rectangle shown in Fig. 8.
In addition, it is necessary to explanation, two-value operation mask image, the two-value colour of skin of the above-mentioned every frame video image being related to are covered
Film image, two-value mask image the equal and every frame video image of width of same size, every frame video image two-value operation cover
Film image, two-value colour of skin mask image, the height of the equal and every frame video image of the height of two-value mask image are identical.
The detection method of past renaturation gesture provided by the invention, detection algorithm is fairly simple, greatly improves toward renaturation hand
The detection efficiency of gesture can faster trigger some time in video monitoring system or Target Tracking System or function
Can, largely improve the experience of user.
Those of ordinary skill in the art will appreciate that: realize that all or part of the steps of above-mentioned each method embodiment can lead to
The relevant hardware of program instruction is crossed to complete.Program above-mentioned can be stored in a computer readable storage medium.The journey
When being executed, execution includes the steps that above-mentioned each method embodiment to sequence；And storage medium above-mentioned include: ROM, RAM, magnetic disk or
The various media that can store program code such as person's CD.
Fig. 9 is the structural schematic diagram of the detection device embodiment one of past renaturation gesture provided by the invention.As shown in figure 9,
The apparatus may include: first, which obtains module 10, second, obtains module 11, the first determining module 12 and the second determining module 13.
Wherein, first module 10 is obtained, for moving exposure mask according to the corresponding two-value of frame video image every in video sequence
Image and the corresponding two-value colour of skin mask image of every frame video image, obtain the corresponding two-value mask image of every frame video image；
Second obtains module 11, for obtaining the view according to the corresponding two-value mask image of every frame video image
Exposure mask sequence of the frequency sequence in preset duration；The exposure mask sequence includes the corresponding all two-value exposure mask figures of the video sequence
Part two-value mask image as in；
First determining module 12, for determining the corresponding response image of the video sequence, institute according to the exposure mask sequence
The value for stating the pixel in response image is used to characterize the probability that the past renaturation gesture occurs；
Second determining module 13, for being determined in the video sequence according to the response image with the presence or absence of described reciprocal
Property gesture.
The detection device of past renaturation gesture provided by the invention, can execute above method embodiment, realization principle and
Technical effect is similar, and details are not described herein.
Figure 10 is the structural schematic diagram of the detection device embodiment two of past renaturation gesture provided by the invention.In above-mentioned Fig. 9
On the basis of illustrated embodiment, further, above-mentioned second acquisition module 11 specifically includes judging unit 111 and processing unit
112；Above-mentioned second determining module 13 includes thresholding unit 131, analytical unit 132 and determination unit 133.
Wherein, in the present embodiment, above-mentioned first module 10 is obtained, be specifically used for every frame video image is corresponding
Two-value move mask image each first pixel and the corresponding two-value colour of skin mask image of every frame video image it is each
A second pixel carries out and operation, obtains the corresponding two-value mask image of every frame video image；Wherein, progress and operation
The first pixel and the second pixel coordinate it is identical.
Whether above-mentioned judging unit 111 is greater than described pre- at the time of for successively judging that every frame video image corresponds to
If duration；
Above-mentioned processing unit 112, for being less than at the time of the judging unit 111 judges that current frame video image corresponds to
Or when being equal to the preset duration, using the corresponding two-value mask image of the current frame video image as the exposure mask sequence
Front end element；And it is greater than at the time of the judging unit 111 judges that current frame video image corresponds to described default
Duration, using the corresponding two-value mask image of the current frame video image as the element of the front end of the exposure mask sequence, and
Delete the element of the most end in the exposure mask sequence.
Above-mentioned first determination unit 133 is specifically used for according to formula:Obtain the response image R；Its
In, the MiFor the two-value mask image in the exposure mask sequence, the N is the length of the exposure mask sequence.
Above-mentioned thresholding unit 131, for according to formula:To the response image
In each third pixel R (x, y) carry out thresholding processing, obtain the corresponding bianry image X of the response image；Wherein,
The x, the y are the coordinate of the third pixel, the TlFor preset threshold range；
Above-mentioned analytical unit 132 obtains rectangular area set O=for carrying out connected domain analysis to the bianry image X
{o1,...om-1,om}；The m is the number of rectangle in the set of the rectangular area；
Above-mentioned determination unit 133, in judging rectangular area set there are width meet the first preset condition,
When height meets the first rectangle of the second preset condition, determine that there are the past renaturation gestures in the video sequence.
The detection device of past renaturation gesture provided by the invention, can execute above method embodiment, realization principle and
Technical effect is similar, and details are not described herein.
Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention., rather than its limitations；To the greatest extent
Pipe present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: its according to
So be possible to modify the technical solutions described in the foregoing embodiments, or to some or all of the technical features into
Row equivalent replacement；And these are modified or replaceed, various embodiments of the present invention technology that it does not separate the essence of the corresponding technical solution
The range of scheme.
Claims (8)
1. a kind of detection method of past renaturation gesture characterized by comprising
According to the corresponding two-value movement mask image of frame video image every in video sequence and the corresponding two-value of every frame video image
Colour of skin mask image obtains the corresponding two-value mask image of every frame video image；
According to the corresponding two-value mask image of every frame video image, exposure mask of the video sequence in preset duration is obtained
Sequence；The exposure mask sequence includes the part two-value mask image in the corresponding all two-value mask images of the video sequence；
The corresponding response image of the video sequence, the value of the pixel in the response image are determined according to the exposure mask sequence
The probability occurred for characterizing the past renaturation gesture；
It is determined in the video sequence according to the response image with the presence or absence of the past renaturation gesture；
It is described corresponding according to the corresponding two-value movement mask image of frame video image every in video sequence and every frame video image
Two-value colour of skin mask image obtains the corresponding two-value mask image of every frame video image, specifically includes:
By each first pixel and every frame video figure of the corresponding two-value movement mask image of every frame video image
As the progress of each second pixel and operation of corresponding two-value colour of skin mask image, it is corresponding to obtain every frame video image
Two-value mask image；Wherein, it carries out identical as the coordinate of the first pixel of operation and the second pixel.
2. the method according to claim 1, wherein described cover according to the corresponding two-value of every frame video image
Film image obtains the exposure mask sequence of the video sequence within a preset time interval, specifically includes:
Successively judge whether be greater than the preset duration at the time of every frame video image corresponds to；
If being less than or equal to the preset duration at the time of current frame video image corresponds to, by the current frame video image
Element of the corresponding two-value mask image as the front end of the exposure mask sequence；
It is if being greater than the preset duration at the time of current frame video image corresponds to, the current frame video image is corresponding
Two-value mask image as the exposure mask sequence front end element, and delete the member of the most end in the exposure mask sequence
Element.
3. method according to claim 1 or 2, which is characterized in that described to determine the video according to the exposure mask sequence
The corresponding response image of sequence, specifically includes:
According to formula:Obtain the response image R；Wherein, the MiIt is covered for the two-value in the exposure mask sequence
Film image, the N are the length of the exposure mask sequence.
4. according to the method described in claim 3, it is characterized in that, described determine the video sequence according to the response image
In whether there is the past renaturation gesture, specifically include:
According to formula:Each third pixel R (x, y) in the response image is carried out
Thresholding processing, obtains the corresponding bianry image X of the response image；Wherein, the x, the y are the third pixel
Coordinate, the TlFor preset threshold range；
Connected domain analysis is carried out to the bianry image X, obtains rectangular area set O={ o1,…om-1,om}；The m is described
The number of rectangle in the set of rectangular area；
If there are width to meet the first preset condition in the rectangular area set, height meets the first square of the second preset condition
Shape, it is determined that there are the past renaturation gestures in the video sequence.
5. a kind of detection device of past renaturation gesture characterized by comprising
First obtains module, for being regarded according to the corresponding two-value movement mask image of frame video image every in video sequence and every frame
The corresponding two-value colour of skin mask image of frequency image obtains the corresponding two-value mask image of every frame video image；
Second obtains module, for obtaining the video sequence according to the corresponding two-value mask image of every frame video image
Exposure mask sequence in preset duration；The exposure mask sequence includes in the corresponding all two-value mask images of the video sequence
Part two-value mask image；
First determining module, for determining the corresponding response image of the video sequence, the response according to the exposure mask sequence
The value of pixel in image is used to characterize the probability that the past renaturation gesture occurs；
Second determining module, for being determined in the video sequence according to the response image with the presence or absence of the past renaturation hand
Gesture；
Described first obtains module, specifically for by each the of the corresponding two-value movement mask image of every frame video image
One pixel and the progress of each second pixel of the corresponding two-value colour of skin mask image of every frame video image and operation, obtain
To the corresponding two-value mask image of every frame video image；Wherein, the first pixel and the second pixel with operation are carried out
Coordinate it is identical.
6. device according to claim 5, which is characterized in that described second obtains module, comprising:
Whether judging unit is greater than the preset duration at the time of for successively judging that every frame video image corresponds to；
Processing unit, it is described pre- for being less than or equal at the time of the judging unit judges that current frame video image corresponds to
If when duration, using the corresponding two-value mask image of the current frame video image as the member of the front end of the exposure mask sequence
Element；And it is greater than the preset duration at the time of judging unit judges that current frame video image corresponds to, it will be described current
Element of the corresponding two-value mask image of frame video image as the front end of the exposure mask sequence, and delete the exposure mask sequence
In most end element.
7. device according to claim 5 or 6, which is characterized in that first determining module is specifically used for according to public affairs
Formula:Obtain the response image R；Wherein, the MiFor the two-value mask image in the exposure mask sequence, institute
State the length that N is the exposure mask sequence.
8. device according to claim 7, which is characterized in that second determining module specifically includes:
Thresholding unit, for according to formula:To each third in the response image
Pixel R (x, y) carries out thresholding processing, obtains the corresponding bianry image X of the response image；Wherein, the x, the y are
The coordinate of the third pixel, the TlFor preset threshold range；
Analytical unit obtains rectangular area set O={ o for carrying out connected domain analysis to the bianry image X1,…om-1,
om}；The m is the number of rectangle in the set of the rectangular area；
Determination unit, for there are width to meet the first preset condition in judging rectangular area set, height meets the
When the first rectangle of two preset conditions, determine that there are the past renaturation gestures in the video sequence.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610137640.6A CN105825170B (en) | 2016-03-10 | 2016-03-10 | Toward the detection method and device of renaturation gesture |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610137640.6A CN105825170B (en) | 2016-03-10 | 2016-03-10 | Toward the detection method and device of renaturation gesture |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105825170A CN105825170A (en) | 2016-08-03 |
| CN105825170B true CN105825170B (en) | 2019-07-02 |
Family
ID=56988100
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610137640.6A Active CN105825170B (en) | 2016-03-10 | 2016-03-10 | Toward the detection method and device of renaturation gesture |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105825170B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110956219B (en) * | 2019-12-09 | 2023-11-14 | 爱芯元智半导体(宁波)有限公司 | Video data processing method, device and electronic system |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102324019A (en) * | 2011-08-12 | 2012-01-18 | 浙江大学 | Method and system for automatically extracting gesture candidate region in video sequence |
| CN103366188A (en) * | 2013-07-08 | 2013-10-23 | 中科创达软件股份有限公司 | Gesture tracking method adopting fist detection as auxiliary information |
| CN103530613A (en) * | 2013-10-15 | 2014-01-22 | 无锡易视腾科技有限公司 | Target person hand gesture interaction method based on monocular video sequence |
| CN104050454A (en) * | 2014-06-24 | 2014-09-17 | 深圳先进技术研究院 | Movement gesture track obtaining method and system |
| CN104423556A (en) * | 2013-09-05 | 2015-03-18 | 华为技术有限公司 | Gesture processing method, server side and terminal |
| US9218669B1 (en) * | 2011-01-20 | 2015-12-22 | Verint Systems Ltd. | Image ghost removal |
| CN105373785A (en) * | 2015-11-30 | 2016-03-02 | 北京地平线机器人技术研发有限公司 | Method and device of hand gesture recognition and detection on the basis of deep neural network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2409030A (en) * | 2003-12-11 | 2005-06-15 | Sony Uk Ltd | Face detection |
-
2016
- 2016-03-10 CN CN201610137640.6A patent/CN105825170B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9218669B1 (en) * | 2011-01-20 | 2015-12-22 | Verint Systems Ltd. | Image ghost removal |
| CN102324019A (en) * | 2011-08-12 | 2012-01-18 | 浙江大学 | Method and system for automatically extracting gesture candidate region in video sequence |
| CN103366188A (en) * | 2013-07-08 | 2013-10-23 | 中科创达软件股份有限公司 | Gesture tracking method adopting fist detection as auxiliary information |
| CN104423556A (en) * | 2013-09-05 | 2015-03-18 | 华为技术有限公司 | Gesture processing method, server side and terminal |
| CN103530613A (en) * | 2013-10-15 | 2014-01-22 | 无锡易视腾科技有限公司 | Target person hand gesture interaction method based on monocular video sequence |
| CN104050454A (en) * | 2014-06-24 | 2014-09-17 | 深圳先进技术研究院 | Movement gesture track obtaining method and system |
| CN105373785A (en) * | 2015-11-30 | 2016-03-02 | 北京地平线机器人技术研发有限公司 | Method and device of hand gesture recognition and detection on the basis of deep neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105825170A (en) | 2016-08-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11450146B2 (en) | Gesture recognition method, apparatus, and device | |
| CN108062526B (en) | Human body posture estimation method and mobile terminal | |
| CN107798653B (en) | Image processing method and device | |
| CN106462242B (en) | Use the user interface control of eye tracking | |
| CN110175514B (en) | Face brushing payment prompting method, device and equipment | |
| KR102465532B1 (en) | Method for recognizing an object and apparatus thereof | |
| CN103353935B (en) | A kind of 3D dynamic gesture identification method for intelligent domestic system | |
| Hsieh et al. | A real time hand gesture recognition system using motion history image | |
| WO2018076622A1 (en) | Image processing method and device, and terminal | |
| RU2016107189A (en) | DEVICES, SYSTEMS AND METHODS OF VIRTUALIZATION OF THE MIRROR | |
| Chaudhary et al. | Deep network for human action recognition using Weber motion | |
| CN113128368B (en) | Method, device and system for detecting character interaction relationship | |
| EP2630616A1 (en) | Method and apparatus for providing hand detection | |
| WO2018103416A1 (en) | Method and device for detecting facial image | |
| Urabe et al. | Cooking activities recognition in egocentric videos using combining 2DCNN and 3DCNN | |
| CN112274909A (en) | Application operation control method and device, electronic equipment and storage medium | |
| CN105825170B (en) | Toward the detection method and device of renaturation gesture | |
| WO2021147473A1 (en) | Model training method, content generation method, and related devices | |
| CN109034106A (en) | human face data cleaning method and device | |
| JP6044633B2 (en) | Information processing apparatus, information processing method, and program | |
| KR101945487B1 (en) | Apparatus and method for recognizing object included in video data | |
| US20140285426A1 (en) | Signal processing device and signal processing method | |
| Avgerinakis et al. | Activities of daily living recognition using optimal trajectories from motion boundaries | |
| CN110991307B (en) | Face recognition method, device, equipment and storage medium | |
| CN108509853B (en) | Gesture recognition method based on camera visual information |
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 | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: Detection method and device of reciprocating gesture Effective date of registration: 20230302 Granted publication date: 20190702 Pledgee: Tongxiang Yunbei Investment Construction Co.,Ltd. Pledgor: ZHEJIANG SHENGHUI LIGHTING Co.,Ltd. Registration number: Y2023980033495 |