CN105320443B - The method and device of gesture switching channels - Google Patents
The method and device of gesture switching channels Download PDFInfo
- Publication number
- CN105320443B CN105320443B CN201410352753.9A CN201410352753A CN105320443B CN 105320443 B CN105320443 B CN 105320443B CN 201410352753 A CN201410352753 A CN 201410352753A CN 105320443 B CN105320443 B CN 105320443B
- Authority
- CN
- China
- Prior art keywords
- gestures
- images
- cluster
- cluster centre
- value
- 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
- 238000000034 method Methods 0.000 title claims abstract description 48
- 239000011159 matrix material Substances 0.000 claims abstract description 101
- 239000013598 vector Substances 0.000 claims abstract description 88
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 18
- 238000005192 partition Methods 0.000 claims description 16
- 239000007787 solid Substances 0.000 claims description 13
- 238000007621 cluster analysis Methods 0.000 claims description 8
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 238000010586 diagram Methods 0.000 description 24
- 235000013399 edible fruits Nutrition 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 241001062009 Indigofera Species 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a kind of method of gesture switching channels, this method includes：Acquire images of gestures；Laplacian Matrix is built, feature decomposition is carried out to the Laplacian Matrix of structure, obtains multiple feature vectors；The neat angle value for calculating each feature vector generates objective matrix according to two feature vectors of neat angle value minimum；Clustering is carried out to the objective matrix of generation, according to clustering as a result, obtaining the images of gestures of binaryzation；Obtain the maximum comparability value of the images of gestures and each sample images of gestures to prestore of binaryzation, switching channels to the corresponding associated channel of sample images of gestures of maximum comparability value.The invention also discloses a kind of devices of gesture switching channels.Compared to the prior art, user of the present invention is without the use of remote controler, and the switching of channel need to be only can be achieved with by gesture, convenient for users to use, and improves the efficiency of switching channels.
Description
Technical field
The present invention relates to graphics process field more particularly to the method and devices of gesture switching channels.
Background technology
User is remotely controlled TV using remote controler, to realize the switching of channel.But it is this use remote controler
There are certain defects for the method for switching channels, for example, when the electricity of remote controler failure or remote controler exhausts, user
Remote controler switching channels can not be used；For remote controler far from user, user need to first take remote controler could switching channels.Therefore, it uses
Remote controler switching channels are because of the problem of being limited by remote controler itself, causing the inconvenience and inefficiency that user uses.
The above is only used to facilitate the understanding of the technical scheme, and is not represented and is recognized that the above is existing skill
Art.
Invention content
It is a primary object of the present invention to solve to use remote controler switching channels, the inconvenience for causing user to use, Yi Jixiao
The low technical problem of rate.
To achieve the above object, the present invention provides a kind of method of gesture switching channels, the side of the gesture switching channels
Method includes the following steps：
Images of gestures is acquired, the images of gestures of acquisition is converted into gray-scale map；
According to the gray-scale map, Laplacian Matrix is built, feature decomposition is carried out to the Laplacian Matrix of structure,
Obtain multiple feature vectors；
According to the element in each described eigenvector, the neat angle value of each described eigenvector is calculated, according to described
Two feature vectors of neat angle value minimum, generate objective matrix；
Clustering is carried out to the objective matrix of generation, according to clustering as a result, by the images of gestures
The gray value of pixel be set as 0 or 255, obtain the images of gestures of binaryzation；
Obtain the maximum comparability value of the images of gestures and each sample images of gestures to prestore of the binaryzation, switching frequency
Road is to the corresponding associated channel of sample images of gestures of the maximum comparability value.
Preferably, the element according in each described eigenvector, calculates the regularity of each described eigenvector
Value, according to two feature vectors of neat angle value minimum, the step of generation objective matrix, includes：
The average value of the value of element in each described eigenvector is obtained, the value of element and institute in described eigenvector are calculated
The summation for stating the mean square deviation of average value obtains the neat angle value；
Merge two feature vectors of the neat angle value minimum, generates objective matrix.
Preferably, the step of objective matrix progress clustering of described pair of generation includes：
Step A arbitrarily selects two elements as the first cluster centre and the second cluster centre in the objective matrix；
Step B is calculated separately in the objective matrix in each element and first cluster centre, second cluster
The distance value of the heart will be less than or equal to the distance value with second cluster centre with the distance value of first cluster centre
Elemental partition to first cluster, by with the distance value of first cluster centre be more than at a distance from second cluster centre
The Elemental partition of value to second cluster；
Step C calculates the average value of all elements in first cluster, regains the first cluster centre, calculate institute
The average value for stating all elements in the second cluster regains the second cluster centre；
Step B is continued to execute, until first cluster centre and second cluster centre and upper one that regain
When secondary first cluster centre regained is identical with second cluster centre, the result of clustering is exported；It is described
The result of clustering includes：Element in corresponding first cluster of first cluster centre that finally regains and
Element in corresponding second cluster of second cluster centre finally regained.
Preferably, the images of gestures for obtaining the binaryzation is similar to the maximum of each sample images of gestures to prestore
The step of property value includes：
By the pixel of the images of gestures of the binaryzation and the pixel of each sample images of gestures to prestore by
One comparison obtains the quantity of the images of gestures of binaryzation pixel identical with each sample images of gestures, obtains
The similarity of the images of gestures of the binaryzation and each sample images of gestures；
In the similarity from the images of gestures of the binaryzation with each sample images of gestures, obtain maximum
Similarity.
Preferably, the acquisition images of gestures, before the step of images of gestures of acquisition is converted to gray-scale map also
Including：
In the environment of solid color background, multiple original sample images of gestures are acquired；
Each original sample images of gestures of acquisition is subjected to binary conversion treatment, obtains the sample images of gestures；
It is associated with the sample images of gestures and channel, stores the sample images of gestures.
In addition, to achieve the above object, the present invention also provides a kind of device of gesture switching channels, the gesture switching frequency
The device in road includes：
The images of gestures of acquisition is converted to gray-scale map by acquisition module for acquiring images of gestures；
Module is built, for according to the gray-scale map, Laplacian Matrix being built, to the Laplacian Matrix of structure
Feature decomposition is carried out, multiple feature vectors are obtained；
Generation module, for according to the element in each described eigenvector, calculating the neat of each described eigenvector
Angle value generates objective matrix according to two feature vectors of the neat angle value minimum；
Cluster Analysis module, for carrying out clustering to the objective matrix of generation, according to clustering as a result,
The gray value of pixel in the images of gestures is set as 0 or 255, obtains the images of gestures of binaryzation；
Handover module, the maximum phase of images of gestures and each sample images of gestures to prestore for obtaining the binaryzation
Like property value, switching channels to the corresponding associated channel of sample images of gestures of the maximum comparability value.
Preferably, the generation module includes：
Acquiring unit, the average value for obtaining the value of element in each described eigenvector calculate described eigenvector
The summation of the value of middle element and the mean square deviation of the average value, obtains the neat angle value；
Combining unit, two feature vectors for merging the neat angle value minimum generate objective matrix.
Preferably, the Cluster Analysis module includes：
Initialization unit, for arbitrarily selecting two elements as the first cluster centre and second in the objective matrix
Cluster centre；
First analytic unit, for calculating separately each element and first cluster centre, institute in the objective matrix
The distance value for stating the second cluster centre will be less than or equal to and second cluster with the distance value of first cluster centre
The Elemental partition of the distance value at center gathers being more than with the distance value of first cluster centre with described second to the first cluster
The Elemental partition of the distance value at class center to second cluster；
Second analytic unit, the average value for calculating all elements in first cluster, regains the first cluster
Center calculates the average value of all elements in second cluster, regains the second cluster centre；
First analytic unit is additionally operable to calculate the average value of all elements in first cluster in the second analytic unit,
The first cluster centre is regained, the average value of all elements in second cluster is calculated, regains the second cluster centre
Later, it continues to execute and calculates separately in the objective matrix in each element and first cluster centre, second cluster
The distance value of the heart will be less than or equal to the distance value with second cluster centre with the distance value of first cluster centre
Elemental partition to first cluster, by with the distance value of first cluster centre be more than at a distance from second cluster centre
The step of Elemental partition of value to the second cluster, until first cluster centre and second cluster centre that regain
When first cluster centre and identical second cluster centre that are regained with the last time, the knot of clustering is exported
Fruit；The result of the clustering includes：In corresponding first cluster of first cluster centre finally regained
Element and corresponding second cluster of second cluster centre that finally regains in element；
Binarization unit, for according to clustering as a result, the gray value of the pixel in the images of gestures is set
It is 0 or 255, obtains the images of gestures of binaryzation.
Preferably, the handover module includes：
First acquisition unit, for by the pixel of the images of gestures of the binaryzation and each sample hand for prestoring
The pixel of gesture image compares one by one, obtains the images of gestures of binaryzation picture identical with each sample images of gestures
The quantity of vegetarian refreshments obtains the similarity of the images of gestures and each sample images of gestures of the binaryzation；
Second acquisition unit, the phase for images of gestures and each sample images of gestures from the binaryzation
Like in property value, maximum comparability value is obtained；
Switch unit, for switching channels to the corresponding associated channel of sample images of gestures of the maximum comparability value.
Preferably, the device of the gesture switching channels further includes：
Acquisition module, in the environment of solid color background, acquiring multiple original sample images of gestures；
Processing module, each original sample images of gestures for that will acquire carry out binary conversion treatment, obtain described
Sample images of gestures；
Relating module stores the sample images of gestures for being associated with the sample images of gestures and channel.
The present invention acquires images of gestures, and the images of gestures of acquisition is converted to gray-scale map；According to the gray-scale map, structure
Laplacian Matrix is built, feature decomposition is carried out to the Laplacian Matrix of structure, obtains multiple feature vectors；According to each
Element in described eigenvector calculates the neat angle value of each described eigenvector, according to the two of the neat angle value minimum
A feature vector generates objective matrix；Clustering is carried out to the objective matrix of generation, according to clustering as a result,
The gray value of pixel in the images of gestures is set as 0 or 255, obtains the images of gestures of binaryzation；Institute is obtained respectively
State the maximum comparability value of the images of gestures and each sample images of gestures to prestore of binaryzation, switching channels to the maximum phase
It is worth the corresponding associated channel of sample images of gestures like property.Compared to the prior art, user of the present invention is without the use of remote controler, only
It need to can be achieved with the switching of channel by gesture, it is convenient for users to use, and improve the efficiency of switching channels.
Description of the drawings
Fig. 1 is the flow diagram of the method first embodiment of gesture switching channels of the present invention；
Fig. 2 is the flow diagram of the method second embodiment of gesture switching channels of the present invention；
Fig. 3 is the regularity schematic diagram of feature vector；
Fig. 4 is the flow diagram of the method 3rd embodiment of gesture switching channels of the present invention；
Fig. 5 is the flow diagram of the method fourth embodiment of gesture switching channels of the present invention；
Fig. 6 is the flow diagram of the 5th embodiment of method of gesture switching channels of the present invention；
Fig. 7 is the high-level schematic functional block diagram of the device first embodiment of gesture switching channels of the present invention；
Fig. 8 is the high-level schematic functional block diagram of the device second embodiment of gesture switching channels of the present invention；
Fig. 9 is the high-level schematic functional block diagram of the device 3rd embodiment of gesture switching channels of the present invention；
Figure 10 is the high-level schematic functional block diagram of the device fourth embodiment of gesture switching channels of the present invention；
Figure 11 is the high-level schematic functional block diagram of the 5th embodiment of device of gesture switching channels of the present invention.
The embodiments will be further described with reference to the accompanying drawings for the realization, the function and the advantages of the object of the present invention.
Specific implementation mode
It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not intended to limit the present invention.
The present invention provides a kind of method of gesture switching channels.
Referring to Fig.1, Fig. 1 is the flow diagram of the method first embodiment of gesture switching channels of the present invention.
In the method first embodiment of gesture switching channels of the present invention, this approach includes the following steps：
Step S10 acquires images of gestures, the images of gestures of acquisition is converted to gray-scale map；
Television set acquires images of gestures by collecting device (such as camera, camera), which may be provided at
In television set, or it is placed outside television set, the present invention is not construed as limiting.The images of gestures of acquisition is converted to gray-scale map by television set,
For example, by the images of gestures of acquisition from RGB (Red, Green, Blue, red, green, blue) color space converts to YUV (gray scale, indigo plant
Coloration, the red coloration to cyan color components to yellow color-separated) chrominance space, obtain gray-scale map.Certainly also have much by hand
The method that gesture image is converted to gray-scale map, does not repeat one by one herein.
It should be noted that the images of gestures of acquisition is not limited to singlehanded images of gestures, can also be the gesture of both hands
Image.
Step S20 builds Laplacian Matrix according to the gray-scale map, is carried out to the Laplacian Matrix of structure
Feature decomposition obtains multiple feature vectors；
This case can build Laplacian Matrix by following technological means, be only that an example help is managed below
Solution, is not intended to limit the present invention.
The gray-scale map that step S10 is obtained can use following data set representations：
X={ x1,x2,...,xn}∈Rd
Wherein, xiIndicate that the arbitrary point in data set, i ∈ (1, n), n are data amount check, d indicates that data dimension, R represent
Entire set of real numbers.
First scale parameter σ is calculated with following formulai：
Wherein, xdIt is arbitrary point x in data level XiD Neighbor Points away from remaining each point select d=7, and n is data amount check；
Again similarity matrix A is calculated with following formula：
Aij=exp (- | | xi-xj||2/σiσj), i, j ∈ (i, n)
Wherein, AijIndicate the arbitrary element of similarity matrix A, σi、σjArbitrary point x in data set is indicated respectivelyiAnd xjIt is corresponding
Scale parameter, | | xi-xj| | indicate point xiAnd xjEuclidean distance.
Finally Laplacian Matrix is built with following formula：
L=D-1/2AD1/2
Wherein D is diagonal matrix, the arbitrary element D on diagonal lineijFor the elements A of the i-th row of similarity matrix AijIt is total
With DijShown in as the following formula：
Feature decomposition is carried out to the Laplacian Matrix of structure, obtains multiple feature vectors, for example, the La Pula of n rows n row
This matrix- eigenvector-decomposition is n characteristic value and n feature vector.Since the circular of feature decomposition is more conventional,
This is not repeated.
Step S30 calculates the neat angle value of each described eigenvector according to the element in each described eigenvector,
According to two feature vectors of the neat angle value minimum, objective matrix is generated；
According to the value of the element in each feature vector, the neat angle value of each described eigenvector is calculated, regularity is
Refer to the degree of scatter of the value of element in feature vector.
Neat angle value is smaller, shows that the value of element in feature vector is more close, the value of element is more close, then feature vector is got over
Neatly, then the information for the images of gestures that this feature vector includes is also more, therefore, two spies of the neat angle value minimum of television set rounding
Sign vector, generates objective matrix.
Step S40 carries out clustering, according to clustering as a result, by the hand to the objective matrix of generation
The gray value of pixel in gesture image is set as 0 or 255, obtains the images of gestures of binaryzation；
Television set carries out clustering to the objective matrix of generation, and the method for clustering includes：K mean cluster method, K
Central point clustering method, partition clustering method, hierarchy clustering method etc..
Television set is divided into two classes by clustering, by the element in objective matrix, before this two class is images of gestures
Scape and background.One type is indicated with 0, another kind of to be indicated with 255, and then realizes the binaryzation of images of gestures.At this point, gesture figure
The TOTAL DIFFERENT COLOR of the color and background of foreground (hand), realizes the segmentation of images of gestures as in.
Step S50 obtains the maximum comparability of the images of gestures and each sample images of gestures to prestore of the binaryzation
Value, switching channels to the corresponding associated channel of sample images of gestures of the maximum comparability value.
Television set obtains the maximum comparability value of the images of gestures and each sample images of gestures to prestore of binaryzation, obtains
The method of similarity includes：The method etc. of histogram matching, matrix decomposition, feature based point can also use pixel
The method compared one by one calculates similarity.The corresponding sample images of gestures of maximum comparability value is obtained, is connect by channel switching
Mouthful, switching channels to the associated channel of sample images of gestures.
The present embodiment acquires images of gestures, and the images of gestures of acquisition is converted to gray-scale map；According to being converted to gray scale
The images of gestures of figure builds Laplacian Matrix, carries out feature decomposition to the Laplacian Matrix of structure, obtains more
A feature vector；According to the element in each described eigenvector, the neat angle value of each described eigenvector is calculated, according to institute
Two feature vectors of neat angle value minimum are stated, objective matrix is generated；Clustering, root are carried out to the objective matrix of generation
According to clustering as a result, the element in objective matrix is divided into two classes, this two class is the foreground and background of images of gestures.Its
Middle one kind is indicated with 0, another kind of to be indicated with 255, and then realizes the binaryzation of images of gestures；The hand of the binaryzation is obtained respectively
The maximum comparability value of gesture image and each sample images of gestures to prestore, switching channels are corresponding to the maximum comparability value
The associated channel of sample images of gestures.Compared to the prior art, the present embodiment user is without the use of remote controler, only need to pass through gesture
The switching of channel is can be achieved with, it is convenient for users to use, and improve the efficiency of switching channels.On the other hand, the present embodiment
It can also make user far from television set, carry out the switching of channel.
It is the flow diagram of the method second embodiment of gesture switching channels of the present invention with reference to Fig. 2, Fig. 2.
In the method second embodiment of gesture switching channels of the present invention, the present embodiment and first embodiment difference lies in,
The present embodiment is on the basis of first embodiment, the element according in each described eigenvector, calculates each spy
The neat angle value for levying vector, according to two feature vectors of neat angle value minimum, the step of generation objective matrix, includes：
Step S31 obtains the average value of the value of element in each described eigenvector, calculates element in described eigenvector
Value and the average value mean square deviation summation, obtain the neat angle value；
Step S32 merges two feature vectors of the neat angle value minimum, generates objective matrix.
Television set calculates the average value A of the value of element in feature vector, under utilization according to the element in each feature vector
Face formula calculates the neat angle value of each feature vector：
Wherein, L is the neat angle value of feature vector, EiFor the value of element in feature vector, n is element in feature vector
Number, i ∈ (1, n).
As shown in following equation, the formula of the neat angle value of above-mentioned calculating feature vector can also carry out suitably as needed
Deformation, deformed formula is as follows：
Wherein, L is the neat angle value of feature vector, EiFor the value of element in feature vector, n is element in feature vector
Number, i ∈ (1, n).
After calculating the neat angle value of feature vector, obtain two feature vectors of neat angle value minimum, merge this two
A feature vector generates objective matrix VR.For example, two feature vectors are respectively A and B, merge two feature vectors, obtain mesh
Mark matrix V R=[A, B].
In the present embodiment, after the element in getting each feature vector, it can be generated according to obtained each feature vector
The regularity schematic diagram of feature vector, for can be visually seen two features of regularity minimum in feature vector according to the schematic diagram
Vector.With reference to Fig. 3, wherein a, b, c are respectively the regularity schematic diagram of three feature vectors, and abscissa X is indicated in images of gestures
The number of a line or a row pixel, ordinate Y indicate the value of the element in feature vector.Pass through the neat angle value of above-mentioned calculating
Formula be calculated the corresponding feature vector of a, b neat angle value it is minimum, also can intuitively see from Fig. 3, a, b are corresponding
In feature vector in the value of element feature vector corresponding compared to c element value, the value of element in the corresponding feature vector of a, b
It is more close, i.e., it is more neat, therefore, can intuitively it be seen by the regularity schematic diagram of feature vector in Fig. 3 whole in feature vector
Two feature vectors of neat angle value minimum.
It is the flow diagram of the method 3rd embodiment of gesture switching channels of the present invention with reference to Fig. 4, Fig. 4.
In the method 3rd embodiment of gesture switching channels of the present invention, the present embodiment and first embodiment and second embodiment
Difference lies in the present embodiment is on the basis of first embodiment and/or second embodiment, the target square of described pair of generation
Battle array carry out clustering the step of include：
Step S41 arbitrarily selects two elements as in the first cluster centre and the second cluster in the objective matrix
The heart；
Step S42 calculates separately each element and first cluster centre, second cluster in the objective matrix
The distance value at center, by with the distance value of first cluster centre be less than or equal to at a distance from second cluster centre
The Elemental partition of value to first cluster, by with the distance value of first cluster centre be more than with second cluster centre away from
Elemental partition from value is to the second cluster；
Step S43 calculates the average value of all elements in first cluster, regains the first cluster centre, calculate
The average value of all elements, regains the second cluster centre in second cluster；
Step S44 continues to execute step S42, until first cluster centre regained and second cluster
When center is with last first cluster centre regained and identical second cluster centre, clustering is exported
As a result；The result of the clustering includes：Corresponding first cluster of first cluster centre finally regained
In corresponding second cluster of element and second cluster centre that finally regains in element.
Television set is random in objective matrix or arbitrarily two elements is selected to be clustered as the first cluster centre and second
Center；Each element is calculated in objective matrix respectively at a distance from the first cluster centre and the second cluster centre using following formula
Value：
Wherein, OD is distance value, and E is the value of any one element in objective matrix, A be the first cluster centre value or
The value of second cluster centre.
It is the flow diagram of the method fourth embodiment of gesture switching channels of the present invention with reference to Fig. 5, Fig. 5.
In the method fourth embodiment of gesture switching channels of the present invention, the present embodiment and first embodiment, second embodiment,
Difference lies in the present embodiment is described on the basis of first embodiment, second embodiment, 3rd embodiment for 3rd embodiment
The step of maximum comparability value for obtaining the images of gestures and each sample images of gestures to prestore of the binaryzation includes：
Step S51, by the pixel of the images of gestures of the binaryzation and each sample images of gestures for prestoring
Pixel compares one by one, obtains the number of the images of gestures of binaryzation pixel identical with each sample images of gestures
Amount, the similarity of the images of gestures of the as described binaryzation and each sample images of gestures；
Step S52, in the similarity from the images of gestures of the binaryzation with each sample images of gestures,
Obtain maximum comparability value.
The pixel of the images of gestures of binaryzation is numbered television set in order, equally by each sample images of gestures
Pixel is numbered in order.The pixel of identical number in the images of gestures of binaryzation and each sample images of gestures is clicked through
Row comparison, identical pixel are labeled as " 1 ", and different pixels is labeled as " 0 ".Count the images of gestures of binaryzation and each
The quantity of the identical pixel of sample images of gestures, the as similitude of the images of gestures of binaryzation and each sample images of gestures
Value.
From all obtained similarities, maximum comparability value is obtained.
The present embodiment is by by the pixel of the images of gestures of binaryzation and the pixel of each sample images of gestures to prestore
Point is compared one by one, obtains the similarity of the images of gestures and each sample images of gestures of binaryzation.The side of the present embodiment
Method can accurately obtain maximum comparability value.
It is the flow diagram of the 5th embodiment of method of gesture switching channels of the present invention with reference to Fig. 6, Fig. 6.
In the 5th embodiment of method of gesture switching channels of the present invention, the present embodiment and first embodiment, second embodiment,
Difference lies in the present embodiment is in first embodiment, second embodiment, 3rd embodiment, for 3rd embodiment, fourth embodiment
On the basis of four embodiments, the acquisition images of gestures, before the step of images of gestures of acquisition is converted to gray-scale map
Including：
Step S60 acquires multiple original sample images of gestures in the environment of solid color background；
Each original sample images of gestures of acquisition is carried out binary conversion treatment, obtains the sample hand by step S70
Gesture image；
Step S80 is associated with the sample images of gestures and channel, stores the sample images of gestures.
Television set is acquired multiple original sample images of gestures, is carried on the back using solid color in the environment of solid color background
Scape is to keep subsequent image processing process simpler.Due to being acquired in the environment of solid color background, it will be original
Sample images of gestures from RGB (Red, Green, Blue, red, green, blue) color space convert be HSV (Hue, Saturation,
Value, hue, saturation, intensity) chrominance space, binary-state threshold can be set, tone value in original sample images of gestures is big
255 are set as in or equal to the gray value of pixel of binary-state threshold, tone value in original sample images of gestures is less than two-value
The gray value for changing the pixel of threshold value is set as 0, it is, of course, also possible to which tone value in original sample images of gestures is greater than or equal to two
The gray value of the pixel of value threshold value is set as 0, and tone value in original sample images of gestures is less than to the pixel of binary-state threshold
The gray value of point is set as 255.At this point, the effect original sample images of gestures of black and white will be presented as sample images of gestures.
Sample images of gestures is associated with channel, i.e., by the number or name of sample images of gestures and channel, frequency
The one-to-one relationship of the generations such as rate.Sample images of gestures is stored in case follow-up use.
The present embodiment acquires original sample images of gestures under solid color background, simplifies to original sample images of gestures
Processing procedure, improve efficiency.
The present invention further provides a kind of devices of gesture switching channels.
It is the high-level schematic functional block diagram of the device first embodiment of gesture switching channels of the present invention with reference to Fig. 7, Fig. 7.
In the device first embodiment of gesture switching channels of the present invention, which includes：
The images of gestures of acquisition is converted to gray-scale map by acquisition module 10 for acquiring images of gestures；
Acquisition module 10 acquires images of gestures by collecting device (such as camera, camera), which can set
It sets in a television set, or is placed outside television set, the present invention is not construed as limiting.Acquisition module 10 is converted to the images of gestures of acquisition
Gray-scale map, for example, by the images of gestures of acquisition from RGB (Red, Green, Blue, red, green, blue) color space converts to YUV
(gray scale, blue coloration, the red coloration to cyan color components to yellow color-separated) chrominance space, obtains gray-scale map.Certainly also have
The method that images of gestures is much converted to gray-scale map, does not repeat one by one herein.
It should be noted that the images of gestures of acquisition is not limited to singlehanded images of gestures, can also be the gesture of both hands
Image.
Module 20 is built, for according to the gray-scale map, Laplacian Matrix being built, to Laplce's square of structure
Battle array carries out feature decomposition, obtains multiple feature vectors；
This case can build Laplacian Matrix by following technological means, be only that an example help is managed below
Solution, is not intended to limit the present invention.
The gray-scale map that acquisition module 10 obtains can use following data set representations：
X={ x1,x2,...,xn}∈Rd
Wherein, xiIndicate that the arbitrary point in data set, i ∈ (1, n), n are data amount check, d indicates that data dimension, R represent
Entire set of real numbers.
First scale parameter σ is calculated with following formulai：
Wherein, xdIt is arbitrary point x in data level XiD Neighbor Points away from remaining each point select d=7, and n is data amount check；
Again similarity matrix A is calculated with following formula：
Aij=exp (- | | xi-xj||2/σiσj), i, j ∈ (i, n)
Wherein, AijIndicate the arbitrary element of similarity matrix A, σi、σjArbitrary point x in data set is indicated respectivelyiAnd xjIt is corresponding
Scale parameter, | | xi-xj| | indicate point xiAnd xjEuclidean distance.
Finally Laplacian Matrix is built with following formula：
L=D-1/2AD1/2
Wherein D is diagonal matrix, the arbitrary element Di on diagonal linejFor the elements A of the i-th row of similarity matrix AijIt is total
With, as the following formula shown in：
Feature decomposition is carried out to the Laplacian Matrix of structure, obtains multiple feature vectors, for example, the La Pula of n rows n row
This matrix- eigenvector-decomposition is n characteristic value and n feature vector.Since the circular of feature decomposition is more conventional,
This is not repeated.
Generation module 30, for according to the element in each described eigenvector, calculating the whole of each described eigenvector
Neat angle value generates objective matrix according to two feature vectors of the neat angle value minimum；
According to the value of the element in each feature vector, the neat angle value of each described eigenvector is calculated using formula,
Regularity refers to the degree of scatter of the value of element in feature vector.
Neat angle value is smaller, shows that the value of element in feature vector is more close, the value of element is more close, then feature vector is got over
Neatly, then the information for the images of gestures that this feature vector includes is also more, therefore, the two of the neat angle value minimum of 30 rounding of generation module
A feature vector generates objective matrix.
Cluster Analysis module 40 carries out clustering, according to the knot of clustering for the objective matrix to generation
The gray value of pixel in the images of gestures is set as 0 or 255, obtains the images of gestures of binaryzation by fruit；
Cluster Analysis module 40 is divided into two classes by clustering, by the element in objective matrix, this two class is gesture
The foreground and background of image.One type is indicated with 0, another kind of to be indicated with 255, and then realizes the binaryzation of images of gestures.This
When, the TOTAL DIFFERENT COLOR of the color and background of foreground (hand), realizes the segmentation of images of gestures in images of gestures.
Handover module 50, the maximum of images of gestures and each sample images of gestures to prestore for obtaining the binaryzation
Similarity, switching channels to the corresponding associated channel of sample images of gestures of the maximum comparability value.
Handover module 50 obtains the maximum comparability value of the images of gestures and each sample images of gestures to prestore of binaryzation,
Obtain similarity method include：The method etc. of histogram matching, matrix decomposition, feature based point can also use picture
The method that vegetarian refreshments compares one by one calculates similarity.The corresponding sample images of gestures of maximum comparability value is obtained, is cut by channel
Alias, switching channels to the associated channel of sample images of gestures.
The present embodiment acquisition module 10 acquires images of gestures, and the images of gestures of acquisition is converted to gray-scale map；Structure
Module 20 builds Laplacian Matrix, to Laplce's square of structure according to the images of gestures for being converted to gray-scale map
Battle array carries out feature decomposition, obtains multiple feature vectors；Generation module 30 calculates each according to the element in each described eigenvector
The neat angle value of a described eigenvector generates objective matrix according to two feature vectors of the neat angle value minimum；Cluster
Analysis module 40 carries out clustering to the objective matrix of generation, according to clustering as a result, by objective matrix
Element is divided into two classes, this two class is the foreground and background of images of gestures.One type is indicated with 0, another kind of to be indicated with 255,
And then realize the binaryzation of images of gestures；Handover module 50 obtains the images of gestures of the binaryzation and each sample to prestore respectively
The maximum comparability value of this images of gestures, switching channels to the corresponding associated frequency of sample images of gestures of the maximum comparability value
Road.Compared to the prior art, the present embodiment user is without the use of remote controler, and the switching of channel need to be only can be achieved with by gesture,
It is convenient for users to use, and improve the efficiency of switching channels.On the other hand, the present embodiment can also make user far from electricity
In the case of depending on machine, the switching of channel is carried out.
It is the high-level schematic functional block diagram of the device second embodiment of gesture switching channels of the present invention with reference to Fig. 8, Fig. 8.
In the device second embodiment of gesture switching channels of the present invention, the present embodiment and first embodiment difference lies in,
On the basis of first embodiment, the generation module 30 includes the present embodiment：
Acquiring unit 31, the average value for obtaining the value of element in each described eigenvector, calculate the feature to
The summation of the value of element and the mean square deviation of the average value, obtains the neat angle value in amount；
Combining unit 32, two feature vectors for merging the neat angle value minimum generate objective matrix.
Acquiring unit 31 calculates the average value A of the value of element in feature vector, profit according to the element in each feature vector
The neat angle value of each feature vector is calculated with following formula：
Wherein, L is the neat angle value of feature vector, EiFor the value of element in feature vector, n is element in feature vector
Number, i ∈ (1, n).
As shown in following equation, the formula of the neat angle value of above-mentioned calculating feature vector can also carry out suitably as needed
Deformation, but be within the scope of the invention.
Wherein, L is the neat angle value of feature vector, EiFor the value of element in feature vector, n is element in feature vector
Number, i ∈ (1, n).
After calculating the neat angle value of feature vector, combining unit 32 obtain two features of neat angle value minimum to
Amount merges the two feature vectors, generates objective matrix VR.For example, two feature vectors are respectively A and B, merge two features
Vector obtains objective matrix VR=[A, B].
It is the high-level schematic functional block diagram of the device 3rd embodiment of gesture switching channels of the present invention with reference to Fig. 9, Fig. 9.
In the device 3rd embodiment of gesture switching channels of the present invention, the present embodiment and first embodiment and second embodiment
Difference lies on the basis of first embodiment and/or second embodiment, the Cluster Analysis module 40 wraps the present embodiment
It includes：
Initialization unit 41, for arbitrarily selecting two elements as the first cluster centre and in the objective matrix
Two cluster centres；
First analytic unit 42, for calculate separately each element and first cluster centre in the objective matrix,
The distance value of second cluster centre gathers being less than or equal to the distance value of first cluster centre with described second
The Elemental partition of the distance value at class center will be more than and described second to the first cluster with the distance value of first cluster centre
The Elemental partition of the distance value of cluster centre to second cluster；
Second analytic unit 43, the average value for calculating all elements in first cluster, it is poly- to regain first
Class center calculates the average value of all elements in second cluster, regains the second cluster centre；
First analytic unit 42 is additionally operable to all elements in second analytic unit 43 calculates first cluster and is averaged
Value regains the first cluster centre, calculates the average value of all elements in second cluster, regains in the second cluster
After the heart, continues to execute and calculate separately each element and first cluster centre, second cluster in the objective matrix
The distance value at center, by with the distance value of first cluster centre be less than or equal to at a distance from second cluster centre
The Elemental partition of value to first cluster, by with the distance value of first cluster centre be more than with second cluster centre away from
The step of Elemental partition from value to the second cluster, until in first cluster centre and second cluster that regain
When the heart is with last first cluster centre regained and identical second cluster centre, the knot of clustering is exported
Fruit；The result of the clustering includes：In corresponding first cluster of first cluster centre finally regained
Element and corresponding second cluster of second cluster centre that finally regains in element；
Binarization unit 44 is used for according to clustering as a result, by the gray value of the pixel in the images of gestures
It is set as 0 or 255, obtains the images of gestures of binaryzation.
Initialization unit 41 is random in objective matrix or arbitrary selects two elements as the first cluster centre and the
Two cluster centres；First analytic unit 42 using following formula calculate objective matrix in each element respectively with the first cluster centre
With the distance value of the second cluster centre：
Wherein, OD is distance value, and E is the value of any one element in objective matrix, A be the first cluster centre value or
The value of second cluster centre.
0, Figure 10 is the handoff functionality module signal of the device fourth embodiment of gesture switching channels of the present invention referring to Fig.1
Figure.
In the device fourth embodiment of gesture switching channels of the present invention, the present embodiment and first embodiment, second embodiment,
Difference lies in the present embodiment is described on the basis of first embodiment, second embodiment, 3rd embodiment for 3rd embodiment
Handover module 50 includes：
First acquisition unit 51, for by the pixel of the images of gestures of the binaryzation and each sample for prestoring
The pixel of images of gestures compares one by one, and the images of gestures for obtaining the binaryzation is identical with each sample images of gestures
The quantity of pixel obtains the similarity of the images of gestures and each sample images of gestures of the binaryzation；
Second acquisition unit 52 is used for from described in the images of gestures of the binaryzation and each sample images of gestures
In similarity, maximum comparability value is obtained；
Switch unit 53, for switching channels to the corresponding associated frequency of sample images of gestures of the maximum comparability value
Road.
The pixel of the images of gestures of binaryzation is numbered first acquisition unit 51 in order, equally by each sample hand
The pixel of gesture image is numbered in order.By identical number in the images of gestures of binaryzation and each sample images of gestures
Pixel is compared, and identical pixel is labeled as " 1 ", and different pixels is labeled as " 0 ".Count the gesture figure of binaryzation
The quantity of picture and the identical pixel of each sample images of gestures, the as images of gestures of binaryzation and each sample images of gestures
Similarity.
Second acquisition unit 52 obtains maximum comparability value from all obtained similarities.
The present embodiment is by by the pixel of the images of gestures of binaryzation and the pixel of each sample images of gestures to prestore
Point is compared one by one, obtains the similarity of the images of gestures and each sample images of gestures of binaryzation.
1, Figure 11 is the high-level schematic functional block diagram of the 5th embodiment of device of gesture switching channels of the present invention referring to Fig.1.
In the 5th embodiment of device of gesture switching channels of the present invention, the present embodiment and first embodiment, second embodiment,
Difference lies in the present embodiment is in first embodiment, second embodiment, 3rd embodiment, for 3rd embodiment, fourth embodiment
On the basis of four embodiments, the device of the gesture switching channels further includes：
Sample collection module 60, in the environment of solid color background, acquiring multiple original sample images of gestures；
Processing module 70, each original sample images of gestures for that will acquire carry out binary conversion treatment, obtain institute
State sample images of gestures；
Relating module 80 stores the sample images of gestures for being associated with the sample images of gestures and channel.
Sample collection module 60 acquires multiple original sample images of gestures, using list in the environment of solid color background
One color background is to keep subsequent image processing process simpler.Due to being acquired in the environment of solid color background, because
This, original sample images of gestures is by processing module 70 from RGB (Red, Green, Blue, red, green, blue) color space convert
HSV (Hue, Saturation, Value, hue, saturation, intensity) chrominance space, can set binary-state threshold, by original sample
Tone value is set as 255 more than or equal to the gray value of the pixel of binary-state threshold in this images of gestures, by original sample gesture
Tone value is set as 0 less than the gray value of the pixel of binary-state threshold in image, it is, of course, also possible to by original sample images of gestures
The gray value that middle tone value is greater than or equal to the pixel of binary-state threshold is set as 0, by tone value in original sample images of gestures
Gray value less than the pixel of binary-state threshold is set as 255.At this point, the effect original sample images of gestures that black and white is presented is made
For sample images of gestures.
Sample images of gestures is associated by relating module 80 with channel, i.e., by the number of sample images of gestures and channel or
The one-to-one relationship of the generations such as person's name, frequency.Sample images of gestures is stored in case follow-up use.
The present embodiment acquires original sample images of gestures under solid color background, simplifies to original sample images of gestures
Processing procedure, improve efficiency.
It these are only the preferred embodiment of the present invention, be not intended to limit the scope of the invention, it is every to utilize this hair
Equivalent structure or equivalent flow shift made by bright specification and accompanying drawing content is applied directly or indirectly in other relevant skills
Art field, is included within the scope of the present invention.
Claims (8)
1. a kind of method of gesture switching channels, which is characterized in that the method for the gesture switching channels includes the following steps：
Images of gestures is acquired, the images of gestures of acquisition is converted into gray-scale map；
According to the gray-scale map, Laplacian Matrix is built, feature decomposition is carried out to the Laplacian Matrix of structure, is obtained
Multiple feature vectors；
The average value of the value of element in each described eigenvector is obtained, the value of element in described eigenvector is calculated and is put down with described
The summation of the mean square deviation of mean value obtains the neat angle value of described eigenvector；
Merge two feature vectors of the neat angle value minimum, generates objective matrix；
Arbitrarily select two elements as the first cluster centre and the second cluster centre in the objective matrix, with described first
Cluster centre, the second cluster centre carry out clustering to the objective matrix of generation, by element in the objective matrix point
At foreground elements and background element；
According to the foreground elements and background element, the gray value of the pixel in the images of gestures is set as 0 or 255,
Obtain the images of gestures of binaryzation；
The maximum comparability value of the images of gestures and each sample images of gestures to prestore of the binaryzation is obtained, switching channels are extremely
The corresponding associated channel of sample images of gestures of the maximum comparability value.
2. the method for gesture switching channels as described in claim 1, which is characterized in that arbitrarily selected in the objective matrix
Two elements are as the first cluster centre and the second cluster centre, with first cluster centre, the second cluster centre to generating
The objective matrix carry out clustering, element in the objective matrix is divided into foreground elements and background element step packet
It includes：
Step A arbitrarily selects two elements as the first cluster centre and the second cluster centre in the objective matrix；
Step B calculates separately each element and first cluster centre in the objective matrix, second cluster centre
Distance value will be less than or equal to the member with the distance value of second cluster centre with the distance value of first cluster centre
Element distribution will be more than and the distance value of second cluster centre to the first cluster with the distance value of first cluster centre
Elemental partition to second cluster；
Step C calculates the average value of all elements in first cluster, regains the first cluster centre, calculates described the
The average value of all elements, regains the second cluster centre in two clusters；
Step B is continued to execute, until first cluster centre and second cluster centre that regain and last time weight
When first cluster centre newly obtained is identical with second cluster centre, the result of clustering is exported；The cluster
The result of analysis includes：Element in corresponding first cluster of first cluster centre that finally regains and final
Element in corresponding second cluster of second cluster centre regained；
Step D, according to element in corresponding first cluster of first cluster centre finally regained and final
Element in corresponding second cluster of second cluster centre regained divides the element in the objective matrix
For foreground elements and background element.
3. the method for gesture switching channels as described in claim 1, which is characterized in that the gesture for obtaining the binaryzation
The step of image and the maximum comparability value of each sample images of gestures to prestore includes：
The pixel of the pixel of the images of gestures of the binaryzation and each sample images of gestures to prestore is right one by one
Than obtaining the quantity of the images of gestures of binaryzation pixel identical with each sample images of gestures, obtaining described
The similarity of the images of gestures of binaryzation and each sample images of gestures；
In the similarity from the images of gestures of the binaryzation with each sample images of gestures, obtain maximum similar
Property value.
4. the method for gesture switching channels as described in claim 1, which is characterized in that the acquisition images of gestures will acquire
Images of gestures the step of being converted to gray-scale map before further include：
In the environment of solid color background, multiple original sample images of gestures are acquired；
Each original sample images of gestures of acquisition is subjected to binary conversion treatment, obtains the sample images of gestures；
It is associated with the sample images of gestures and channel, stores the sample images of gestures.
5. a kind of device of gesture switching channels, which is characterized in that the device of the gesture switching channels includes：
The images of gestures of acquisition is converted to gray-scale map by acquisition module for acquiring images of gestures；
Module is built, for according to the gray-scale map, building Laplacian Matrix, the Laplacian Matrix of structure is carried out
Feature decomposition obtains multiple feature vectors；
Generation module, the average value for obtaining the value of element in each described eigenvector calculate member in described eigenvector
The summation of the value of element and the mean square deviation of the average value obtains the neat angle value of described eigenvector, merges the neat angle value
Two minimum feature vectors generate objective matrix；
Cluster Analysis module, for arbitrarily selecting two elements poly- as the first cluster centre and second in the objective matrix
Class center carries out clustering with first cluster centre, the second cluster centre to the objective matrix of generation, will be described
Element is divided into foreground elements and background element in objective matrix, according to the foreground elements and background element, by the gesture figure
The gray value of pixel as in is set as 0 or 255, obtains the images of gestures of binaryzation；
Handover module, the maximum comparability of images of gestures and each sample images of gestures to prestore for obtaining the binaryzation
Value, switching channels to the corresponding associated channel of sample images of gestures of the maximum comparability value.
6. the device of gesture switching channels as claimed in claim 5, which is characterized in that the Cluster Analysis module includes：
Initialization unit, for arbitrarily selecting two elements to be clustered as the first cluster centre and second in the objective matrix
Center；
First analytic unit, for calculating separately each element and first cluster centre in the objective matrix, described the
The distance value of two cluster centres will be less than or equal to and second cluster centre with the distance value of first cluster centre
Distance value Elemental partition to first cluster, will with the distance value of first cluster centre be more than with it is described second cluster in
The Elemental partition of the distance value of the heart to second cluster；
Second analytic unit：Average value for calculating all elements in first cluster, regains the first cluster centre,
The average value for calculating all elements in second cluster, regains the second cluster centre；
First analytic unit is additionally operable to calculate the average value of all elements in first cluster in the second analytic unit, again
The first cluster centre is obtained, the average value of all elements in second cluster is calculated, after regaining the second cluster centre,
Continue to execute calculate separately each element and first cluster centre in the objective matrix, second cluster centre away from
From value, the element with the distance value of second cluster centre will be less than or equal to the distance value of first cluster centre
Distribution will be more than the member with the distance value of second cluster centre to the first cluster with the distance value of first cluster centre
The step of element distribution to the second cluster, until first cluster centre and second cluster centre and upper one that regain
When secondary first cluster centre regained is identical with second cluster centre, the result of clustering is exported；It is described
The result of clustering includes：Element in corresponding first cluster of first cluster centre that finally regains and
Element in corresponding second cluster of second cluster centre finally regained；
Binarization unit, for according to the member in corresponding first cluster of first cluster centre finally regained
Element in element second cluster corresponding with second cluster centre finally regained will be in the objective matrix
Element be divided into foreground elements and background element, the gray value of the pixel in the images of gestures is set as 0 or 255, is obtained
To the images of gestures of binaryzation.
7. the device of gesture switching channels as claimed in claim 5, which is characterized in that the handover module includes：
First acquisition unit, for by the pixel of the images of gestures of the binaryzation and each sample gesture figure for prestoring
The pixel of picture compares one by one, obtains the images of gestures of binaryzation pixel identical with each sample images of gestures
Quantity, obtain the similarity of the images of gestures and each sample images of gestures of the binaryzation；
Second acquisition unit, the similitude for images of gestures and each sample images of gestures from the binaryzation
In value, maximum comparability value is obtained；
Switch unit, for switching channels to the corresponding associated channel of sample images of gestures of the maximum comparability value.
8. the device of gesture switching channels as claimed in claim 5, which is characterized in that the device of the gesture switching channels is also
Including：
Sample collection module, in the environment of solid color background, acquiring multiple original sample images of gestures；
Processing module, each original sample images of gestures for that will acquire carry out binary conversion treatment, obtain the sample
Images of gestures；
Relating module stores the sample images of gestures for being associated with the sample images of gestures and channel.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410352753.9A CN105320443B (en) | 2014-07-23 | 2014-07-23 | The method and device of gesture switching channels |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410352753.9A CN105320443B (en) | 2014-07-23 | 2014-07-23 | The method and device of gesture switching channels |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105320443A CN105320443A (en) | 2016-02-10 |
| CN105320443B true CN105320443B (en) | 2018-09-04 |
Family
ID=55247911
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410352753.9A Active CN105320443B (en) | 2014-07-23 | 2014-07-23 | The method and device of gesture switching channels |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105320443B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113552940A (en) * | 2021-06-23 | 2021-10-26 | 九牧厨卫股份有限公司 | Intelligent closestool and control method and device thereof |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102200830A (en) * | 2010-03-25 | 2011-09-28 | 夏普株式会社 | Non-contact control system and control method based on static gesture recognition |
| CN102265250A (en) * | 2008-12-31 | 2011-11-30 | 微软公司 | Control function gestures |
| CN202907113U (en) * | 2012-06-14 | 2013-04-24 | 深圳市同洲电子股份有限公司 | A TV set controlled by gesture identification |
| CN103179359A (en) * | 2011-12-21 | 2013-06-26 | 北京新岸线移动多媒体技术有限公司 | Method and device for controlling video terminal and video terminal |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5845002B2 (en) * | 2011-06-07 | 2016-01-20 | ソニー株式会社 | Image processing apparatus and method, and program |
-
2014
- 2014-07-23 CN CN201410352753.9A patent/CN105320443B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102265250A (en) * | 2008-12-31 | 2011-11-30 | 微软公司 | Control function gestures |
| CN102200830A (en) * | 2010-03-25 | 2011-09-28 | 夏普株式会社 | Non-contact control system and control method based on static gesture recognition |
| CN103179359A (en) * | 2011-12-21 | 2013-06-26 | 北京新岸线移动多媒体技术有限公司 | Method and device for controlling video terminal and video terminal |
| CN202907113U (en) * | 2012-06-14 | 2013-04-24 | 深圳市同洲电子股份有限公司 | A TV set controlled by gesture identification |
Non-Patent Citations (2)
| Title |
|---|
| 基于谱聚类的图像分割方法研究;由里;《中国优秀硕士学位论文全文数据库 信息科技辑》;20120715;I138-1901 * |
| 谱聚类中选取特征向量的动态选择性集成方法;王兴良等;《模式识别与人工智能》;20140531;第27卷(第5期);第452-462页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105320443A (en) | 2016-02-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Crowd counting with deep structured scale integration network | |
| Yu et al. | Cascading convolutional color constancy | |
| CN107122826B (en) | Processing method and system for convolutional neural network, and storage medium | |
| US9741137B2 (en) | Image-based color palette generation | |
| Wan et al. | A novel framework for optimal RGB to grayscale image conversion | |
| US20160104303A1 (en) | Image-based color palette generation | |
| CN107204034B (en) | A kind of image processing method and terminal | |
| CN112328345B (en) | Method, apparatus, electronic device and readable storage medium for determining theme colors | |
| KR102634186B1 (en) | Method for verifying the identity of a user by identifying an object by identifying an object in an image containing the user's biometric characteristics and isolating the part of the image containing the biometric characteristics from other parts of the image within an image that has a biometric characteristic of the user and separating a portion of the image comprising the biometric characteristic from other portions of the image) | |
| WO2017059620A1 (en) | Image processing method and device for led display screen | |
| CN106530361A (en) | Color correction method for color face image | |
| CN111783803A (en) | Image processing method and device for realizing privacy protection | |
| Lee et al. | Depth-guided adaptive contrast enhancement using 2D histograms | |
| JP5615344B2 (en) | Method and apparatus for extracting color features | |
| KR101334794B1 (en) | Flower recognition method and apparatus using feature extraction | |
| CN105320443B (en) | The method and device of gesture switching channels | |
| JP2019062482A (en) | Image processing apparatus, image processing system, and program | |
| JP2013164832A (en) | Object recognition device and method | |
| CN113762058B (en) | Video synthesis method, device, computer equipment and storage medium | |
| RU2004117830A (en) | METHOD FOR CONVERTING IMAGES | |
| CN108353133B (en) | Apparatus and method for reducing exposure time sets for high dynamic range video/imaging | |
| Buzzelli et al. | Consensus-driven illuminant estimation with GANs | |
| CN101561316A (en) | On-line test visual data processing system based on region of interest (ROI) | |
| Uruma et al. | Image colorization algorithm using series approximated sparse function | |
| JP2021197599A5 (en) |
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 |