CN108073872A - Hand region recognition methods and device based on similarity between pixel - Google Patents
Hand region recognition methods and device based on similarity between pixel Download PDFInfo
- Publication number
- CN108073872A CN108073872A CN201611020599.0A CN201611020599A CN108073872A CN 108073872 A CN108073872 A CN 108073872A CN 201611020599 A CN201611020599 A CN 201611020599A CN 108073872 A CN108073872 A CN 108073872A
- Authority
- CN
- China
- Prior art keywords
- pixel
- value
- hand region
- similarity
- presumptive area
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/107—Static hand or arm
Landscapes
- Engineering & Computer Science (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 a kind of hand region recognition methods and device based on similarity between pixel, the described method includes:The color value of the pixel in hand region to be identified is compared respectively to obtain similarity w1 with the color value of the first presumptive area, wherein first presumptive area is located in the hand region to be identified;The color value of the pixel is compared respectively to obtain similarity w2 with the color value of the second presumptive area, wherein second presumptive area and the hand region to be identified are misaligned;Respectively the color value of the pixel is compared to obtain similarity value w3 with the color value of the pixel surrounding pixel;The steady angle value w4 of the pixel color variation in the pixel to the shortest path and the shortest path of first presumptive area is determined respectively;Determine that the pixel is the probability value of pixel in hand region according to w1, w2, w3 and w4;Hand region is determined according to the probability value.
Description
Technical field
The present invention relates to field of image recognition, and in particular to a kind of hand region recognition methods based on similarity between pixel
And device.
Background technology
With integrating for the fast development of software and hardware relevant technical, such as wrist intelligent apparatus, smartwatch, Intelligent bracelet etc.
Spend higher and higher, function is increasingly abundanter, and the cell-phone function of significant proportion can be realized by smartwatch, Intelligent bracelet, greatly
The big method for simplifying user and receiving and transferring information.But compare with traditional smart phone, wrist intelligent apparatus is limited to small size
Display screen, on the one hand, user can not complete the operation of correlation function using touch screen or button well when in use, easily
Maloperation is caused, on the other hand, when smartwatch is worn on one on hand, to be operated on it, except waking up, dormancy etc.
Simple operations are not required the operation of another hand outer, remaining more complicated operation has the completion of another hand, can not use single
Hand independently smartwatch is operated, therefore, smartwatch on content shows and operates there are still it is very big the defects of.
The control mode based on user's hand gesture is some products provides in view of the above-mentioned problems, having at present, user can be with
By wearing the finger movement of wrist-watch come control device, principle is to utilize the photographic device acquisition hand figure set on wrist-watch
Picture, processor determine control content according to the variation of image.Such scheme needs identify the hand of user exactly from image
Portion region or profile need to reject background and other noises, only determine control content according to the variation of hand region.
Existing image identifying schemes are typically the shape feature according to target object, according to default characteristic parameter, profit
Target object is identified from image with means such as machine learning models.But this identification method resists the ability of background interference
It is poor, such as when user's local environment background is sufficiently complex, be likely to judge by accident according to shape feature, it can be seen that existing
Image identifying schemes accuracy is poor.
The content of the invention
The present invention is to solve the problem of existing image identifying schemes accuracy is poor.
In view of this, the present invention provides a kind of hand region recognition methods based on similarity between pixel, which is characterized in that
Including:The color value of the pixel in hand region to be identified is compared respectively to obtain with the color value of the first presumptive area
To similarity w1, wherein first presumptive area is located in the hand region to be identified;Respectively by the face of the pixel
Color value is compared to obtain similarity w2 with the color value of the second presumptive area, wherein second presumptive area is treated with described
The hand region of identification is misaligned;The color value of the pixel is compared with the color value of the pixel surrounding pixel respectively
To obtain similarity value w3;Shortest path and the shortest path of the pixel to first presumptive area are determined respectively
On pixel color variation steady angle value w4;Determine that the pixel is the general of pixel in hand region according to w1, w2, w3 and w4
Rate value;Hand region is determined according to the probability value.
Preferably, before the step of calculating w1, w2, w3 and w4, further include:
Hand region to be identified is determined in the picture;
First presumptive area is rejected from the hand region to be identified.
Preferably, determine that the pixel is the step of the probability value of pixel in hand region according to w1, w2, w3 and w4 described
In rapid, the probability value and w1, w3, w4 correlation, the probability value and the negatively correlated relations of w2.
Preferably, it is described to determine that the pixel is that the probability value of pixel in hand region includes according to w1, w2, w3 and w4:
The each similarity, the corresponding weights of the smoothness are determined respectively;
According to determining the product and the product of the smoothness and corresponding weight value of each similarity and corresponding weight value
Probability value.
Preferably, the hand region is determined using active contour model, and the probability value is as the active
The input value of Active contour model, the active contour model is according to the probability value, predetermined pull parameter, predetermined gravitational parameter
Hand region is obtained with predetermined initial profile.
Preferably, the similarity and smoothness are the similitudes according to the color value under RGB, HSV and YCrCb space
What the summation of indication information determined.
Correspondingly, the present invention also provides a kind of hand region identification device based on similarity between pixel, including:First phase
Like degree determination unit, the color value of the pixel in hand region to be identified and the color value of the first presumptive area are carried out respectively
It compares to obtain similarity w1, wherein first presumptive area is located in the hand region to be identified;Second similarity
Determination unit the color value of the pixel is compared to obtain similarity w2 respectively with the color value of the second presumptive area,
Wherein described second presumptive area and the hand region to be identified are misaligned;Third phase seemingly spends determination unit, respectively by institute
The color value for stating pixel is compared to obtain similarity value w3 with the color value of the pixel surrounding pixel;Smoothness determines list
Member determines that the pixel color in the pixel to the shortest path and the shortest path of first presumptive area becomes respectively
The steady angle value w4 changed;Identifying unit, for determining that the pixel is the general of pixel in hand region according to w1, w2, w3 and w4
Rate value;Determination unit, for determining hand region according to the probability value.
Preferably, further include:
Recognition unit, for true like spending in first similarity determining unit, the second similarity determining unit, third phase
Before order member and smoothness determination unit are handled, hand region to be identified is determined in the picture;
Culling unit, for rejecting first presumptive area from the hand region to be identified.
Preferably, the probability value and w1, w3, w4 correlation, the probability value and the negatively correlated relations of w2.
Preferably, the identifying unit includes:
Weights determination unit, for determining each similarity, the corresponding weights of the smoothness respectively;
Probability value determination unit, for the product according to each similarity and corresponding weight value and the smoothness and accordingly
The product of weights determines the probability value.
The hand region recognition methods based on similarity between pixel provided according to embodiments of the present invention and device, pass through by
The color value of pixel in hand images to be identified is compared with the pixel color on each position, to determine in image
Each pixel is the probability value of the pixel in hand region, so according to the probability value and each pixel and presumptive area away from
From hand region is determined, compared to the existing scheme that hand region is determined according to shape feature, the embodiment of the present invention possesses
Higher accuracy.And this programme is by pixel to be identified and itself surrounding pixel, the color value of the first presumptive area, second
Pixel in presumptive area is compared and thus integrates definite above-mentioned probability value so that this programme possesses stronger anti-interference
Ability.
Description of the drawings
It, below will be to specific in order to illustrate more clearly of the specific embodiment of the invention or technical solution of the prior art
Embodiment or attached drawing needed to be used in the description of the prior art are briefly described, it should be apparent that, in describing below
Attached drawing is some embodiments of the present invention, for those of ordinary skill in the art, before not making the creative labor
It puts, can also be obtained according to these attached drawings other attached drawings.
Fig. 1 is a kind of structure diagram of wearable device;
Fig. 2 is the image of the shooting of equipment shown in Fig. 1;
Fig. 3 is to carry out discoloration treated image to Fig. 2;
Fig. 4 is the flow chart provided in an embodiment of the present invention based on the hand region recognition methods of similarity between pixel;
Fig. 5 is the structure chart provided in an embodiment of the present invention based on the hand region identification device of similarity between pixel.
Specific embodiment
Technical scheme is clearly and completely described below in conjunction with attached drawing, it is clear that described implementation
Example is part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill
Personnel's all other embodiments obtained without making creative work, belong to the scope of protection of the invention.
As long as in addition, technical characteristic involved in invention described below different embodiments non-structure each other
It can be combined with each other into conflict.
An embodiment of the present invention provides a kind of hand region recognition methods based on similarity between pixel, handled by this method
Image be the image shot by the wearable device with photographic device, the equipment is as shown in Figure 1, wherein 01 edge of photographic device
Wearer's wrist gathers wearer's hand images to palm of the hand direction.Its image collected is as shown in Fig. 2, the palm of wearer exists
Integral position in image is more fixed, and simply finger areas can be sent out in a fixed scope with the movement of user
Changing.As shown in figure 3, method provided in this embodiment includes the following steps:
S1 respectively compares the color value of the pixel in hand region to be identified and the color value of the first presumptive area
To obtain similarity w1, wherein first presumptive area is located in the hand region to be identified;
S2 the color value of the pixel is compared to obtain similarity respectively with the color value of the second presumptive area
W2, wherein second presumptive area and the hand region to be identified are misaligned;
The color value of the pixel is compared to obtain similar by S3 with the color value of the pixel surrounding pixel respectively
Angle value w3;
S4 determines the picture in the pixel to the shortest path and the shortest path of first presumptive area respectively
The steady angle value w4 of plain color change;
As shown in Fig. 2, the first presumptive area 11,12 and second presumptive area 13 of hand region to be identified are wherein included,
Usually also referred to as the first presumptive area 11 is seed region.On the selection in above three region, fixed position can be preset
It puts, as long as user normally wears, the first presumptive area 11 must be a part for user's palm, and the second presumptive area 13 must not
Including user's palm.In order to more clearly describe the present embodiments relate to each area, below use to Fig. 2 carry out discoloration
Fig. 4 afterwards is illustrated, it is necessary to illustrate, the embodiment of the present invention needs the color for relying on pixel to carry out subsequent processing, so
Fig. 4 is intended merely to clearly demonstrate given image, and when practical application need not carry out discoloration processing.
On above-mentioned steps S1, since the pixel in the first presumptive area must be the pixel in hand region, mesh
It marks pixel and its similarity is higher, then it represents that object pixel is that the probability of hand region pixel is bigger.First presumptive area
Color can be wherein all pixels color value average.
On above-mentioned steps S2, since the pixel in the second presumptive area is not centainly the pixel in hand region,
Object pixel is lower with its similarity, then it represents that object pixel is that the probability of hand region pixel is bigger.Second fate
The color in domain can be the average of wherein all pixels color value.
On above-mentioned steps S3, hand region is close as color and closed area, each pixel therein and surrounding picture
The color similarity of element should be higher, be then the pixel in hand edge if there is larger situation is differed, therefore
Object pixel and adjacent pixel similarity are higher, then it represents that object pixel is that the probability of hand region pixel is bigger.Alignments
Including a variety of, for example, by pixel a and N number of pixel around its own be compared to obtain respectively N number of similarity value (such as
8 neighborhoods around N=8, i.e. pixel), then calculate the average of N number of similarity;Or the color average of N number of pixel is first calculated, so
The color of pixel a is compared afterwards to obtain similarity with the average, is all feasible.
On above-mentioned steps S4, hand region is close as color and closed area, each pixel to first
The color change between each pixel on the shortest path of presumptive area should be subtleer and gentle, if there is change
The situation for having interruption on path may then be represented by changing larger situation, which may not be the pixel in hand region.Therefore
Object pixel and the smoothness of the pixel color variation on the shortest path of the first presumptive area are higher, then it represents that object pixel is
The probability of hand region pixel is bigger.
It is compared on the color value between pixel, the embodiment of the present invention can use the color value under a certain color space
It is compared, the combination of the color value under a variety of kinds of color spaces can also be used to be compared, such as can be according to RGB, HSV
And the summation of the similarity indices information of the color value under YCrCb spaces determines the similarity of two pixels.
It will be understood by those skilled in the art that complementary relation is not present between above-mentioned steps S1-S4, thus it is above-mentioned
It is all feasible that step execution sequence performs in no particular order or simultaneously.In above-mentioned steps, by hand region 12 to be identified
In pixel be compared respectively with the pixel in above-mentioned various positions after can to obtain 3 similarity values and 1 steady
Angle value.After all pixels in hand region 12 to be identified are compared, each pixel is corresponding with 3 similarity values and 1
A steady angle value.
S5 determines that the pixel is the probability value of pixel in hand region according to w1, w2, w3 and w4.I.e. for be identified
Hand region 12 in each pixel, using its corresponding similarity value and smoothness value add up to calculate a probability value,
The probability value is more high, and it is that probability in hand region is also big to represent the pixel, on the contrary then smaller.To all pixels COMPREHENSIVE CALCULATING
After finishing, each pixel corresponds to a probability value.
Comprehensive multiple numerical value calculate there are many modes of a probability value, in calculating formula, probability value and above-mentioned w1,
W2, w3, w4 can in a linear relationship, non-linear relation, and should determine in the case where w2 is constant, with w1, w3, w4
Increase, the probability value calculated should increase, on the contrary then reduce;In the case where w1, w3, w4 are constant, with the increase of w2, meter
The probability value of calculating should reduce, on the contrary then increase, i.e., described probability value and w1, w3, w4 correlation, the probability value
With the negatively correlated relations of w2.
S6 determines hand region according to probability value, and there are many methods of determination, such as can use simplest mode, leads to
Setting probability threshold value probability value corresponding with each pixel is crossed to compare, you can the pixel in all hand regions is filtered out, these
The summation of pixel is hand region.Preferably, in the present embodiment, a kind of active contour model may be employed (also referred to as
Snake models) identify hand region.Specifically, input value of the above-mentioned probability value as active contour model, active profile
Line model obtains in one's hands according to the corresponding probability value of each pixel, predetermined pull parameter, predetermined gravitational parameter and predetermined initial profile
Portion region.Each point on model will move to interest edges of regions under gravitation and pulling force effect, while profile keeps certain
Curvature, possess smooth characteristic.Finally, after each point movement stops, that is, the light of interest region (hand region) is identified
Sliding edge to get to be continuously smooth hand region outer profile.
The hand region recognition methods based on similarity between pixel provided according to embodiments of the present invention, by will be to be identified
Hand images in the color value of pixel be compared with the pixel color on each position, to determine each picture in image
Element is the probability value of the pixel in hand region, and then according to the probability value and the distance of each pixel and presumptive area come really
Determine hand region, compared to the existing scheme that hand region is determined according to shape feature, the embodiment of the present invention possesses higher
Accuracy.And this programme is by pixel to be identified and itself surrounding pixel, the color value of the first presumptive area, the second fate
Pixel in domain is compared and thus integrates definite above-mentioned probability value so that this programme possesses stronger antijamming capability.
As a preferred embodiment, before above-mentioned steps S1-S4, can also include the following steps:
S01 determines hand region to be identified in the picture.As described above, according to the characteristics of image, hand to be identified
The position in region 12 can also be demarcated in advance, 1/2,3/4,4/5 region etc. e.g. from image border to image, specifically
Predetermined focal distance and viewfinder range when can be according to captured image determine.
S02 rejects the first presumptive area from hand region to be identified, since the first presumptive area must be palm
A part, therefore can this partial pixel be gone divided by be reduced calculation amount.
Above-mentioned preferred embodiment determines hand region to be identified in the picture, and must be hand region from wherein rejecting
Part so that subsequent step can be calculated just for remaining region, it is possible thereby to reduce calculation amount, improve identification effect
Rate.
Further, above-mentioned steps S4 may include steps of:
S41 calculates pixel to the cost value in each path of the first presumptive area respectively, and cost value is according on path
What the value of pixel determined;
S42 chooses the path with minimum cost value from each path.
Further, above-mentioned steps S41 may include steps of:
The color value of pixel on path is built curve by S411 respectively;
S412, the second-order partial differential coefficient of calculated curve;
S413 calculates the variance of second order local derviation as cost value.
According to above-mentioned preferred embodiment, such image upper pathway can be searched in the picture first so that object pixel
O'clock to the center of the first presumptive area 11 Least-cost, the calculating of cost can be based on one passed through on image path
What the color value of series of pixels point determined, it can calculate in the following way:The value of pixel builds a curve successively on path,
The second-order partial differential coefficient of the curve is calculated, counts the variance of each point second order local derviation.Variance yields is cost, it is clear that second order local derviation
The stationarity of the variation of picture point on delegated path, the variation that the variance of local derviation represents picture point pixel value in whole upper pathway are steady
Property.
As described above, the comparison situation between pixel, including a variety of, above-mentioned 4 kinds of situations are set forth in detail in the present embodiment, this
The reliability of a little comparison results may be incorporated into the concept of weights to calculate above-mentioned probability value there may be different,
That is above-mentioned steps S5 may include steps of:
S51 determines that above-mentioned similarity and the corresponding weights of smoothness, i.e., above-mentioned w1, w2, w3, w4 can be corresponded to not respectively
Same weights;
S52 is determined according to the product and the product of the smoothness and corresponding weight value of each similarity and corresponding weight value
The probability value, the weights corresponding to above-mentioned various similarities, smoothness can be it is different can also be it is identical, specifically may be used
Reliability according to content is compared is set, and is finally determined generally according to the product of a variety of similarities and corresponding weights to integrate
Thus rate value further improves identification accuracy.
An alternative embodiment of the invention additionally provides a kind of hand region identification device based on similarity between pixel, such as
Shown in Fig. 5, which includes:
First similarity determining unit 51, respectively makes a reservation for the color value of the pixel in hand region to be identified and first
The color value in region is compared to obtain similarity w1, wherein first presumptive area is located at the hand area to be identified
In domain;
Second similarity determining unit 52 respectively carries out the color value of the pixel and the color value of the second presumptive area
It compares to obtain similarity w2, wherein second presumptive area and the hand region to be identified are misaligned;
Third phase is like determination unit 53 is spent, respectively by the color value of the pixel and the color value of the pixel surrounding pixel
It is compared to obtain similarity value w3;
Smoothness determination unit 54 determines the pixel to the shortest path of first presumptive area and described respectively
The steady angle value w4 of pixel color variation on shortest path;
Identifying unit 55, for determining that the pixel is the probability value of pixel in hand region according to w1, w2, w3 and w4;
Determination unit 56, for determining hand region according to the probability value.
The hand region identification device based on similarity between pixel provided according to embodiments of the present invention, by will be to be identified
Hand images in the color value of pixel be compared with the pixel color on each position, to determine each picture in image
Element is the probability value of the pixel in hand region, and then according to the probability value and the distance of each pixel and presumptive area come really
Determine hand region, compared to the existing scheme that hand region is determined according to shape feature, the embodiment of the present invention possesses higher
Accuracy.And this programme is by pixel to be identified and itself surrounding pixel, the color value of the first presumptive area, the second fate
Pixel in domain is compared and thus integrates definite above-mentioned probability value so that this programme possesses stronger antijamming capability.
Optionally, which can also further include:
Recognition unit 501, in first similarity determining unit, the second similarity determining unit, third phase seemingly
Before degree determination unit and smoothness determination unit are handled, hand region to be identified is determined in the picture;
Culling unit 502, for rejecting first presumptive area from the hand region to be identified.
Above-mentioned preferred embodiment determines hand region to be identified in the picture, and must be hand region from wherein rejecting
Part so that subsequent step can be calculated just for remaining region, it is possible thereby to reduce calculation amount, improve identification effect
Rate.
Preferably, the probability value and w1, w3, w4 correlation, the probability value and the negatively correlated relations of w2.
Preferably, the identifying unit can include:
Weights determination unit, for determining each similarity, the corresponding weights of the smoothness respectively;
Probability value determination unit, for the product according to each similarity and corresponding weight value and the smoothness and accordingly
The product of weights determines the probability value.
Obviously, the above embodiments are merely examples for clarifying the description, and is not intended to limit the embodiments.It is right
For those of ordinary skill in the art, can also make on the basis of the above description it is other it is various forms of variation or
It changes.There is no necessity and possibility to exhaust all the enbodiments.And the obvious variation thus extended out or
Among changing still in the protection domain of the invention.
Claims (10)
1. a kind of hand region recognition methods based on similarity between pixel, which is characterized in that including:
The color value of the pixel in hand region to be identified is compared respectively to obtain with the color value of the first presumptive area
To similarity w1, wherein first presumptive area is located in the hand region to be identified;
The color value of the pixel is compared respectively to obtain similarity w2 with the color value of the second presumptive area, wherein institute
It states the second presumptive area and the hand region to be identified is misaligned;
Respectively the color value of the pixel is compared to obtain similarity value w3 with the color value of the pixel surrounding pixel;
The pixel color in the pixel to the shortest path and the shortest path of first presumptive area is determined respectively
The steady angle value w4 of variation;
Determine that the pixel is the probability value of pixel in hand region according to w1, w2, w3 and w4;
Hand region is determined according to the probability value.
2. according to the method described in claim 1, it is characterized in that, before the step of calculating w1, w2, w3 and w4, further include:
Hand region to be identified is determined in the picture;
First presumptive area is rejected from the hand region to be identified.
3. according to the method described in claim 1, it is characterized in that, determine that the pixel is according to w1, w2, w3 and w4 described
In hand region the step of the probability value of pixel in, the probability value and w1, w3, w4 correlation, the probability value with
The negatively correlated relations of w2.
4. method according to any one of claim 1-3, which is characterized in that described that institute is determined according to w1, w2, w3 and w4
Stating the probability value that pixel is pixel in hand region includes:
The each similarity, the corresponding weights of the smoothness are determined respectively;
The probability is determined according to the product and the product of the smoothness and corresponding weight value of each similarity and corresponding weight value
Value.
5. according to the described method of any one of claim 1-4, which is characterized in that the hand region is to utilize active profile
What line model determined, input value of the probability value as the active contour model, the active contour model according to
The probability value, predetermined pull parameter, predetermined gravitational parameter and predetermined initial profile obtain hand region.
6. according to the method any one of claim 1-5, which is characterized in that the similarity and smoothness are bases
What the summation of the similarity indices information of the color value under RGB, HSV and YCrCb space determined.
7. a kind of hand region identification device based on similarity between pixel, which is characterized in that including:
First similarity determining unit, respectively by the color value of the pixel in hand region to be identified and the first presumptive area
Color value is compared to obtain similarity w1, wherein first presumptive area is located in the hand region to be identified;
Second similarity determining unit, the color value of the pixel is compared with the color value of the second presumptive area respectively with
Similarity w2 is obtained, wherein second presumptive area and the hand region to be identified are misaligned;
Third phase seemingly spends determination unit, respectively compares the color value of the pixel and the color value of the pixel surrounding pixel
To obtain similarity value w3;
Smoothness determination unit determines shortest path and the shortest path of the pixel to first presumptive area respectively
The steady angle value w4 of pixel color variation on footpath;
Identifying unit, for determining that the pixel is the probability value of pixel in hand region according to w1, w2, w3 and w4;
Determination unit, for determining hand region according to the probability value.
8. device according to claim 7, which is characterized in that further include:
Recognition unit, for definite single like spending in first similarity determining unit, the second similarity determining unit, third phase
Before member and smoothness determination unit are handled, hand region to be identified is determined in the picture;
Culling unit, for rejecting first presumptive area from the hand region to be identified.
9. the device according to claim 7 or 8, which is characterized in that the probability value and w1, w3, w4 correlation,
The probability value and the negatively correlated relations of w2.
10. according to the device any one of claim 7-9, which is characterized in that the identifying unit includes:
Weights determination unit, for determining each similarity, the corresponding weights of the smoothness respectively;
Probability value determination unit, for the product according to each similarity and corresponding weight value and the smoothness and corresponding weight value
Product determine the probability value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611020599.0A CN108073872A (en) | 2016-11-18 | 2016-11-18 | Hand region recognition methods and device based on similarity between pixel |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611020599.0A CN108073872A (en) | 2016-11-18 | 2016-11-18 | Hand region recognition methods and device based on similarity between pixel |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108073872A true CN108073872A (en) | 2018-05-25 |
Family
ID=62160529
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611020599.0A Pending CN108073872A (en) | 2016-11-18 | 2016-11-18 | Hand region recognition methods and device based on similarity between pixel |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108073872A (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101144158B1 (en) * | 2010-12-20 | 2012-05-10 | 전남대학교산학협력단 | Method of lip region for lip-reading in a mobile device |
CN104063876A (en) * | 2014-01-10 | 2014-09-24 | 北京理工大学 | Interactive image segmentation method |
CN104537355A (en) * | 2015-01-12 | 2015-04-22 | 中南大学 | Remarkable object detecting method utilizing image boundary information and area connectivity |
CN105184763A (en) * | 2014-06-23 | 2015-12-23 | 索尼公司 | Image processing method and device |
-
2016
- 2016-11-18 CN CN201611020599.0A patent/CN108073872A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101144158B1 (en) * | 2010-12-20 | 2012-05-10 | 전남대학교산학협력단 | Method of lip region for lip-reading in a mobile device |
CN104063876A (en) * | 2014-01-10 | 2014-09-24 | 北京理工大学 | Interactive image segmentation method |
CN105184763A (en) * | 2014-06-23 | 2015-12-23 | 索尼公司 | Image processing method and device |
CN104537355A (en) * | 2015-01-12 | 2015-04-22 | 中南大学 | Remarkable object detecting method utilizing image boundary information and area connectivity |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11546505B2 (en) | Touchless photo capture in response to detected hand gestures | |
US20220206588A1 (en) | Micro hand gestures for controlling virtual and graphical elements | |
US11531402B1 (en) | Bimanual gestures for controlling virtual and graphical elements | |
US11925863B2 (en) | Tracking hand gestures for interactive game control in augmented reality | |
US20220334649A1 (en) | Hand gestures for animating and controlling virtual and graphical elements | |
US11520399B2 (en) | Interactive augmented reality experiences using positional tracking | |
US20110193778A1 (en) | Device and method for controlling mouse pointer | |
EP3899644A1 (en) | Multi-camera cross reality device | |
CN106464800B (en) | Method, system, machine readable media and the calculating equipment that automatic anti-dazzle for imaging device exposes | |
US20210405363A1 (en) | Augmented reality experiences using social distancing | |
US20230017052A1 (en) | Head-related transfer function | |
US20140044342A1 (en) | Method for generating 3d coordinates and mobile terminal for generating 3d coordinates | |
CN109819206A (en) | Object method for tracing and its system and computer-readable storage medium based on image | |
CN109328355A (en) | Method and system for intelligent group portrait | |
EP4281901A1 (en) | Action recognition using pose data and machine learning | |
CN108073935A (en) | Method and device based on adjacent pixels and communication path identification hand region | |
CN108073872A (en) | Hand region recognition methods and device based on similarity between pixel | |
CN108073871A (en) | Method and device based on two kinds of presumptive area identification hand regions | |
CN108073271A (en) | Method and device based on presumptive area identification hand region | |
CN108073870A (en) | Method and device based on seed region and communication path identification hand region | |
CN108073878A (en) | Method and device based on presumptive area and communication path identification hand region | |
CN108073881A (en) | Method and device based on presumptive area and adjacent pixels identification hand region | |
CN108073937A (en) | A kind of method and device for identifying hand region in the picture | |
CN108073877A (en) | Method and device based on seed region and adjacent pixels identification hand region | |
CN108073882A (en) | Hand region recognition methods and device based on communication path |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180525 |
|
RJ01 | Rejection of invention patent application after publication |