CN106651879A - Method and system for extracting nail image - Google Patents
Method and system for extracting nail image Download PDFInfo
- Publication number
- CN106651879A CN106651879A CN201611215028.2A CN201611215028A CN106651879A CN 106651879 A CN106651879 A CN 106651879A CN 201611215028 A CN201611215028 A CN 201611215028A CN 106651879 A CN106651879 A CN 106651879A
- Authority
- CN
- China
- Prior art keywords
- image
- region
- hand
- pin
- nail
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30196—Human being; Person
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a method and system for extracting a nail image. The method comprises the steps of: by machine vision equipment, carrying out shooting on a hand or foot to be analyzed so as to acquire an image of the hand or foot in a designated gesture; according to preset human skin color information, establishing a classification model of a background of the image and the hand or foot, and according to the classification model, extracting an image region of the hand or foot in the image; and according to the designated gesture, extracting a first nail region in the image region of the hand or foot, and extracting the nail image in the first nail region by adopting an image segmentation method. According to the method and system disclosed by the invention, the nail image is extracted according to outline feature information of human skin and nails under color features and in the specific gesture, so that both accuracy and robustness are greatly improved.
Description
Technical field
The present invention relates to technical field of image processing, the extracting method and system of more particularly to a kind of thumbnail image.
Background technology
Image Segmentation Technology application in business activity instantly is extremely wide, including medicine CT image analysis, mobile phone are clapped
According to background blurring etc..Because image is ever-changing, the characteristics of image in each field is different, thus different field is accomplished by phase
The image Segmentation Technology answered is processed.
In medical field, doctor can carry out the prevention of disease, for example, nail semilunar phase by the thumbnail image for extracting
Diminish than usual, illustrate that current metabolism goes wrong, grey nail represents that people has fungus-caused infectious disease etc..Utilize
Image Segmentation Technology automatically extracts out fingernail region and the state to history nail is tracked, it will play certain pre- to disease
Alert effect, meets Future Internet medical trend.On the other hand, carry out making automatic U.S. using the fingernail region image for extracting
First machine will produce the property changed to nail salon to be affected.
The achievement in research for extracting fingernail region currently with image segmentation is less, Kumuda, N.S., and M.S.Dinesh
["Human fingernail segmentation."Emerging Research in Electronics,Computer
Science and Technology(ICERECT),2015 International Conference on.IEEE,2015.]
The method of proposition mainly realizes the nail based on color and extracts, and the segmentation of nail regional, and the method only uses face
Color information, its limitation is larger, and is not suitable for the shallower ethnic group of the colour of skin.Sun,Yu,et.al.["Estimation of
fingertip force direction with computer vision."IEEE Transactions on Robotics
25.6(2009):1356-1369.] used in fingernail region extracting method only used the marginal information of nail, robustness is not
Height, and the method needs carry out in advance corresponding image calibration collection, practicality to the thumbnail image of each user
It is poor.
Thus prior art could be improved and improve.
The content of the invention
The technical problem to be solved in the present invention is, for the deficiencies in the prior art, there is provided a kind of extraction of thumbnail image
Method and system, its contour feature information extraction nail according to human skin and nail under color characteristic and given pose
Image so that accuracy and robustness are improved.
In order to solve above-mentioned technical problem, the technical solution adopted in the present invention is as follows:
A kind of extracting method of thumbnail image, it includes:
Hand or pin to be analyzed are shot by machine vision equipment, obtains the image for specifying the hand or pin of posture;
The background of described image and hand or the disaggregated model of pin are set up according to default human skin colouring information, and according to
The disaggregated model extracts the image-region of the hand or pin in described image;
The hand or the first fingernail region in pin image-region are extracted according to the specified posture, and adopts image segmentation
Method extracts thumbnail image in first fingernail region.
The extracting method of the thumbnail image, wherein, the Background color information of the identification described image, and according to described
Background color and human skin colouring information set up the background of image and hand or pin disaggregated model are specifically included:
By the color space conversion of described image to YCbCr space, and obtain value and Cr of each pixel in Cb passages
The value of passage;
Background for distinguishing image is determined in the value of Cb passages and the value of Cr passages according to the human skin color
With hand or the class condition of pin;
The image-region of the hand or pin is extracted in described image according to the class condition.
The extracting method of the thumbnail image, wherein, the class condition is:
Wherein, Pb be pixel P Cb passages value, Pr be pixel P Cr passages value, b0, r0, θ, a and b are constant.
The extracting method of the thumbnail image, wherein, it is described that the figure is set up according to default human skin colouring information
The background of picture and hand or the disaggregated model of pin, and the image of the hand or pin is extracted in described image according to the disaggregated model
Region specifically includes:
Obtain the nominal contour and default mankind's skin of the image-region of the corresponding standard hand of the specified posture or pin
Skin colouring information;
The background and hand or the disaggregated model of pin of described image are set up according to the skin color and the nominal contour;
The image-region of the hand or pin is extracted in described image according to the disaggregated model.
The extracting method of the thumbnail image, wherein, it is described according to the skin color information and the nominal contour
The background and the disaggregated model of hand or pin for setting up described image is specifically included:
The Background color information of identification described image, and according to the Background color information and skin color information to institute
The color for stating image is clustered;
Calculated using comparative silhouette method and cluster current outline and institute that all sort merges for obtaining form image-region
State the affinity score of the corresponding nominal contour of specified posture;Obtain affinity score highest classification, and by sort merge shape
Into image-region be set in described image the image-region of hand or pin and extract.
The extracting method of the thumbnail image, wherein, it is described to be extracted in described image region according to the specified posture
First fingernail region, and thumbnail image is extracted in first fingernail region using image partition method specifically include:
Described image region is modified so that described image region mark corresponding with the default specified posture
The image-region alignment of quasi- hand or pin;
First nail in described image region is determined according to the fingernail region of the image-region of the standard hand or pin setting
Region, wherein, first fingernail region includes some first nail subregions;
Nail subregion, some subregion structures are extracted in each first nail subregion using image partition method
Into the thumbnail image.
The extracting method of the thumbnail image, wherein, described image dividing method is based on nail colour and skin color
The image segmentation algorithm of difference, based on the curve-fitting method of image edge information, based on nail colour information, image border believe
It is a kind of in the energy minimization method of standard profile information under breath and specified posture.
A kind of extraction system of thumbnail image, it includes:
Taking module, for being shot to hand or pin to be analyzed by machine vision equipment, is obtained and specifies posture
The image of hand or pin;
First extraction module, for setting up the background and hand or pin of described image according to default human skin colouring information
Disaggregated model, and the image-region of the hand or pin is extracted in described image according to the disaggregated model;
Second extraction module, for extracting the hand or the first nail area in pin image-region according to the specified posture
Domain, and thumbnail image is extracted in first fingernail region using image partition method.
The extraction system of the thumbnail image, wherein, first extraction module is specifically included:
Acquiring unit, for obtain the image-region of the corresponding standard hand of the specified posture or pin nominal contour and
Default human skin colouring information;
Set up unit, for set up according to the skin color and the nominal contour background of described image and hand or
The disaggregated model of pin;
First extraction unit, for extracting the image district of the hand or pin in described image according to the disaggregated model
Domain.
The extraction system of the thumbnail image, wherein, second extraction module is specifically included:
Amending unit, for being modified to described image region so that described image region is described specified with default
The image-region alignment of the corresponding standard hand of posture or pin;
Determining unit, the fingernail region for being set according to the image-region of the standard hand or pin determines described image area
First fingernail region in domain, wherein, first fingernail region includes some first nail subregions;
Second extraction unit, for nail sub-district to be extracted in each first nail subregion using image partition method
Domain, some subregions constitute the thumbnail image.
A kind of extraction element of thumbnail image, it includes the extraction system of as above arbitrary described thumbnail image.
The extraction element of the thumbnail image, it also includes placing device, and the placing device is included with accommodating chamber
Housing, the bottom of the accommodating chamber is provided with holding plane, and the housing one side relative with the holding plane is provided with use
In external equipment shooting image first is open, and the housing is provided with second stretched into for palm and is open, and described first opens
Plane residing for mouthful intersects with the residing plane of the described second opening.
The extraction element of the thumbnail image, wherein, the LED for light filling is provided with the housing.
Beneficial effect:Compared with prior art, the invention provides the control method and system of a kind of application self-starting, institute
The method of stating includes:Hand or pin to be analyzed are shot by machine vision equipment, obtains the figure for specifying the hand or pin of posture
Picture;The background and hand or the disaggregated model of pin of described image are set up according to default human skin colouring information, and according to described
Disaggregated model extracts the image-region of the hand or pin in described image;Described image region is extracted according to the specified posture
The first interior fingernail region, and thumbnail image is extracted in first fingernail region using image partition method.Root of the present invention
Contour feature information according to human skin and nail under color characteristic and given pose is extracted to thumbnail image so that
Accuracy and robustness have obtained larger raising.
Description of the drawings
The flow chart that the extracting method of the thumbnail image that Fig. 1 is provided for the present invention is preferably implemented.
The schematic diagram of gesture is specified in the extracting method of the thumbnail image that Fig. 2 is provided for the present invention.
The hand region image shot in the extracting method of the thumbnail image that Fig. 3 is provided for the present invention.
The schematic diagram of the area-of-interest extracted in the extracting method of the thumbnail image that Fig. 4 is provided for the present invention.
The schematic diagram of the thumbnail image extracted in the extracting method of the thumbnail image that Fig. 5 is provided for the present invention.
Curve matching is from region of interest used in the embodiment one of the extracting method of the thumbnail image that Fig. 6 is provided for the present invention
The schematic diagram of the first fingernail region is extracted in domain.
The structure principle chart of the extraction control system of the thumbnail image that Fig. 7 is provided for the present invention.
The topology view of the placing device that Fig. 8 is provided for the present invention.
The use state figure of the placing device that Fig. 9 is provided for the present invention.
Specific embodiment
The present invention provides a kind of extracting method and system of thumbnail image, to make the purpose of the present invention, technical scheme and effect
Fruit is clearer, clear and definite, and the present invention is described in more detail for the embodiment that develops simultaneously referring to the drawings.It should be appreciated that this place
The specific embodiment of description only to explain the present invention, is not intended to limit the present invention.
In the present invention, using the suffix of such as " module ", " part " or " unit " for being used to represent element only for favourable
In the explanation of the present invention, itself do not have specific meaning.Therefore, module ", " part " or " unit " can mixedly make
With.
Terminal device can be implemented in a variety of manners.For example, the terminal described in the present invention can include such as moving
Phone, smart phone, notebook computer, digit broadcasting receiver, PDA (personal digital assistant), PAD (panel computer), PMP
The mobile terminal of (portable media player), guider etc. and such as numeral TV, desktop computer etc. are consolidated
Determine terminal.However, it will be understood by those skilled in the art that, in addition to being used in particular for moving the element of purpose, according to this
The construction of bright embodiment can also apply to the terminal of fixed type.
Below in conjunction with the accompanying drawings, by the description to embodiment, the content of the invention is described further.
Refer to Fig. 1, the flow chart of the preferred embodiment of the extracting method of the thumbnail image that Fig. 1 is provided for the present invention.Institute
The method of stating includes:
S100, hand or pin to be analyzed are shot by machine vision equipment, obtain the hand of specifying posture or pin
Image.
Specifically, described machine vision equipment can be the camera of mobile phone, the camera of panel computer or pen
Remember that the shooting of this computer is first-class.As long as the equipment of thumbnail image can be taken pictures and obtained to nail can be set as machine vision
It is standby.When the specified posture is to be set in advance in the image for obtaining hand or pin, the posture residing for hand and pin.Here it is right by taking hand as an example
The specified posture is illustrated, as shown in Fig. 2 the specified posture can upwards palm lies against a plane for the back of the hand
On, and the five fingers open.Certainly, in the variant embodiment of the present embodiment, the specified posture can also be user's palm of the hand face
To camera, the five finger gripping thenad hearts are not still mutually blocked.Certainly, the specified posture can also be other postures, here just not
Enumerate.
S200, the background and hand of described image or the disaggregated model of pin are set up according to default human skin colouring information,
And the image-region of the hand or pin is extracted in described image according to the disaggregated model.
Specifically, the human skin color pre-sets, its mankind that can be obtained by substantial amounts of experiment statisticses
Skin color standard value.It can also user voluntarily arranged according to autologous skin color.In actual applications, can be pre-
If multiple human skin colors are stored, and the human skin color is corresponding with ethnic group, and different ethnic group correspondences is different
Human skin color, can so avoid the classification produced because the skin variations between Black people and white man are excessive inaccurate
Problem.
In the present embodiment, the background and hand or pin that described image is set up according to default human skin colouring information
Disaggregated model, and the image-region that the hand or pin are extracted in described image according to the disaggregated model specifically can wrap
Include:
The Background color information of S201, identification described image, and believed according to the background color and human skin color
Breath sets up the background and hand or pin disaggregated model of image;
S202, the image-region for extracting hand or pin in described image according to the disaggregated model.
Specifically, the disaggregated model for setting up image background and hand or pin according to skin color can be according to model of ellipse
Set up based on the disaggregated model of color.The process of setting up of the disaggregated model is specifically as follows:First by the color of described image
Space is transformed into YCbCr space, if certain pixel P is Pb in the value of Cb passages in described image, the value of Cr passages is Pr, if P is full
Foot states condition, then judge that P belongs to hand or pin region, is otherwise then background area.
The condition be according to human skin color information value from color be transformed into YCbCr space when, in the value of Cb passages
The relation met with the value in Cr passages.In such that it is able to judge the corresponding color space of described image according to the relation
Belong to the pixel of hand or pin image-region, and then by background color and the color classification of hand or pin, to realize extracting hand or pin
Image-region.In the present embodiment, the condition can be expressed as:
Wherein, the b0, r0, θ, a and b are constant.
In another embodiment of the present invention, the employing image segmentation extracts the hand or pin in described image
Image-region specifically can include:
S201a, the nominal contour of image-region for obtaining the corresponding standard hand of the specified posture or pin and default
Human skin colouring information;
S202a, background and hand that described image is set up according to the skin color and the nominal contour or pin point
Class model;
S203a, the image-region for extracting the hand or pin in described image according to the sort module.
Specifically, the image-region of the standard hand or pin is to be preset as that the specified posture shoots comprising hand or pin
Image setting, and the storage corresponding with the specified posture of the image-region of the standard hand or pin, so that when user is with institute
When stating posture shooting image, the image-region of the standard hand or pin can be found according to the specified posture.Actually should
With in, can preset and set up a database, for storing the image-region of specified posture and standard hand or pin, and the finger
The image-region for determining posture and the standard hand or pin is mutually bound.Meanwhile, described specifying can also be stored in the database
Also there is its corresponding specified posture binding storage the corresponding standard fingernail region of posture, the standard fingernail region, in order to root
The standard fingernail region is inquired according to the specified posture.
It is described according to image background and hand or pin disaggregated model using the method for color cluster and comparative silhouette set up hand or
Pin disaggregated model.The process of setting up specifically can include:Initially with Expectation Maximization (EM) algorithm
All pixels in described image are carried out with k-means clusters and obtains several classification.Then the method using comparative silhouette is true
Fixed all sort merges form in the profile of image-region hand or pin composition in standard picture corresponding with the specified posture
Image-region nominal contour affinity score highest sort merge, and the image-region that the sort merge is formed is set to
The image-region of hand or pin and extract in described image.
In the present embodiment, the method for the comparative silhouette can be using it is described classify formed the girth of image-region/
Area ratio determines profile similarity fraction for reduced factor.That is, calculating the figure that a certain category combinations are formed first
As the girth and the ratio of area in region, the image-region that the hand of its standard picture corresponding with specified posture and pin are constituted
The ratio of girth and area is compared, and according to the comparative result profile similarity fraction is determined.
In the variant embodiment of the present embodiment, the method for the comparative silhouette can be procrustes analysis
Method.The detailed process for adopting procrustes analysis (Pu Shi analyses) method to determine affinity score can be for:
Point set P is selected in the sampling in the profile of the image-region that described each sort merge is formed, corresponding to the specified posture
Point set P0, and the point set P and point set P0 are selected in the nominal contour sampling of the image-region that hand or pin are constituted in standard picture
Point quantity it is identical.Then, (for example, translation, rotation, minute surface are anti-to calculate a linear transformation according to point set P and point set P0
Penetrate, scale) so that point set P point sets P0 aligns, then the variance of the point set P after conversion and point set P0 is calculated, the variance is point
The diversity factor of collection P and point set P0.Finally, the affinity score for obtaining point set P and point set P0 inverted to the variance.
S300, the first fingernail region in the hand or pin image-region is extracted according to the specified posture, and using figure
As dividing method extracts thumbnail image in first fingernail region.
Specifically, it is described to be referred to according to the posture extraction hand or the first fingernail region in pin image-region
It is rough in the image-region of the hand or pin to extract the first fingernail region.The employing image partition method is in first nail
Extraction thumbnail image refers to fine image of the extraction comprising nail in first fingernail region and obtains nail in region
Image.
Exemplary, it is described that the hand or the first fingernail region in pin image-region are extracted according to the specified posture,
And thumbnail image is extracted in first fingernail region using image partition method specifically can include:
S301, the hand or pin image-region are modified so that the hand or pin image-region are described with default
The image-region alignment of the corresponding standard hand of specified posture or pin.
Specifically, described being modified to the hand or pin image-region refers to carry out in the hand or pin image-region
The alignment operations such as rotation, scaling cause the position of the hand or pin image-region standard hand corresponding with the specified posture or pin
Image-region alignment.The image-region of the standard hand or pin is in all specified posture for pre-setting shared by hand or pin
Image-region.In actual applications, the image-region of the standard hand or pin is in the finger according to a large amount of mankind's hands or pin
Determine the shared the average image region of hand that the image study under posture arrives or pin.
S302, that the hand or pin image-region are determined according to the fingernail region of the standard hand or the image-region of pin
One fingernail region, wherein, first fingernail region includes some first nail subregions.
Specifically, using the region comprising nail in the image-region of the standard hand or pin as when remote holder or pin image district
To determine first fingernail region in described image region, first fingernail region includes the area-of-interest comprising nail in domain
Some first nail subregions.The first nail subregion is the region of hand or pin each nail.In actual applications, in root
Determine behind the first fingernail region according to the standard picture, the colouring information that can be combined with nail further optimizes region of interest
Domain, such as disaggregated model based on color remove the edge of fingernail region to optimize area-of-interest, i.e. the first fingernail region.
S303, nail subregion, some fingers are extracted in each first nail subregion using image partition method
The cycle of sixty years region composition thumbnail image.
Specifically, described image dividing method can be the image segmentation calculation based on nail colour and skin color difference
Method, or the curve-fitting method based on image edge information, or based on nail colour information, image edge information and specified posture
The energy minimization method of lower standard profile information.
In the present embodiment, described image dividing method is the image segmentation calculation based on nail colour and skin color difference
Method.It is described to be with the process of the image segmentation algorithm acquisition nail subregion of skin color difference based on nail colour:It is right
Each first nail subregion carries out principal component analysis and sets up nail colour model, and the nail colour model is to calculate the second He
The weighted sum of the 3rd principal component, and adopt threshold method to extract region of the nail colour value higher than predetermined threshold value for nail area
Domain.In the present embodiment, the principal component analysis (PCA) is if be that image all pixels rgb value is orthogonally transformed into into main line
Property incoherent variable, i.e., described some linear incoherent variables are principal component.Also, the image pixel after the conversion
It is maximum that first principal component meets variance, and it is maximum that Second principal component, meets variance on the premise of first principal component is orthogonal to, and the 3rd
It is maximum that principal component meets variance on the premise of the first and second principal components are orthogonal to.
In another embodiment of the present invention, described image partitioning algorithm is that image edge information is carried out curve fitting
Method.It is described to image edge information carry out curve fitting obtain nail subregion process can be:To each the first finger
The cycle of sixty years, region obtained its corresponding marginal information using edge detection operator (e.g., canny operators);The marginal information is entered again
Row curve matching tries to achieve the boundary curve of nail, for example, the marginal information extracted is intended using cubic spline function curve
Close, then solve the minimum cubic spline function curve of variance as the edge of final fingernail region.
In yet another embodiment of the present invention, described image partitioning algorithm is that for energy minimization method, such as figure cuts method
Or level set method (level set) etc. (graph-cut).Energy term in the energy minimization method can include nail face
Standard skeleton pattern under color information, gradient information and specified posture.Preferably, the nail colour model can be staff or
Second and the 3rd factor weighted method value of pin area image;The gradient information can pass through sobel, prewitt, Laplace
The operators such as of Gaussian are calculated;The known profile model can be current thumbnail image profile and the specified posture
The difference of nominal contour.
In order to further illustrate the extracting method of the thumbnail image of present invention offer, below to the nail figure by taking hand as an example
The extracting method of picture is described further.For convenience of description, if the region of hand images is Ω, the hand figure of required extraction
As region is ΩH, the fingernail region of required extraction is ΩN, these definition are suitable for follow-up full text.
Embodiment one
A kind of extracting method of thumbnail image is present embodiments provided, it includes:
S10, when palm is positioned over placing device by user according to specified posture, obtain the hand images.
Specifically, palm is positioned over user that placing device refers to according to prompting according to specified appearance by user according to specified posture
Will definitely palm extend in the placing device, wherein, the specified posture be the back of the hand upwards, the five fingers open, and the palm of the hand is adjacent to load
Put the holding plane of device.After user places on palm, palm image is shot by smart mobile phone.
The placing device is the device for shooting hand images in order to smart mobile phone for placing palm.The placing dress
Put including the housing with accommodating chamber, the bottom of the accommodating chamber is provided with holding plane, the housing and the holding plane
Relative one side is provided with and is open for the first of external equipment shooting image, and the housing is provided with the stretched into for palm
Two openings, plane intersects with the residing plane of the described second opening residing for first opening.Can be with the inwall of the housing
Be provided with for for shoot light filling LED.In actual applications, the placing device is also provided with camera and leads to
News device so that with direct access hand images, and the hand images can be sent to outside set by the placing device
It is standby.
S20, the internal color according to the placing device and the human skin color set up disaggregated model, according to described
Disaggregated model extracts the hand images region in the hand images.
Specifically, the internal color of the placing device is obtained according to measuring to the color inside placing device
Background color B (lb,ab,bb).The human skin color can pass through average H of the human skin color that a large amount of statistics are obtained
(lh,ah,bh).In actual applications, the internal color is preferably the color big with face skin color difference, e.g., green,
Blueness etc., so that the internal color and hand skin color of placing device has very big contrast, it is possible to will according to color
The all pixels of hand images are sorted out, and the background of hand images is separated with palm, obtain hand region image, such as Fig. 3
It is shown.
The internal color and the human skin color according to the placing device sets up disaggregated model specifically can be with
For:Hand images are transformed in CIE LAB color spaces first, in vision between the color space color and true colors
Gap in impression is consistent (uniform).Then, it is calculated with background color B to each pixel p (l, a, b) in image
(lb,ab,bb) and human skin average H (lh,ah,bh) distance than r (p), when distance is more than 1 than r (p), the pixel category
In palm area, i.e. p ∈ Ω H;When distance is less than or equal to 1 than r (p), the pixel belongs to background area.The tool of the r (p)
Body formula is:
S30, the area-of-interest comprising nail is extracted in the hand images region according to the specified posture.
Specifically, it is described that the region of interest comprising nail is extracted in the hand images region according to the specified posture
Domain is specially corrects the hand images region to corresponding with the corresponding standard hand images region of the specified posture, it
Afterwards according in the standard hand images region comprising nail area-of-interest to determine the hand images region in include
The area-of-interest of nail, to realize extracting the area-of-interest comprising nail in the hand images region, such as Fig. 4 institutes
Show.
In the present embodiment, it is described that the sense comprising nail is extracted in the hand images region according to the specified posture
Interest region specifically includes:
S31, the hand images area is modified, so that the hand images region is corresponding with the specified posture
Standard hand images region it is consistent;
S32, according to the standard hand region determine in the hand images region protect nail area-of-interest simultaneously
Extract.
Specifically, in step S31, described being modified to hand image-region is referred to the hand
Image-region such as is rotated, is scaled at the alignment operation so that the hand images region can be with standard hand images region again
Close, i.e., both size and shapes are identical.In actual applications, the makeover process can be:To hand image-region coordinate meter
Principal component is calculated, the maximum vector of coordinate variance change is extracted, with mark after then causing the vector rotated to image rotation
Quasi- hand images Y-axis is put down, and finally makes it in the same size with standard picture image scaling.
The standard hand images region is to be analyzed study to the hand images of the specified postures in a large number to obtain.Institute
State learning process to be specifically as follows:The hand images that a large number of users is shot with the specified posture are obtained first, then will be all
Hand images are adjusted to uniform sizes, and all hand images carry out study and obtain the standard hand images after to adjustment.
In practical application, the standard hand images can be obtained by the method for image averaging.
In step S32, the corresponding area-of-interest of the standard hand region can be defined as the hand
The corresponding area-of-interest of image-region.That is, obtaining the standard hand can directly use standard drawing according to us
The area-of-interest of picture also can further calculate present image hand or pin image as when the area-of-interest of remote holder or pin image
Convex closure summit, select five minimum summits of Y value, the image-region as sense comprising nail of five summit near zones
Interest region.
S40, thumbnail image is extracted in the area-of-interest using image segmentation, as shown in Figure 5.
Specifically, described image split plot design is based on the approximating method of marginal information, i.e., edge to be detected and is used
The method of cubic spline interpolation.As shown in fig. 6, the process for extracting image is specifically as follows:S41, first to feel it is emerging
Interesting area image carries out edge extracting.In the present embodiment, carry out edge to region of interest area image using canny operators to carry
Take.Canny operators carry out first Gaussian Blur to image and use sobel operators to calculate image gradient information, afterwards to gradient
Result of calculation carries out non-maximum removal, finally sets up high threshold to sieve to choose the strong edge of the region of interest area image, if
Vertical Low threshold by the edge extracting for having neighborhood relationships with strong edge in weak edge out, using weak edge and the strong edge extracted as
The edge of the region of interest area image, such as Fig. 8 .c are figure of the region of interest area image after canny edge extractings
Picture.
Further, because the image after edge extracting may be mingled with the edge letter for being not belonging to real nail edge
Breath, e.g., the overexposure region of a crease in the skin, the vertical line of nail, skin or nail or the edge of skin and background etc..Accordingly, it would be desirable to
Carry out sieves to the image after canny edge extractings to select to obtain accurate nail marginal information, Fig. 8 .d are exhibition
Show the marginal information after filtering.In actual applications, the edge can be screened according to edge character.For example, can be with
A crease in the skin and nail vertical line are removed according to the neighborhood information of each of the edges, that is, calculate the neighborhood information mean square deviation of each of the edges,
If the variance of the edge neighborhood image is more than predetermined threshold value, the edge belongs to real nail edge;If the edge neighborhood figure
The variance of picture is less than or equal to predetermined threshold value, then the edge belongs to a crease in the skin or nail vertical line.Such as overexposure region, we
Setting pixel grey scale is then judged to overexposure when being higher than certain threshold value (such as 220), if a line edge neighborhood overexposure pixel exceeds one
Certainty ratio, then judge that the edge is overexposure edge.
Finally, after nail marginal information is got, the edge after screening is fitted using the method for curve matching.
In the present embodiment, it is fitted using traditional cubic spline function pair edge, the traditional cubic spline function is one point
Section function, each section of function is all a cubic polynomial function, and each section of function is with its neighborhood section function in critical node
Value and first derivative values are continuous and second dervative is 0.If the fit procedure is specially choosing first in all marginal points
Dry initial control point, according to some initial control point interpolation initial natural cubic spline function is gone out, and calculates remaining marginal point
With the distance of the function, if average distance is more than a certain threshold value, adds the farthest marginal point of distance function and recalculate one
Traditional cubic spline function, repeats the process until the distance of all marginal point and functions obtains nail figure less than a certain threshold value
The edge of picture, the image-region of the surrounded by edges is thumbnail image.In the present embodiment, the selection of the initial control point
Process can be the center of gravity for determining marginal point first, penetrate n bars (such as n=36 bars) are launched to around as starting with the center of gravity
Line, the angle between each pair adjacent ray is 360/n degree, and every ray is initial control point with the intersection point of marginal point.It is described
Can being determined according to spline function perimeter L for threshold value, such as can be L/ (2 π * 10).
Embodiment two
In the thumbnail image extracting method that this enforcement is provided, directly the nail of staff is extracted using intelligent terminal,
Methods described is specifically included:.
H10, hand or pin to be analyzed are shot by intelligent terminal, obtain the hand images for specifying posture.
Specifically, when user installation specifies posture to be positioned in hand in one plane, show on the smart machine
The corresponding guidance diagram of the specified posture, and point out user to adjust the position of hand according to the guidance diagram, so that the use
During imaging of the hand at family on smart machine is as the guidance diagram, the difficulty of image segmentation can be so reduced.
H20, background and hand that hand images are set up according to the profile information and human skin information of the specified posture
The disaggregated model in portion, and palm image-region is extracted according to the disaggregated model.
Specifically, the profile information and face skin information of the specified posture is setting up classification mould to pre-set
The profile information and face skin information of the specified posture for prestoring, the profile of the specified posture can be first read during type
Information and face skin information can be stored in smart machine, can be stored in background server etc..
The profile information and human skin information according to the specified posture sets up the background and hand of hand images
The disaggregated model in portion is specifically as follows:All pixels in graphics field are instructed to carry out clustering by color distance so that all by described
Pixel is divided into k (k>=2) individual classification.In the present embodiment, clustered using Expectation Maximization algorithms,
Its detailed process can be the average for initializing k classification first, then calculate the distance of each pixel and each classification average
And the pixel is referred in closest classification, circulation successively completes to sort out up to all pixels, then to each classification
Average is updated, and repeats the process of above-mentioned classification and average renewal until completing convergence.
It is two classifications by the k category division, respectively after basis is divided into k classification to pixel according to color
Background and hand.The partition process is specifically as follows:Being selected according to the average of the k classification may belong to skin area
Part, is then combined to the classification selected, and the profile in the region formed to these combinations judges, searches out
Combine as final hand classification with the posture profile similarity fraction highest.For example, if k=4, will the guidance diagram
All pixels in shape are gathered into 4 classifications according to color.Calculate each classification average and human skin average color respectively again
The distance between, and given threshold selects the classification that may belong to skin.For example, when pixel p is transformed into into hsv passages, and to h
When zooming to [0,1] interval with two passages of s, the threshold value can be 0.1.That is, the value of each passage and mankind's skin
When gap of the skin average color between the value of the passage is all necessarily less than 0.1, could judge that the pixel p belongs to human skin face
Color.
Further, it is assumed here that the average of classification { 1,2,4 } is close to human skin average color.Then, obtain 1,2,
4 } all subsets, respectively including { 1 }, { 1,2 }, { 1,2,4 } ... waits 7 combinations, calculate each combination current outline with
The affinity score of the nominal contour in the corresponding standard hand images region of the specified posture.The computational methods of the affinity score
Girth/the area of profile can be adopted than determination current outline and the affinity score of nominal contour, i.e., according to profile girth and wheel
The affinity score of ratio-dependent current outline and nominal contour between the area included in exterior feature.It can also be selected
Procrustes analysis (Pu Shi analyses) method, i.e., be sampled to select point set P0, and to working as front-wheel to nominal contour
Exterior feature is sampled to select point set P, wherein, the point set P is identical with the point quantity of P0, and according to point set P and point set P0 one is calculated
It is individual go out a linear transformation (e.g., translation, rotation, mirror-reflection, scaling) so that the point set P and point set P0 alignment, meter
Calculation makes the variance minimum of the P after conversion and P0, and the variance is the diversity factor of P and P0, inverted to the variance to be
The affinity score of P and P0 is obtained.It should be noted that needing before procrustes analysis (Pu Shi analyses) method is carried out
Current outline is alignd with nominal contour, for strengthening the corresponding relation of each point of P and P0, alignment operation can be counted respectively
Calculate the finger north orientation amount of current outline and current outline and carry out rotating the finger for making it refer to north orientation amount and nominal contour to current outline
North orientation amount is parallel.It is starting point for regional barycenter contained by profile to herein refer to north orientation amount, terminal be on profile the minimum point of Y value to
Amount.
Further, after the disaggregated model of the background and hand of setting up hand images, hand is extracted according to the disaggregated model
Portion's image-region.In the present embodiment, extract hand images region and method (graph-cut) extraction palm image is cut using figure.Institute
State extraction process and be specifically as follows and first energy cost function is set up to image, the problem of dividing the image into is converted to the energy function
Minimization problem, and cut (graph-cut) method using figure and realize energy minimization process.The energy letter of the energy minimization
Number includes two, respectively area item and border item, and the expression of the energy function is:
Wherein, Ep(Lp) be area item, Lp,Lq∈{ΩH,Ω/ΩH}、Ep(Lp,Lq) it is border item, p, q represent adjacent two
Individual pixel;λ is the coefficient of border item.
In the present embodiment, the area item can determine according to above-mentioned classification results.For example, it is assumed that pixel p is classified
For hand region, then Ep(Lp∈ΩH)=0, Ep(Lp∈Ω/ΩH)=1.The border item can be believed according to the gradient of pixel p
Breath definition, it can be expressed asWherein,For the Grad of pixel p and q, k is a constant, β
=1 or 2.
It is worth explanation, if the front profile and nominal contour affinity score higher (for example, higher than 90 points), then can be straight
Result of the selecting classification results as image segmentation.
H30, the area-of-interest comprising nail is extracted in the hand images region according to the specified posture.
The method for extracting area-of-interest is identical with the method for embodiment one, just repeats no more here.
H40, thumbnail image is extracted in the area-of-interest using image segmentation.
Specifically, described image split plot design is level set (level set) method, and methods described is also to divide the image into ask
Topic is converted to energy minimization problem, and its detailed process can be the profile C0 for initializing a round sealed first, and by institute
Shu Yuan centers are set to region of interest centers, and the circle is set to into a quarter of region of interest field width.Then, iteration institute successively
Profile is stated so that predetermined energy is progressively reduced until reaches extreme value, during each iteration on profile each point according to its normal direction
To external expansion, wherein, the speed of the extension is relevant with energy functional.Each time profile C (t) during iteration is all looked at as letter
Several zero level collection, i.e. every time the process of profile iteration is exactly the process of function Temporal Evolution.
The energy functional is that, based on the energy functional of marginal information, its expression formula is:
E (C)=∫ g (C (s)) ds
Wherein, g is edge indicator function, and the g can be expressed asWherein,For pixel x gradient to
Amount, σ is a constant.
In the present embodiment, extreme value can be asked to E (C) using gradient descent method, obtains the EVOLUTION EQUATION of iteration each time
For:
Wherein,ForGradient vector, Div (v) for vector v divergence, t is the time
In actual applications, can be from initialStart, define the time period Δ t of each iteration, then each iteration
Current function can be calculatedUntil restraining, final profile is final functionZero level collection.It is worth explanation,
The energy functional can also be adjusted to it and add area information and the priori profile information under given pose, just differ here
One explanation.
Present invention also offers a kind of extraction system of thumbnail image, as shown in fig. 7, it includes:
Taking module, for being shot to hand or pin to be analyzed by machine vision, obtain specify posture hand or
The image of pin;
First extraction module, for setting up the background and hand or pin of described image according to default human skin colouring information
Disaggregated model, and the image-region of the hand or pin is extracted in described image according to the disaggregated model;
Second extraction module, for extracting the hand or the first nail area in pin image-region according to the specified posture
Domain, and thumbnail image is extracted in first fingernail region using image partition method.
The extraction system of the thumbnail image, wherein, first extraction module is specifically included:
Acquiring unit, for obtaining the nominal contour of the image of the corresponding standard hand of the specified posture or pin and presetting
Human skin colouring information;
Set up unit, for set up according to the skin color and the nominal contour background of described image and hand or
The disaggregated model of pin;
First extraction unit, for extracting the image district of the hand or pin in described image according to the disaggregated model
Domain.
The extraction system of the thumbnail image, wherein, second extraction module is specifically included:
Amending unit, for being modified to described image region so that described image region is described specified with default
The image-region alignment of the corresponding standard hand of posture or pin;
Determining unit, the fingernail region for being set according to the standard picture region determines the first of described image region
Fingernail region, wherein, first fingernail region includes some first nail subregions;
Second extraction unit, for nail sub-district to be extracted in each first nail subregion using image partition method
Domain, some subregions constitute the thumbnail image.
Present invention also offers a kind of extraction element of thumbnail image, it includes carrying for as above arbitrary described thumbnail image
Take system.
The extraction element of the thumbnail image, it also includes placing device and outside filming apparatus.As shown in Figure 8 and Figure 9,
The placing device includes the housing 1 with accommodating chamber 11, and the bottom of the accommodating chamber 11 is provided with holding plane 12, the shell
The one side relative with the holding plane 12 of body 1 is provided with the first opening 2 for external equipment shooting image, the housing 1
Be provided with for palm stretch into second opening 3, it is described first opening 1 residing for plane and described second opening 3 residing for plane phase
Hand over.
The extraction element of the thumbnail image, wherein, the LED for light filling is provided with the housing.
The extraction system of above-mentioned thumbnail image and the modules of device have been described in detail in the above-mentioned methods, at this
In just no longer state one by one.
In embodiment provided by the present invention, it should be understood that disclosed system and method, can pass through other
Mode is realized.For example, device embodiment described above is only schematic, and for example, the division of the module is only
A kind of division of logic function, can there is an other dividing mode when actually realizing, such as multiple units or component can with reference to or
Person is desirably integrated into another system, or some features can be ignored, or does not perform.Another, shown or discussed is mutual
Between coupling or direct-coupling or communication connection can be INDIRECT COUPLING or communication link by some interfaces, device or unit
Connect, can be electrical, mechanical or other forms.
The unit as separating component explanation can be or may not be it is physically separate, it is aobvious as unit
The part for showing can be or may not be physical location, you can with positioned at a place, or can also be distributed to multiple
On NE.Some or all of unit therein can according to the actual needs be selected to realize the mesh of this embodiment scheme
's.
In addition, each functional unit in each embodiment of the invention can be integrated in a processing unit, it is also possible to
It is that unit is individually physically present, it is also possible to which two or more units are integrated in a unit.Above-mentioned integrated list
Unit both can be realized in the form of hardware, it would however also be possible to employ hardware adds the form of SFU software functional unit to realize.
The above-mentioned integrated unit realized in the form of SFU software functional unit, can be stored in an embodied on computer readable and deposit
In storage media.Above-mentioned SFU software functional unit is stored in a storage medium, including some instructions are used so that a computer
Equipment (can be personal computer, server, or network equipment etc.) or processor (processor) perform the present invention each
The part steps of embodiment methods described.And aforesaid storage medium includes:USB flash disk, portable hard drive, read-only storage (Read-
Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disc or CD etc. it is various
Can be with the medium of store program codes.
Finally it should be noted that:Above example only to illustrate technical scheme, rather than a limitation;Although
The present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those within the art that:It still may be used
To modify to the technical scheme described in foregoing embodiments, or equivalent is carried out to which part technical characteristic;
And these modification or replace, do not make appropriate technical solution essence depart from various embodiments of the present invention technical scheme spirit and
Scope.
Claims (13)
1. a kind of extracting method of thumbnail image, it is characterised in that it includes:
Hand or pin to be analyzed are shot by machine vision equipment, obtains the image for specifying the hand or pin of posture;
The background and hand or the disaggregated model of pin of described image are set up according to default human skin colouring information, and according to described
Disaggregated model extracts the image-region of the hand or pin in described image;
The hand or the first fingernail region in pin image-region are extracted according to the specified posture, and adopts image partition method
Thumbnail image is extracted in first fingernail region.
2. the extracting method of thumbnail image according to claim 1, it is characterised in that described according to default human skin face
Color information sets up the background and hand or the disaggregated model of pin of described image, and is extracted in described image according to the disaggregated model
The image-region of the hand or pin is specifically included:
By the color space conversion of described image to YCbCr space, and obtain value and Cr passage of each pixel in Cb passages
Value;
Background and hand for distinguishing image is determined in the value of Cb passages and the value of Cr passages according to the human skin color
Or the class condition of pin;
The image-region of the hand or pin is extracted in described image according to the class condition.
3. the extracting method of the thumbnail image according to claim 2, it is characterised in that the class condition is:
Wherein, Pb be pixel P Cb passages value, Pr be pixel P Cr passages value, b0, r0, θ, a and b are constant.
4. the extracting method of thumbnail image according to claim 1, it is characterised in that described according to default human skin face
Color information sets up the background and hand or the disaggregated model of pin of described image, and is extracted in described image according to the disaggregated model
The image-region of the hand or pin is specifically included:
Obtain the nominal contour and default human skin color letter of the image of the corresponding standard hand of the specified posture or pin
Breath;
The background and hand or the disaggregated model of pin of described image are set up according to the skin color and the nominal contour;
The image-region of the hand or pin is extracted in described image according to the disaggregated model.
5. according to the claim 4 thumbnail image extracting method, it is characterised in that it is described according to the skin color
Information and the nominal contour set up the background of described image and the disaggregated model of hand or pin is specifically included:
The Background color information of identification described image, and according to the Background color information and skin color information to the figure
The color of picture is clustered;
The all sort merges obtained using the comparative silhouette method calculating cluster form current outline and the institute of image-region
State the affinity score of the corresponding nominal contour of specified posture;
The affinity score highest sort merge is obtained, and the image-region that sort merge is formed is set in described image
The image-region of hand or pin is simultaneously extracted.
6. the extracting method of thumbnail image according to claim 1, it is characterised in that described to be extracted according to the specified posture
The first fingernail region in described image region, and nail figure is extracted in first fingernail region using image partition method
As specifically including:
Described image region is modified so that described image region standard hand corresponding with the default specified posture
Or the image-region alignment of pin;
First fingernail region in described image region is determined according to the fingernail region of standard picture region setting, wherein, institute
The first fingernail region is stated including some first nail subregions;
Nail subregion is extracted in each first nail subregion using image partition method, some subregions constitute institute
State thumbnail image.
7. the extracting method of thumbnail image according to claim 6, it is characterised in that described image dividing method is based on finger
The image segmentation algorithm of first color and skin color difference, based on the curve-fitting method of image edge information, based on nail face
It is a kind of in the energy minimization method of standard profile information under color information, image edge information and specified posture.
8. a kind of extraction system of thumbnail image, it is characterised in that it includes:
Taking module, for being shot to hand or pin to be analyzed by machine vision equipment, obtain specify posture hand or
The image of pin;
First extraction module, for setting up the background and hand of described image or dividing for pin according to default human skin colouring information
Class model, and the image-region of the hand or pin is extracted in described image according to the disaggregated model;
Second extraction module, for extracting the hand or the first fingernail region in pin image-region according to the specified posture,
And thumbnail image is extracted in first fingernail region using image partition method.
9. the extraction system of thumbnail image according to claim 8, it is characterised in that first extraction module is specifically wrapped
Include:
Acquiring unit, for the nominal contour for obtaining the image of the corresponding standard hand of the specified posture or pin and default people
Class skin color information;
Unit is set up, for setting up the background of described image and hand or pin according to the skin color and the nominal contour
Disaggregated model;
First extraction unit, for extracting the image-region of the hand or pin in described image according to the disaggregated model.
10. the extraction system of thumbnail image according to claim 8, it is characterised in that second extraction module is specifically wrapped
Include:
Amending unit, for being modified to described image region so that described image region and the default specified posture
The image-region alignment of corresponding standard hand or pin;
Determining unit, the fingernail region for being set according to the standard picture region determines the of the hand or pin image-region
One fingernail region, wherein, first fingernail region includes some first nail subregions;
Second extraction unit, for nail subregion, institute to be extracted in each first nail subregion using image partition method
State some subregions and constitute the thumbnail image.
11. a kind of extraction elements of thumbnail image, it is characterised in that it includes the nail as described in claim 10-12 is arbitrary
The extraction system of image.
12. according to claim 11 thumbnail image extraction element, it is characterised in that its also include placing device, it is described
Placing device includes the housing with accommodating chamber, and the bottom of the accommodating chamber is provided with holding plane, and the housing is put with described
The relative one side in horizontalization face is provided with and is open for the first of external equipment shooting image, and the housing is provided with to be stretched for palm
The second opening for entering, plane intersects with the residing plane of the described second opening residing for first opening.
13. according to claim 11 thumbnail image extraction element, it is characterised in that be provided with the housing for mending
The LED of light.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611215028.2A CN106651879B (en) | 2016-12-23 | 2016-12-23 | Method and system for extracting nail image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611215028.2A CN106651879B (en) | 2016-12-23 | 2016-12-23 | Method and system for extracting nail image |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106651879A true CN106651879A (en) | 2017-05-10 |
CN106651879B CN106651879B (en) | 2020-04-14 |
Family
ID=58827874
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611215028.2A Active CN106651879B (en) | 2016-12-23 | 2016-12-23 | Method and system for extracting nail image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106651879B (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111062962A (en) * | 2019-12-06 | 2020-04-24 | 昆明理工大学 | Multi-threshold ultrasonic image segmentation method based on differential search algorithm |
CN111091519A (en) * | 2019-12-20 | 2020-05-01 | 维沃移动通信有限公司 | Image processing method and device |
WO2020140198A1 (en) * | 2019-01-02 | 2020-07-09 | 深圳市邻友通科技发展有限公司 | Fingernail image segmentation method, apparatus and device, and storage medium |
CN111639562A (en) * | 2020-05-15 | 2020-09-08 | 圣点世纪科技股份有限公司 | Intelligent positioning method for palm region of interest |
CN112183388A (en) * | 2020-09-30 | 2021-01-05 | 北京字节跳动网络技术有限公司 | Image processing method, apparatus, device and medium |
CN113194323A (en) * | 2021-04-27 | 2021-07-30 | 口碑(上海)信息技术有限公司 | Information interaction method, multimedia information interaction method and device |
CN113435982A (en) * | 2021-07-22 | 2021-09-24 | 创优数字科技(广东)有限公司 | Nail polish recommendation method, device, equipment and medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CA2285734A1 (en) * | 1998-10-14 | 2000-04-14 | Dino Tessarolo | Nail covering system |
CN103473563A (en) * | 2013-09-23 | 2013-12-25 | 程涛 | Fingernail image processing method and system, and fingernail feature analysis method and system |
CN105353634A (en) * | 2015-11-30 | 2016-02-24 | 北京地平线机器人技术研发有限公司 | Household appliance and method for controlling operation by gesture recognition |
CN105433577A (en) * | 2014-09-22 | 2016-03-30 | 卡西欧计算机株式会社 | Nail information detection device, drawing apparatus, and nail information detection method |
CN106127181A (en) * | 2016-07-02 | 2016-11-16 | 乐活无限(北京)科技有限公司 | One is virtual tries manicure method, system on |
-
2016
- 2016-12-23 CN CN201611215028.2A patent/CN106651879B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CA2285734A1 (en) * | 1998-10-14 | 2000-04-14 | Dino Tessarolo | Nail covering system |
CN103473563A (en) * | 2013-09-23 | 2013-12-25 | 程涛 | Fingernail image processing method and system, and fingernail feature analysis method and system |
CN105433577A (en) * | 2014-09-22 | 2016-03-30 | 卡西欧计算机株式会社 | Nail information detection device, drawing apparatus, and nail information detection method |
CN105353634A (en) * | 2015-11-30 | 2016-02-24 | 北京地平线机器人技术研发有限公司 | Household appliance and method for controlling operation by gesture recognition |
CN106127181A (en) * | 2016-07-02 | 2016-11-16 | 乐活无限(北京)科技有限公司 | One is virtual tries manicure method, system on |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2020140198A1 (en) * | 2019-01-02 | 2020-07-09 | 深圳市邻友通科技发展有限公司 | Fingernail image segmentation method, apparatus and device, and storage medium |
CN111062962A (en) * | 2019-12-06 | 2020-04-24 | 昆明理工大学 | Multi-threshold ultrasonic image segmentation method based on differential search algorithm |
CN111062962B (en) * | 2019-12-06 | 2022-09-27 | 昆明理工大学 | Multi-threshold ultrasonic image segmentation method based on differential search algorithm |
CN111091519A (en) * | 2019-12-20 | 2020-05-01 | 维沃移动通信有限公司 | Image processing method and device |
CN111091519B (en) * | 2019-12-20 | 2023-04-28 | 维沃移动通信有限公司 | Image processing method and device |
CN111639562A (en) * | 2020-05-15 | 2020-09-08 | 圣点世纪科技股份有限公司 | Intelligent positioning method for palm region of interest |
CN112183388A (en) * | 2020-09-30 | 2021-01-05 | 北京字节跳动网络技术有限公司 | Image processing method, apparatus, device and medium |
CN113194323A (en) * | 2021-04-27 | 2021-07-30 | 口碑(上海)信息技术有限公司 | Information interaction method, multimedia information interaction method and device |
CN113194323B (en) * | 2021-04-27 | 2023-11-10 | 口碑(上海)信息技术有限公司 | Information interaction method, multimedia information interaction method and device |
CN113435982A (en) * | 2021-07-22 | 2021-09-24 | 创优数字科技(广东)有限公司 | Nail polish recommendation method, device, equipment and medium |
Also Published As
Publication number | Publication date |
---|---|
CN106651879B (en) | 2020-04-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106651879A (en) | Method and system for extracting nail image | |
CN104798101B (en) | Makeup auxiliary device, cosmetic auxiliary method and makeup auxiliary program | |
CN103927016B (en) | Real-time three-dimensional double-hand gesture recognition method and system based on binocular vision | |
EP3101594A1 (en) | Saliency information acquisition device and saliency information acquisition method | |
CN106682601B (en) | A kind of driver's violation call detection method based on multidimensional information Fusion Features | |
Maly et al. | Built-in current testing-feasibility study | |
CN104268583B (en) | Pedestrian re-recognition method and system based on color area features | |
CN106897675A (en) | The human face in-vivo detection method that binocular vision depth characteristic is combined with appearance features | |
CN108717524A (en) | It is a kind of based on double gesture recognition systems and method for taking the photograph mobile phone and artificial intelligence system | |
CN107844736A (en) | iris locating method and device | |
CN106485222A (en) | A kind of method for detecting human face being layered based on the colour of skin | |
CN108197534A (en) | A kind of head part's attitude detecting method, electronic equipment and storage medium | |
CN106097354B (en) | A kind of hand images dividing method of combining adaptive Gauss Face Detection and region growing | |
Wang et al. | Pill recognition using minimal labeled data | |
CN104598889B (en) | The method and apparatus of Human bodys' response | |
CN110032932B (en) | Human body posture identification method based on video processing and decision tree set threshold | |
CN111160194B (en) | Static gesture image recognition method based on multi-feature fusion | |
CN110415212A (en) | Abnormal cell detection method, device and computer readable storage medium | |
CN110363747A (en) | Intelligent abnormal cell judgment method, device and computer readable storage medium | |
CN107506738A (en) | Feature extracting method, image-recognizing method, device and electronic equipment | |
Chung et al. | A realtime hand gesture recognition based on haar wavelet representation | |
CN103440633A (en) | Digital image automatic speckle-removing method | |
CN103778430B (en) | Rapid face detection method based on combination between skin color segmentation and AdaBoost | |
CN105631456B (en) | A kind of leucocyte method for extracting region based on particle group optimizing ITTI model | |
CN110728242A (en) | Image matching method and device based on portrait recognition, storage medium and application |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |