CN103679130B - Hand method for tracing, hand tracing equipment and gesture recognition system - Google Patents
Hand method for tracing, hand tracing equipment and gesture recognition system Download PDFInfo
- Publication number
- CN103679130B CN103679130B CN201210375618.7A CN201210375618A CN103679130B CN 103679130 B CN103679130 B CN 103679130B CN 201210375618 A CN201210375618 A CN 201210375618A CN 103679130 B CN103679130 B CN 103679130B
- Authority
- CN
- China
- Prior art keywords
- hand
- frame
- template
- target
- candidate
- 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
Abstract
The present invention discloses a kind of hand method for tracing, hand tracing equipment and gesture recognition system.Hand method for tracing includes:Judge whether covered in the frame before present frame;According to masking judging result, proficiency target is selected in the hand target adaptively tracked from frame above as template;And template matching method is carried out to obtain the hand target of present frame for present frame based on the template.
Description
Technical field
The present invention relates to field of image recognition, and relate more specifically to hand method for tracing, hand tracing equipment and gesture and know
Other system.
Background technology
With becoming more and more indispensable in the life Computer of current people and numerous portable intelligent devices,
People will wish that more natural and more efficient between people and computer interacts.Therefore, as freely touching human-computer interaction
The gesture remote control system at interface is by as noticeable selection.
Substantially, gesture remote control system will follow the trail of hand and analyze the expression of significant hand, if they are identified as
One kind in pre-defined gesture, then corresponding operational order will be triggered to operate certain application.As it can be seen that as premise
The hand method for tracing of robustness is very important for analyzing work manually and identification gesture.In addition, for low-power
The realization of equipment, it is also necessary to consider the complexity of tracing algorithm.
Currently, some algorithms or its variation have been introduced into follows the trail of for real time hand.For example, optical flow (Optical Flow)
Algorithm reflects that frame changes by the velocity for the movement pixel being attached in image, however, this needs a large amount of calculate;
CamShift algorithms can effectively follow the trail of the probability distribution dynamically changed in visual scene, however, this depends on suitable brightness
And it will become do not have robustness when there are other objects with the color similar with the colour of skin in tracing process;Template
Matching (Template Matching) algorithm can provide optimal selection so as to high by comparing the difference between candidate and template
Effect ground tracking target, but in some cases, tracking accuracy is unsatisfactory.
Therefore, it is necessary to it is a kind of can be when some objects with the color similar with the colour of skin be spanned through when hand
There can be the gesture remote control system of the hand method for tracing and application of high tracking accuracy the hand method for tracing.
The content of the invention
According to an aspect of the present invention, there is provided a kind of hand method for tracing, including:Judge whether sent out in the frame before present frame
Raw masking;According to masking judging result, proficiency target is selected in the hand target adaptively tracked from frame above
As template;And template matching method is carried out to obtain the hand target of present frame for present frame based on the template.
In one embodiment, be adaptive selected includes as the hand target of template:If in the frame before present frame
Cover, then tightly masking occur before frame obtained in hand target by as the template;If before present frame
Do not covered in frame, then the hand target obtained in the former frame of present frame is by as the template.
In one embodiment, the template matching method includes:The search range defined in the broca scale picture of present frame
With multiple candidate's hand scopes;Found out based on described search scope and the template in described search scope with the template most
The candidate's hand scope matched somebody with somebody is as the hand target tracked in the current frame.
In one embodiment, defining described search scope includes:By will be previous in the broca scale picture of present frame
The surrounding of the hand target tracked in frame expands the pixel of predetermined number to define initial search frequency range;And will be described initial
Search range is punctured into the hand target tracked included in former frame and the broca scale picture and former frame that represent present frame
Broca scale picture between differences in motion current colour of skin difference image including minimum rectangle as described search scope.
Wherein, template matching method further comprises:Multiple first candidate hand scopes defined in described search scope, should
A little first candidate hand scopes have the size identical with the template, also, in the broca scale picture and former frame for representing present frame
Broca scale picture between differences in motion current colour of skin difference image defined in second candidate's hand scope, the second candidate hand scope
With the size identical with the template, circulated for the multiple first candidate hand scope and perform matching judgment processing until this
Untill multiple first candidate hand scopes are all by matching judgment processing, so that it is determined that going out and the most matched candidate's hand model of the template
Enclose includes as the hand target tracked in the current frame, the matching judgment processing:Calculate first candidate's hand scope with
The average value of the absolute difference of the value of each pixel of the template is as the first error;If it is predetermined that first error is more than first
Threshold value, then current first candidate hand scope be judged as not with the template matches and thereby being excluded;If described first misses
Difference is less than first predetermined threshold, then calculates the second error, second error is by the way that first error is subtracted institute
The average value for stating the value of each pixel of second candidate's hand scope is multiplied by difference obtained from the value that modifying factor obtains;It is if described
Second error is less than the second predetermined threshold, then current first candidate hand scope be judged as with the template matches, it is and described
The average value of the value of each pixel of the value of second error and current first candidate hand scope is respectively as next
The value of each pixel of the second predetermined threshold and second candidate's hand scope in the matching judgment processing of one candidate's hand scope is averaged
Value.
In one embodiment, in the case of not covered in the frame before present frame, the modifying factor is the
One predetermined value, and in the case of being covered in the frame before present frame, the modifying factor for second predetermined value with it is tight
The average value of the value of each pixel in the hand target obtained in colour of skin difference image of the frame before masking occurs with its former frame
The sum of percentage.
In one embodiment, which further includes:Judge whether covered in present frame, it is described to judge bag
Include:If not following the trail of target in one's hands in the current frame, directly judge not cover in present frame;And if in present frame
Middle tracking target in one's hands, then perform masking condition judgment to determine whether cover in the current frame.
In one embodiment, the masking condition judgment comprises determining whether to meet the following conditions:(((PreOccDiff
<=A) && (DeltaDiff>B))||((PreOccDiff>A)&&(DeltaDiff>C)))&&(OccNum<D), wherein,
Each picture in the hand target that PreOccDiff expressions tightly obtain in colour of skin difference image of the frame before masking occurs with its former frame
The average value of the value of element, DeltaDiff is the difference of PreOccDiff and AvgTarDiff, and AvgTarDiff is current bad skin
In partial image with the average value of the value of each pixel in the range of the most matched candidate's hand of the template, OccNum represent present frame before
The number continuously covered occurs for face, wherein, C is more than B,
And wherein, judge to be covered in the current frame, if above-mentioned condition is discontented with if above-mentioned condition meets
Foot, then judge not cover in present frame.
According to the second aspect of the invention, there is provided a kind of hand tracing system, the system include:Judgment means, judge current
Whether covered in frame before frame;Selection device, according to masking judging result, is adaptively chased after from frame above
Track to hand target in select proficiency target as template;And device is obtained, mould is carried out for present frame based on the template
Plate matching process is to obtain the hand target of present frame.
According to the third aspect of the invention we, there is provided a kind of gesture recognition system, the gesture recognition system are chased after including above-mentioned hand
Track equipment.
Hand method for tracing and hand tracing equipment according to the present invention, span through with similar with the colour of skin even in when hand
Also there can be high tracking accuracy during some objects of color, so as to fulfill preparatively gesture identification.
Brief description of the drawings
Fig. 1 shows the frame diagram of general gesture recognition system;
Fig. 2 shows that the diagrammatic illustration of trace error occurs in conventional template matching algorithm when hand is moved before face;
Fig. 3 shows the flow chart of hand method for tracing according to the present invention;
Fig. 4 shows the more detailed flow chart of hand method for tracing according to the present invention;And
Fig. 5 shows diagrammatic illustration of the hand method for tracing according to the present invention in the tracking result when hand is moved before face.
Embodiment
The feature and exemplary embodiment of various aspects of the present invention is described more fully below.Following description covers many
Detail, in order to provide complete understanding of the present invention.It will be apparent, however, to one skilled in the art that
The present invention can be implemented in the case of some details being not required in these details.Description to embodiment below is only
It is to be provided by showing the example of the present invention to the clearer understanding of the present invention.The present invention is not limited to set forth below
Any concrete configuration and algorithm, but cover coherent element, component and calculation under the premise of without departing from the spirit of the present invention
Any modification, replacement and the improvement of method.
<1. gesture recognition system>
Fig. 1 shows the frame diagram of general gesture recognition system.As shown in Figure 1, gesture recognition system includes three functions
Part:Hand detection part, hand tracking part and gesture identification part.Specifically, the action of people is caught by common web camera
The picture frame for obtaining and being captured is input into the gesture recognition system.For each picture frame, first determine whether in former frame
In whether detect target in one's hands.If being also not detected by hand target in former frame, hand detection part will be in current whole frame
Middle detection hand target.If having detected that hand target in former frame, hand tracking part will be in present frame and ensuing
Hand target is followed the trail of in frame.At the same time, hand identification division by analysis of history tracing path to determine whether certain prearranged gesture quilt
Identify.If certain prearranged gesture is identified, corresponding gesture command will be output so that rear class uses, for example, to touch
Send out certain application.
If as described above, have detected that hand target in former frame, hand tracking part will be in present frame and under connecing
Hand target is followed the trail of in the frame come.As it can be seen that the calculation amount and tracking accuracy involved by tracking hand target are for gesture identification system
It is critically important for system.However, as described above, the calculation amount of existing hand method for tracing demand is larger and follow the trail of accuracy
It is unsatisfactory, particularly in the situation for being moved through some objects with the color similar with the colour of skin when hand, very may be used
Trace error can occur due to masking (Occlusion).For example, in traditional template matching algorithm, always former frame
Obtained in hand target be used as in present frame carry out template matching algorithm template, therefore when hand be moved through have with
In the situation of some objects of the similar color of the colour of skin due to masking occur and hand target template is misused or improper in use, chasing after
Track mistake can then occur.
For example, Fig. 2 is shown, and the example of trace error is occurring when hand is moved through face.As shown in Figure 2, given in figure
5 successive image frame 331-335 are gone out, in picture frame 331-333, normal template matching algorithm is performed, i.e. picture frame
Hand tracking in 331 based on the hand target for detecting or tracking in preceding a burst of picture frame as template, the hand in picture frame 322
Tracking is used as template based on the hand target tracked in picture frame 311, and the hand tracking in picture frame 333 is based in picture frame 332
The hand target tracked is as template.
However, as shown in the figure, due in picture frame 333 there is a situation where hand and face overlap, the institute in the picture frame 333
The hand target of acquisition contains a part for the face for being all the colour of skin, that is, the hand target obtained is inaccurate or mistake
's.Thus, when carrying out hand tracking for picture frame 334, with the inaccurate or wrong hand target obtained in picture frame 333
As template, stayed in the picture frame 334 by template matches by hand is obtained in error tracking result on the face.So
Continue, the tracking result of mistake will be also obtained in ensuing picture frame.As it can be seen that trace error occurs in a picture frame,
This mistake will cause the tracking in ensuing picture frame that mistake also accordingly occurs, therefore follow the trail of result inaccuracy, from
And influence the correct identification of gesture.
Confirm in view of the above problems, to provide a kind of hand tracking side using adaptive template matching algorithm in the present invention
Suitable template is adaptive selected to carry out mould by judging whether to occur in the frame before present frame masking in method, this method
Plate matching algorithm, so as to can also be obtained even if when in the situation that hand is moved through some objects with the color similar with the colour of skin
Obtain high tracking accuracy.
<2. hand method for tracing>
Hand method for tracing according to the present invention is described below with reference to Fig. 3.
Fig. 3 shows the flow chart of hand method for tracing according to the present invention.As shown in figure 3, being directed to each picture frame, step is performed
Rapid S301-S303.Specifically, in step S301, judge whether covered in the frame before present frame.In step S302
In, according to masking judging result, proficiency target conduct is selected in the hand target adaptively tracked from frame above
Template.If covered in the frame before present frame, tightly the hand target obtained in the frame before masking occurs is by conduct
The template;If do not covered in the frame before present frame, the hand target obtained in the former frame of present frame is made
For the template.Then, in step S303, template matching method is carried out to obtain currently for present frame based on the template
The hand target of frame.By using the present invention hand method for tracing, can with according to whether occur masking come select suitable template with
Template matching algorithm is performed, so as to avoid the error tracking because caused by occurring masking.
Next, hand method for tracing according to the present invention will be more fully described with reference to figure 4.
Fig. 4 shows the more detailed flow chart of hand method for tracing according to the present invention.As described in Figure 4, first, in step
In S401, the fortune between the broca scale picture and the broca scale picture of expression present frame and the broca scale picture of former frame of present frame is calculated
The current colour of skin difference image of moment.
Then, in step S402, search range is adaptively defined.The definition of search range includes following two steps
Suddenly:(1) predetermined number is expanded by the surrounding for the hand target that will be tracked in former frame in the broca scale picture of present frame
Pixel (for example, 15 pixels) defines initial search frequency range;And the initial search frequency range is punctured into comprising previous by (2)
The hand target and the minimum rectangle of the current colour of skin difference image tracked in frame.So just obtain and perform template
Search range used in method of completing the square.
Then, in step S403, judge whether covered in the frame before present frame.Afterwards, sentenced according to masking
Disconnected result selects proficiency target as carrying out template in present frame in the hand target that is tracked from the frame before present frame
The template of matching process.
Specifically, it is as shown in Figure 3, if judging not cover in the frame before present frame in step S403,
That is the judging result in step S403 is "No", then process proceeds to step S405.In step S405, such as traditional template
The same with algorithm, the hand target tracked in the former frame of present frame is as the mould that template matching method is carried out in present frame
Plate.At this time, the modifying factor k of differences in motion is fixed predetermined value, for example, k=0.3, wherein modifying factor k be used for later into
Capable template matching method.
On the other hand, if judging to be covered in the frame before present frame in step S403, i.e. step S403
In judging result be "Yes", then process proceeds to step S404.In step s 404, tightly chased after in the frame before masking occurs
Track target in one's hands is by as the template that template matching method is carried out in present frame.At this time, the modifying factor k of differences in motion is calculated as
Predetermined value and PreOccDiff percentage and, for example, k=0.4+PreOccDiff/100, wherein, PreOccDiff is represented
The average value of the value of each pixel in the hand target tightly obtained in colour of skin difference image of the frame before masking occurs with its former frame.
Judge for the masking in present frame, it will introduced in later step, and for the judgement whether covered of each frame
As a result can be stored for judging whether occur to cover and directly use in frame above in a later frame.
The search range defined in step S402 and after in step S404 or 405 have selected template, in step
In S406, template matching method is performed.In the present invention, which is to carry the modified template matches side of differences in motion
Method.The key of the present invention is to select template based on judging result is covered, and performs template based on the template so selected
Matching algorithm.Therefore, based on otherwise consideration, the modified template matching algorithm of differences in motion can also be used without.
Specifically, in the template matching method, multiple first candidate hands first defined in defined search range
Scope, each first candidate hand scope have the size identical with selected template, and in the current colour of skin difference image
Second candidate's hand scope is defined, this second candidate hand scope equally has the size identical with selected template.Exemplary
In embodiment, the hand target and the time in search range and defined in colour of skin difference image that are tracked in every frame
Player's scope is rectangle, and to simplify calculation amount, and the position of rectangle is represented with the vertex in the upper left corner of rectangle, so as to
With the position of unified Policy Updates hand target.
In the template matching method, for the multiple first candidate hand scope circulate perform matching judgment processing until
Untill this multiple first candidate hand scope is all passed through with judgement processing, so that it is determined that going out and selecting the most matched candidate's hand model of template
Enclose and include the following steps as the hand target tracked in the current frame, the matching judgment processing:
First candidate's hand scope is calculated with the average value of the absolute difference of the value of each pixel of selected template as first
Error;
If the first error is more than the first predetermined threshold, current first candidate hand scope be judged as not with selected template
Match and so as to be excluded;
If the first error is less than the first predetermined threshold, the second error is calculated, the second error is by by the first error
The average value for subtracting the value of each pixel of second candidate's hand scope is multiplied by difference obtained from the value that modifying factor obtains;
If the second error is less than the second predetermined threshold, current first candidate hand scope is judged as and selected template
Match somebody with somebody, and the average value of the value of the value of second error and each pixel in current first candidate hand scope is respectively as in pin
To the second predetermined threshold and each pixel of second candidate's hand scope in the matching judgment processing of next first candidate hand scope
Value average value.
Then, in step S 407, result based on the template matching algorithm performed in step S406 to determine whether
There are candidate's hand scope and selected template matches.Afterwards, judged in the current frame based on the information for judging result and former frame
Whether generation covers and updates relevant parameter for the processing in next frame.Specifically, if determined in step S 407
There is no candidate's hand scope and selected template matches, then processing proceeds to step S409, and directly judges to work as in step S409
Do not covered in previous frame.If having determined candidate's hand scope and selected template matches in step S 407, processing carries out
To step S408, and masking condition judgment is performed to determine whether cover in the current frame in step S408.
The masking condition judgment performed in step S408 comprises determining whether to meet the following conditions:(((PreOccDiff<
=A) && (DeltaDiff>B))||((PreOccDiff>A)&&(DeltaDiff>C)))&&(OccNum<D) wherein,
It is each in the hand target that PreOccDiff expressions tightly track in colour of skin difference image of the frame before masking occurs with its former frame
The average value of the value of pixel, DeltaDiff are the difference of PreOccDiff and AvgTarDiff, and AvgTarDiff is the current colour of skin
In difference image present frame is represented with the average value of the value of each pixel in the range of the most matched candidate's hand of selected template, OccNum
The number continuously covered above occurs, wherein, C is more than B, and wherein, if above-mentioned condition meets, sentences in step S410
It is fixed to be covered in the current frame, if above-mentioned condition is unsatisfactory for, judge not cover in present frame in step S410.
In above-mentioned formula, threshold value A-D is set based on experience.In the case of such as hand is move across face, usually,
It can probably be occurred in two frames with being covered during common camera shooting image, therefore for example, threshold value D can be set to 2, of course depended upon
Video camera uses the other factors such as frame per second and hand translational speed, and threshold value D's is also set to other suitable values.Separately
Outside, in fact, the intensity that PreOccDiff reflections are sold mobile, PreOccDiff is bigger, represent that hand moves faster or hand
Movement is stronger, and vice versa.Therefore, in the value to B, C, C can take the value bigger than B, for example, all taking 15 in threshold value A
In the case of, B, C can be taken as B=and C=7 respectively.It is exactly to have used given here specifically to show in the example that subsequent figures 4 provide
Example, i.e. A=15, B=4, C=5 and D=2.
Afterwards, in step S411 and step S412, related ginseng is updated according to the result in step S409, step S410
Number and tracing path.Specifically, in the case of successfully following the trail of target in one's hands in the current frame, if it is decided that do not send out in the current frame
Raw masking, then follow the trail of target in one's hands by as the template that template matching method is carried out in next frame in present frame, if it is decided that
Covered in present frame, then template used in template matching method is carried out in present frame and is still used to carry out template in next frame
The template of matching process.Also it is still before masking occurs in addition, in the case of not following the trail of target in one's hands successfully in the current frame
Target in one's hands is followed the trail of in frame by as the template that template matching method is carried out in next frame.
In the case of successfully following the trail of target in one's hands in the current frame, if it is decided that do not cover, then chase after in the current frame
Track track is updated to:PN(x, y)=P (x, y), wherein, PN(x, y) represents a left side for the rectangle hand target tracked in present frame
The coordinate position at upper angle, P (x, y) represent the position with the upper left corner of the most matched candidate's hand scope of selected template.That is,
The position in the upper left corner of the hand target tracked in present frame is the left side being determined with the most matched candidate's hand scope of selected template
The position at upper angle.If it is determined that being covered in the current frame, then tracing path is updated to:
Wherein PN-1(x, y) represents the position in the upper left corner of the hand target tracked in the former frame of present frame, and PN-2
The position in the upper left corner of the hand target tracked before (x, y) present frame in the second frame.In this way, the tracking process knot in present frame
Beam, afterwards into next frame.The above method so is performed for each frame, untill all frames are all processed, so as to obtain
The position of the hand target tracked in each frame is to identify the movement of hand.
Fig. 5 shows diagrammatic illustration of the hand method for tracing according to the present invention in the tracking result when hand is moved before face.
In Fig. 5, frame array above is the colour of skin difference image between each frame and its former frame, and is also shown for every frame
For AvgTargetDiff values to illustrate the effect of masking, this value, which can be used in step S408, judges whether masking condition meets.This
Outside, frame array below shows reference sequence of the renewal for the template of tracking process.As shown in FIG. 5, sentence in frame 80
Masking is made, so carrying out the template of template matching method in subsequent frame 81 based on being chased after in the frame 79 before masking
Track target in one's hands and be not based on the wrong hand target tracked in frame 80 so that even if when hand be moved through have and colour of skin class
As color object (being in this example face) when can also improve tracking accuracy.
The influence changed according to the present invention using the hand back tracking method of adaptive template matching algorithm from brightness, and not
Influenced by the object with the color similar with the colour of skin, and there is relatively low computation complexity to meet to realize that low-power is set
Standby requirement.
Hand method for tracing according to the present invention can be realized with software, hardware, firmware and other modes.Should for example, working as
When hand method for tracing is realized in a manner of software, realizing the program of the hand method for tracing can be embedded in storage medium, and lead to
Cross the computer being loaded and be operative to implement the present invention.When hand method for tracing according to the present invention is realized in a manner of hardware
When, which can be completed by corresponding component.Such as, it is possible to achieve such a hand tracing equipment, including:Sentence
Disconnected device, judges whether covered in the frame before present frame;Selection device, according to masking judging result, adaptively from
Proficiency target is selected in the hand target tracked in frame above as template;And device is obtained, based on the template
Template matching method is carried out for present frame to obtain the hand target of present frame.
In addition, hand method for tracing can be used alone according to the present invention, also can be as the part during gesture identification
Use, equally, hand method for tracing equipment according to the present invention can be used alone, and also can be attached to gesture identification as a component
In system.
It is to be described according to the order in given flow chart in the present specification, but the hand method for tracing of the present invention
Can sequentially it be realized with other, for example, some steps can be executed in parallel or certain step can perform before another step,
As long as meeting subject of the present invention thought, i.e., come from for each frame according to the judgement for whether being covered in frame above
Template is adaptively selected to carry out template matching algorithm, you can.
Hand method for tracing and hand tracing equipment according to the present invention are not limited to the above-mentioned purposes referred to, but can be applied to
Any occasion related with hand tracking.
Although with specific embodiment, the invention has been described above, it should be appreciated to those skilled in the art that according to design
It is required that various modifications, combination, sub-portfolio and replacement are may occur in which with other factors, as long as they fall in appended claims and its wait
In the range of jljl.
Claims (9)
1. a kind of hand method for tracing, including:
Judge whether covered in the frame before present frame;
According to masking judging result, proficiency target conduct is selected in the hand target adaptively tracked from frame above
Template;And
Template matching method is carried out to obtain the hand target of present frame for present frame based on the template,
Wherein, be adaptive selected includes as the hand target of template:
If covered in the frame before present frame, tightly the hand target obtained in the frame before masking occurs is by as institute
State template;
If do not covered in the frame before present frame, the hand target obtained in the former frame of present frame is by as described
Template.
2. hand method for tracing according to claim 1, wherein, the template matching method includes:
The search range defined in the broca scale picture of present frame and multiple candidate's hand scopes;
Based on described search scope and the template come find out in described search scope with the most matched candidate's hand model of the template
Enclose as the hand target tracked in the current frame.
3. hand method for tracing according to claim 2, wherein, defining described search scope includes:
Predetermined number is expanded by the surrounding for the hand target that will be tracked in former frame in the broca scale picture of present frame
Pixel defines initial search frequency range;And
The initial search frequency range is punctured into the hand target tracked included in former frame and the skin for representing present frame
Described in minimum rectangle including the current colour of skin difference image of the differences in motion of color image and the broca scale of former frame as between is used as
Search range.
4. the hand method for tracing according to Claims 2 or 3, wherein, the template matching method further comprises:
Multiple first candidate hand scopes defined in described search scope, those the first candidate hand scopes have and the template phase
Same size, also, the current colour of skin of the differences in motion between the broca scale picture and the broca scale picture of former frame for representing present frame
Second candidate's hand scope defined in difference image, the second candidate hand scope have the size identical with the template,
For the multiple first candidate hand scope circulate perform matching judgment processing until this multiple first candidate hand scope all
Untill being handled by matching judgment, followed the trail of in the current frame so that it is determined that going out and being used as with the most matched candidate's hand scope of the template
The hand target arrived,
The matching judgment processing includes:
First candidate's hand scope is calculated with the average value of the absolute difference of the value of each pixel of the template as the first error;
If first error is more than the first predetermined threshold, current first candidate hand scope be judged as not with the template
Match and thereby be excluded;
If first error is less than first predetermined threshold, calculate the second error, second error be pass through by
The average value that first error subtracts the value of each pixel of the second candidate hand scope be multiplied by the value that modifying factor obtains and
Obtained difference;
If second error is less than the second predetermined threshold, current first candidate hand scope is judged as and the template
Match somebody with somebody, and the average value of the value of each pixel of the value of second error and current first candidate hand scope is respectively as in pin
To the second predetermined threshold and each pixel of second candidate's hand scope in the matching judgment processing of next first candidate hand scope
Value average value.
5. hand method for tracing according to claim 4, wherein, there is a situation where cover in the frame before present frame
In, the modifying factor is first predetermined value, and in the case of being covered in the frame before present frame, the modifying factor
It is each in the hand target that son obtains for second predetermined value and tightly in colour of skin difference image of the frame before masking occurs with its former frame
The sum of the percentage of the average value of the value of pixel.
6. hand method for tracing according to claim 1, further includes:Judge whether covered in present frame, the judgement
Including:
If not following the trail of target in one's hands in the current frame, directly judge not cover in present frame;And
If following the trail of target in one's hands in the current frame, masking condition judgment is performed to determine whether hide in the current frame
Cover.
7. hand method for tracing according to claim 6, the masking condition judgment comprises determining whether to meet the following conditions:
(((PreOccDiff<=A) && (DeltaDiff>B))||((PreOccDiff>A)&&(DeltaDiff>C)))&&
(OccNum<D) wherein, PreOccDiff represents tightly to obtain in colour of skin difference image of the frame before masking occurs with its former frame
Hand target in each pixel value average value, DeltaDiff is the difference of PreOccDiff and AvgTarDiff,
AvgTarDiff is flat with the value of each pixel in the range of the most matched candidate's hand of the template in current colour of skin difference image
Average, OccNum represent that the number that continuously covers occurs before present frame, and wherein C is more than B,
And wherein, judge to be covered in the current frame if above-mentioned condition meets, if above-mentioned condition is unsatisfactory for,
Then judge not cover in present frame.
8. a kind of hand tracing equipment, including:
Judgment means, judge whether covered in the frame before present frame;
Selection device, according to masking judging result, one is selected in the hand target adaptively tracked from frame above
Hand target is as template;And
Device is obtained, template matching method is carried out to obtain the hand target of present frame for present frame based on the template,
Wherein, be adaptive selected includes as the hand target of template:If covered in the frame before present frame, tightly exist
The hand target obtained in frame before masking generation is by as the template;If do not covered in the frame before present frame,
Then the hand target obtained in the former frame of present frame is by as the template.
9. a kind of gesture recognition system, which includes hand tracing equipment as claimed in claim 8.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210375618.7A CN103679130B (en) | 2012-09-24 | 2012-09-24 | Hand method for tracing, hand tracing equipment and gesture recognition system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210375618.7A CN103679130B (en) | 2012-09-24 | 2012-09-24 | Hand method for tracing, hand tracing equipment and gesture recognition system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103679130A CN103679130A (en) | 2014-03-26 |
CN103679130B true CN103679130B (en) | 2018-04-13 |
Family
ID=50316622
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201210375618.7A Active CN103679130B (en) | 2012-09-24 | 2012-09-24 | Hand method for tracing, hand tracing equipment and gesture recognition system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103679130B (en) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104050454B (en) * | 2014-06-24 | 2017-12-19 | 深圳先进技术研究院 | A kind of motion gesture track acquisition methods and system |
US9558389B2 (en) * | 2015-03-24 | 2017-01-31 | Intel Corporation | Reliable fingertip and palm detection |
WO2018205120A1 (en) * | 2017-05-09 | 2018-11-15 | 深圳市速腾聚创科技有限公司 | Target tracking method, smart device and storage medium |
WO2020080535A1 (en) * | 2018-10-18 | 2020-04-23 | 国立研究開発法人科学技術振興機構 | Target tracking method, target tracking system, and target tracking program |
CN111507192A (en) * | 2020-03-19 | 2020-08-07 | 北京捷通华声科技股份有限公司 | Appearance instrument monitoring method and device |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102509079A (en) * | 2011-11-04 | 2012-06-20 | 康佳集团股份有限公司 | Real-time gesture tracking method and tracking system |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20080073933A (en) * | 2007-02-07 | 2008-08-12 | 삼성전자주식회사 | Object tracking method and apparatus, and object pose information calculating method and apparatus |
CN101127122A (en) * | 2007-09-13 | 2008-02-20 | 复旦大学 | Content self-adaptive gradual-progression type sheltering analysis target tracking algorism |
CN101251928A (en) * | 2008-03-13 | 2008-08-27 | 上海交通大学 | Object tracking method based on core |
JP5601045B2 (en) * | 2010-06-24 | 2014-10-08 | ソニー株式会社 | Gesture recognition device, gesture recognition method and program |
-
2012
- 2012-09-24 CN CN201210375618.7A patent/CN103679130B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102509079A (en) * | 2011-11-04 | 2012-06-20 | 康佳集团股份有限公司 | Real-time gesture tracking method and tracking system |
Non-Patent Citations (2)
Title |
---|
分块跟踪中的目标模板更新方法;齐美彬等;《中国图象图形学报》;20110630;全文 * |
基于三维模型的单目图象序列头部姿态跟踪;宋杰;《中国优秀硕士学位论文全文数据库信息科技辑》;20080815;全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN103679130A (en) | 2014-03-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
KR102150776B1 (en) | Face location tracking method, apparatus and electronic device | |
CN109711304B (en) | Face feature point positioning method and device | |
EP3084577B1 (en) | Selection and tracking of objects for display partitioning and clustering of video frames | |
CN108960163B (en) | Gesture recognition method, device, equipment and storage medium | |
US9317772B2 (en) | Method for improving tracking using dynamic background compensation with centroid compensation | |
CN106951346B (en) | Response time testing method and device | |
US10620826B2 (en) | Object selection based on region of interest fusion | |
JP5498454B2 (en) | TRACKING DEVICE, TRACKING METHOD, AND PROGRAM | |
JP6939111B2 (en) | Image recognition device and image recognition method | |
WO2019023921A1 (en) | Gesture recognition method, apparatus, and device | |
CN103679130B (en) | Hand method for tracing, hand tracing equipment and gesture recognition system | |
US20150205997A1 (en) | Method, apparatus and computer program product for human-face features extraction | |
US20120243733A1 (en) | Moving object detecting device, moving object detecting method, moving object detection program, moving object tracking device, moving object tracking method, and moving object tracking program | |
KR100660725B1 (en) | Portable terminal having apparatus for tracking human face | |
JP2008538832A (en) | Estimating 3D road layout from video sequences by tracking pedestrians | |
CN109829933B (en) | Image target tracking method and device | |
CN103677274A (en) | Interactive projection method and system based on active vision | |
US20130036389A1 (en) | Command issuing apparatus, command issuing method, and computer program product | |
US20150104067A1 (en) | Method and apparatus for tracking object, and method for selecting tracking feature | |
CN113688820A (en) | Stroboscopic stripe information identification method and device and electronic equipment | |
CN113194253A (en) | Shooting method and device for removing image reflection and electronic equipment | |
KR20160079531A (en) | Method and apparatus for processing gesture input | |
Nguyen et al. | Real time human tracking using improved CAM-shift | |
KR101853276B1 (en) | Method for detecting hand area from depth image and apparatus thereof | |
JP2007510994A (en) | Object tracking in video images |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |