CN108269290A - Skin complexion recognition methods and device - Google Patents
Skin complexion recognition methods and device Download PDFInfo
- Publication number
- CN108269290A CN108269290A CN201810053781.9A CN201810053781A CN108269290A CN 108269290 A CN108269290 A CN 108269290A CN 201810053781 A CN201810053781 A CN 201810053781A CN 108269290 A CN108269290 A CN 108269290A
- Authority
- CN
- China
- Prior art keywords
- skin
- pixel
- area
- skin area
- identified
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- 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
- G06T2207/30201—Face
Abstract
The present invention provides a kind of skin complexion recognition methods and device.The method includes:The face skin confidence level of each pixel in facial image to be identified is calculated based on partitioning into skin model, and the face skin credible degree identification based on each pixel obtains the skin area in facial image to be identified;Face skin confidence level based on each pixel in skin area extracts target skin area from skin area, and the target skin area extracted is transformed into LAB color spaces;Color similarity of the target skin area in LAB color spaces between default skin color model standard is calculated, and the skin complexion of facial image to be identified is identified based on the color similarity being calculated.The method skin segmentation precision is high, and skin color model precision is high, and high precisely land for building skin region segmentation can be carried out to facial image to be identified, interference of the environmental factor to skin area skin color model is excluded, finally identifies corresponding high-precision Skin Color Information.
Description
Technical field
The present invention relates to technical field of image processing, in particular to a kind of skin complexion recognition methods and device.
Background technology
With the continuous development of image processing techniques, the application of technology of skin analysis is carried out for facial image also more
Extensively.The facial image skin analysis scheme that industry mainstream uses at present is that facial image directly is transformed into YCrCb colors sky
In, the mode of colour of skin Threshold segmentation is carried out to the colouring information in each region of the facial image based on the YCrCb color spaces, it will
Skin area in the facial image is split, and be correspondingly made available the Skin Color Information of skin area in the facial image.
But the skin segmentation precision of this facial image skin analysis scheme is not high, and skin color model precision is not high, and the program exists
During carrying out skin analysis, the background image that color is similar to the colour of skin in facial image can not be eliminated, segmentation is caused to obtain
Face skin area often carry the content of background image, the segmenting edge of skin area is inaccurate, and in skin area
There are stronger environmental disturbances factor (for example, bloom, shades etc.), influence the identification to the colour of skin and judge, are obtained so as to cause final
To Skin Color Information miss by a mile compared with actual conditions.
Invention content
In order to overcome above-mentioned deficiency of the prior art, the purpose of the present invention is to provide a kind of skin complexion recognition methods
And device, the skin complexion method skin segmentation precision is high, and skin color model precision is high, can be to face figure to be identified
As carrying out accurately skin segmentation, side by side except interference of the environment disturbing factor to skin area skin color model process, thus
Identify the corresponding high-precision Skin Color Information of skin area.
For method, preferred embodiments of the present invention provide a kind of skin complexion recognition methods, the method includes:
The face skin confidence level of each pixel in facial image to be identified is calculated based on partitioning into skin model, and based on each
The face skin credible degree identification of pixel obtains the skin area in the facial image to be identified;
Face skin confidence level based on each pixel in the skin area extracts target from the skin area
Skin area, and the target skin area extracted is transformed into LAB color spaces;
It is similar to calculate color of the target skin area in LAB color spaces between default skin color model standard
It spends, and the skin complexion of the facial image to be identified is identified based on the color similarity being calculated.
For device, preferred embodiments of the present invention provide a kind of skin complexion identification device, and described device includes:
Skin identification module, for calculating the face skin of each pixel in facial image to be identified based on partitioning into skin model
Skin confidence level, and the face skin credible degree identification based on each pixel obtains the skin region in the facial image to be identified
Domain;
Region extraction module, for the face skin confidence level based on each pixel in the skin area from the skin
Target skin area is extracted in skin region, and the target skin area extracted is transformed into LAB color spaces;
Skin color model module, for calculate the target skin area in LAB color spaces with default skin color model mark
Color similarity between standard, and identify based on the color similarity being calculated the skin skin of the facial image to be identified
Color.
In terms of existing technologies, the skin complexion recognition methods and device that preferred embodiments of the present invention provide have
Following advantageous effect:The skin complexion method skin segmentation precision is high, and skin color model precision is high, can be to be identified
Facial image carries out accurately skin segmentation, side by side except environment disturbing factor does skin area skin color model process
It disturbs, so as to identify the corresponding high-precision Skin Color Information of skin area.First, the method is treated based on the calculating of partitioning into skin model
Identify the face skin confidence level of each pixel in facial image, and the face skin credible degree identification based on each pixel obtains
Skin area in the facial image to be identified;Then, people of the method based on each pixel in the skin area
Face skin confidence level extracts target skin area from the skin area, and the target skin area extracted is converted
Into LAB color spaces;Finally, the method by calculate the target skin area in LAB color spaces with default skin
Color similarity between color criterion of identification, and the facial image to be identified is identified based on the color similarity being calculated
Skin complexion.Wherein described partitioning into skin model trains to obtain based on a large amount of face skin area distributed datas, the method
Accurately skin segmentation can carry out facial image to be identified by the partitioning into skin model, identification obtains precision
High skin area;The target skin area is the region that face skin confidence level is higher in the skin area that identification obtains,
The method carries out Skin Color Information identification by obtaining target skin area, and based on the target skin area, can exclude
Interference of the environmental disturbances factor to skin color model process, so as to obtain the higher Skin Color Information of precision.
For the above objects, features and advantages of the present invention is enable to be clearer and more comprehensible, present pre-ferred embodiments cited below particularly,
And attached drawing appended by coordinating, it is described in detail below.
Description of the drawings
It in order to illustrate the technical solution of the embodiments of the present invention more clearly, below will be to needed in the embodiment attached
Figure is briefly described, it should be understood that the following drawings illustrates only certain embodiments of the present invention, therefore is not construed as pair
The restriction of the claims in the present invention protection domain, for those of ordinary skill in the art, what is do not made the creative labor
Under the premise of, it can also be obtained according to these attached drawings other relevant attached drawings.
Fig. 1 is a kind of block diagram of image processing equipment that preferred embodiments of the present invention provide.
Fig. 2 is a kind of flow diagram of skin complexion recognition methods that preferred embodiments of the present invention provide.
Fig. 3 is a kind of flow diagram of sub-step that step S210 shown in Fig. 2 includes.
Fig. 4 is a kind of flow diagram of sub-step that step S220 shown in Fig. 2 includes.
Fig. 5 is a kind of flow diagram of sub-step that step S230 shown in Fig. 2 includes.
Fig. 6 is a kind of block diagram of skin complexion identification device that preferred embodiments of the present invention provide.
Icon:10- image processing equipments;11- memories;12- processors;13- communication units;14- display units;100-
Skin complexion identification device;110- skin identification modules;120- region extraction modules;130- skin color model modules.
Specific embodiment
Purpose, technical scheme and advantage to make the embodiment of the present invention are clearer, below in conjunction with the embodiment of the present invention
In attached drawing, the technical solution in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is
Part of the embodiment of the present invention, instead of all the embodiments.The present invention being usually described and illustrated herein in the accompanying drawings is implemented
The component of example can be configured to arrange and design with a variety of different.
Therefore, below the detailed description of the embodiment of the present invention to providing in the accompanying drawings be not intended to limit it is claimed
The scope of the present invention, but be merely representative of the present invention selected embodiment.Based on the embodiments of the present invention, this field is common
Technical staff's all other embodiments obtained without creative efforts belong to the model that the present invention protects
It encloses.
It should be noted that:Similar label and letter represents similar terms in following attached drawing, therefore, once a certain Xiang Yi
It is defined in a attached drawing, does not then need to that it is further defined and explained in subsequent attached drawing.
In the description of the present invention, it should be noted that term " first ", " second ", " third " etc. are only used for distinguishing and retouch
It states, and it is not intended that instruction or hint relative importance.
Below in conjunction with the accompanying drawings, it elaborates to some embodiments of the present invention.In the absence of conflict, it is following
Feature in embodiment and embodiment can be combined with each other.
Fig. 1 is please referred to, is a kind of block diagram for the image processing equipment 10 that preferred embodiments of the present invention provide.
In the embodiment of the present invention, described image processing equipment 10 is for handling image, to obtain the figure included in correspondence image
As information, wherein described image processing equipment 10 can be by carrying out facial image to be identified skin area identification and colour of skin knowledge
Not, obtain corresponding to the Skin Color Information of skin area in the facial image to be identified.In the present embodiment, described image processing equipment
10 may be, but not limited to, smart mobile phone, PC (Personal Computer, PC), server, tablet computer, a
Personal digital assistant (Personal Digital Assistant, PDA), mobile internet surfing equipment (Mobile Internet
Device, MID) etc..
In the present embodiment, described image processing equipment 10 includes skin complexion identification device 100, memory 11, processing
Device 12, communication unit 13 and display unit 14.The memory 11, processor 12, communication unit 13 and display unit 14 are each
Element is directly or indirectly electrically connected between each other, to realize the transmission of data or interaction.For example, these elements are mutual
It can be realized and be electrically connected by one or more communication bus or signal wire.The skin complexion identification device 100 is included at least
One can be stored in the software function module in the memory 11, the processing in the form of software or firmware (firmware)
Device 12 is stored in the 100 corresponding software function module of the skin complexion identification device in memory 11 by operation, so as to
Perform various functions application and data processing.
In the present embodiment, the memory 11 can be used for storing facial image to be identified, it can also be used to which knowledge is treated in storage
Others' face image carries out the partitioning into skin model of face skin segmentation.The memory 11 may be, but not limited to, at random
Memory (Random Access Memory, RAM) is accessed, read-only memory (Read Only Memory, ROM) may be programmed
Read-only memory (Programmable Read-Only Memory, PROM), Erasable Programmable Read Only Memory EPROM
(Erasable Programmable Read-Only Memory, EPROM), electrically erasable programmable read-only memory
(Electric Erasable Programmable Read-Only Memory, EEPROM) etc..Wherein, memory 11 may be used also
For storing various application programs, the processor 12 performs the application program after execute instruction is received.Further
Ground, software program and module in above-mentioned memory 11 may also include operating system, may include various for managing system
The component software of task (such as memory management, storage device control, power management etc.) and/or driving, and can be with various hardware
Or component software is in communication with each other, so as to provide the running environment of other software component.
In the present embodiment, the processor 12 can be a kind of IC chip with signal handling capacity.Its
Described in processor 12 can be general processor, including central processing unit (Central Processing Unit, CPU), net
Network processor (Network Processor, NP) etc..It can realize or perform the disclosed each side in the embodiment of the present invention
Method, step and logic diagram.General processor can be microprocessor or the processor can also be any conventional processing
Device etc..
In the present embodiment, the communication unit 13 be used for by network establish described image processing equipment 10 and other outside
Communication connection between portion's equipment, and pass through the network and carry out data transmission, wherein described image processing equipment 10 can pass through
The communication unit 13 obtains facial image to be identified at other described external equipments or will be with the facial image pair to be identified
The Skin Color Information answered is sent to corresponding external equipment.
In the present embodiment, the display unit 14 is used to carry out calculation process to image data, and correspondingly to image
It is shown.Wherein described display unit 14 can include GPU (Graphics Processing Unit, graphics processor) and
Display, the GPU are used to the facial image to be identified shown needed for image processing equipment 10 carrying out conversion driving, and control
The display is shown.
In the present embodiment, described image processing equipment 10 is identified by the skin complexion being stored in the memory 11
Device 100 carries out accurately skin segmentation to facial image to be identified, side by side except environment disturbing factor is to skin area skin
The interference of color identification process, so as to identify the corresponding high-precision Skin Color Information of skin area.
It is understood that structure shown in FIG. 1 is only a kind of structure diagram of image processing equipment 10, described image
Processing equipment 10 may also include than shown in Fig. 1 more either less components or with the configuration different from shown in Fig. 1.
Hardware may be used in each component shown in Fig. 1, software or combination is realized.
Fig. 2 is please referred to, is a kind of flow diagram for the skin complexion recognition methods that preferred embodiments of the present invention provide.
In embodiments of the present invention, the skin complexion recognition methods is applied to the image processing equipment 10 shown in Fig. 1, wherein described
The partitioning into skin mould for being partitioned into high-precision face skin area from facial image is stored in image processing equipment 10
Type is below described in detail the idiographic flow and step of skin complexion recognition methods shown in Fig. 2.
In embodiments of the present invention, the skin complexion recognition methods includes the following steps:
Step S210 calculates the face skin confidence of each pixel in facial image to be identified based on partitioning into skin model
Degree, and the face skin credible degree identification based on each pixel obtains the skin area in the facial image to be identified.
In the present embodiment, the partitioning into skin model is trained based on a large amount of face skin area distributed data
It arrives, wherein the face skin area distributed data includes different skin distribution data texturing and different skin on face
Colouring information at corresponding region.The partitioning into skin model is the people to be identified for facial image to be identified
The pattern mask of face image, described image processing equipment 10 can be by the partitioning into skin models from the facial image to be identified
Middle extraction identifies corresponding skin area.In an embodiment of the present embodiment, the partitioning into skin model is to be based on
Deep learning algorithm is iterated what training obtained.
In the present embodiment, described image processing equipment 10 can be based on the partitioning into skin model and calculate face figure to be identified
The face skin confidence level associated with face skin of each pixel as in, and according to the face skin confidence level of each pixel
Identify the skin area in the facial image to be identified.Wherein described face skin confidence level is used to characterize corresponding pixel points
The degree of reliability of the image compared with real skin.
In an embodiment of the present embodiment, described image processing equipment 10 can be by will be in facial image to be identified
The corresponding image information of each pixel is input in the partitioning into skin model, the skin included based on the partitioning into skin model
The colouring information of distribution data texturing and skin of the skin on face is believed with the image of each pixel in the facial image to be identified
Breath is compared, and is based on belief propagation algorithm by each pixel in the facial image to be identified and the partitioning into skin mould
Comparison result between the data that type includes is converted to face skin confidence level associated with face skin, so as to which correspondence is defeated
Go out one and the corresponding single channel mask of facial image to be identified, include the people to be identified in the single channel mask
The corresponding face skin confidence level of each pixel in face image.
Optionally, Fig. 3 is please referred to, is a kind of flow diagram for the sub-step that step S210 shown in Fig. 2 includes.
In the present embodiment, the face skin credible degree identification based on each pixel obtains the face figure to be identified in the step S210
As in skin area the step of can include sub-step S211 and sub-step S212:
Sub-step S211, by the face skin confidence level of each pixel in the facial image to be identified and the first confidence level
Threshold value is compared, and obtains the corresponding comparison result of each pixel.
In the present embodiment, the single-pass corresponding with the facial image to be identified that described image processing equipment 10 is got
In road mask, the face skin confidence level that can be characterized corresponding to the pixel of skin area is using numberical range as 1~255
What gray value was indicated, and it is with numerical value to characterize skin area with the face skin confidence level corresponding to the pixel of exterior domain
Gray value for 0 is indicated, if the image information of corresponding pixel points is closer to healthy skin, corresponding face skin confidence
The value of degree is bigger, i.e., corresponding gray value is closer to 255.In the present embodiment, first confidence threshold value is 0, the figure
As processing equipment 10 is by by the face skin confidence level of pixel each in facial image to be identified and first confidence threshold value
The mode being compared, judges whether each pixel belongs to skin area.
Sub-step S212 extracts corresponding skin region according to the comparison result from the facial image to be identified
Domain, wherein the face skin confidence level of each pixel is more than first confidence threshold value in the skin area.
In the present embodiment, the people of the corresponding each pixel of skin area extracted from the facial image to be identified
Face skin confidence level is all higher than first confidence threshold value, and described image processing equipment 10 can be by from the face to be identified
Side of the face skin confidence level more than the pixel of first confidence threshold value is filtered out in all pixels point that image includes
Formula accordingly extracts skin area from the facial image to be identified.
Referring once again to Fig. 2, step S220, the face skin confidence level based on each pixel in the skin area from
Target skin area is extracted in the skin area, and the target skin area extracted is transformed into LAB color spaces
It is interior.
In the present embodiment, described image processing equipment 10 extracts skin area in the facial image to be identified
Afterwards, it will be extracted from the skin area according to the face skin confidence level of pixel each in the skin area and effectively arranged
In addition to the target skin area of the interference of environmental disturbances factor (for example, the factors such as bloom, shade, flush, acne print, color spot), and
The target skin area is transformed into LAB color spaces, obtains the pattern colour of each pixel in the target skin area
Coloured silk corresponding L * component, A components and B component in the LAB color spaces.
Fig. 4 is please referred to, is a kind of flow diagram for the sub-step that step S220 shown in Fig. 2 includes.In this implementation
In example, the face skin confidence level based on each pixel in the skin area in the step S220 is from the skin region
The step of target skin area is extracted in domain can include sub-step S221 and sub-step S222:
Sub-step S221, by the face skin confidence level of each pixel in the skin area respectively with the second confidence level
Threshold value is compared, and obtains the corresponding comparison result of each pixel in the skin area.
In the present embodiment, the target skin area for the facial image to be identified skin area in confidence level compared with
High region, described image processing equipment 10 can be by the way that the face skin confidence level of each pixel in the skin area be divided
The mode not being compared with the second confidence threshold value, judges whether each pixel in skin area belongs to purported skin area
Domain.
Sub-step S222 extracts corresponding target skin area according to the comparison result from the skin area,
The face skin confidence level of each pixel is more than second confidence threshold value in wherein described target skin area.
In the present embodiment, the face skin confidence level of each pixel is all higher than described second in the target skin area
Confidence threshold value, described image processing equipment 10 can be by filtering out face from all pixels point that the skin area includes
Skin confidence level is more than the mode of the pixel of second confidence threshold value, and corresponding mesh is extracted from the skin area
Mark skin area.Wherein described second confidence threshold value is more than first confidence threshold value, second confidence threshold value
Numerical value may be, but not limited to, and 120,150,180 or 200 etc., specific numerical value can carry out different set according to actual demand
It puts.
Step S230 calculates the target skin area in LAB color spaces between default skin color model standard
Color similarity, and identify based on the color similarity being calculated the skin complexion of the facial image to be identified.
In the present embodiment, the default skin color model standard is preset for identifying including for face skin complexion
The color standards of various standard Skin Color Informations, described image processing equipment 10 can pass through the mesh to being transformed into LAB color spaces
Color similarity in mark skin area and default skin color model standard between each colour of skin calculated, and according to being calculated
Each color similarity identifies the skin complexion of the facial image to be identified.
Optionally, Fig. 5 is please referred to, is a kind of flow diagram for the sub-step that step S230 shown in Fig. 2 includes.
In the present embodiment, the calculating target skin area in the step S230 in LAB color spaces with default skin color model
The step of color similarity between standard, can include sub-step S231, sub-step S232 and sub-step S233:
Sub-step S231 carries out mean value meter to tri- components of L, A, B of each pixel in the target skin area respectively
It calculates, obtains the L * component average value, A components average value and B component average value of the target skin area.
Sub-step S232, L * component, A based on each colour of skin in the default skin color model standard in LAB color spaces divide
Amount and B component, L * component average value, A components average value and B component average value with the target skin area, calculating acquire institute
State the aberration between each colour of skin in target skin area and the default skin color model standard.
In the present embodiment, described image processing equipment 10 can be public according to CIDE2000 colour difference formulas or CIE1976 aberration
The Colorimetry formula based on LAB color spaces of formula etc., to the target skin area and the default skin color model standard
In aberration between each colour of skin calculated.In an embodiment of the present embodiment, described image processing equipment 10 uses
CIDE2000 colour difference formulas calculate the aberration between each colour of skin in the mark skin area and the default skin color model standard.
Sub-step S233 obtains the purported skin based on the Colorimetry between the target skin area and each colour of skin
Color similarity between region and each colour of skin.
In the present embodiment, if the aberration between two kinds of colors is bigger, the color similarity between both colors is got over
It is low.Described image processing equipment 10 can be according in the target skin area and the default skin color model standard being calculated
Aberration between each colour of skin, the color similarity between the target skin area and the corresponding colour of skin calculate, and obtain institute
State the color similarity between target skin area and each colour of skin.
In the present embodiment, purported skin of the described image processing equipment 10 in the facial image to be identified is obtained
It, will be based on the color phase being calculated after color similarity in region and the default skin color model standard between each colour of skin
The skin complexion of the facial image to be identified is identified like degree.It is wherein described to be identified based on the color similarity being calculated
The step of skin complexion of the facial image to be identified, includes:
According to the color similarity between each colour of skin in the target skin area and the default skin color model standard, from
The colour of skin for the similarity maximum that gets colors in the corresponding each colour of skin of the default skin color model standard, as treating described in identifying
Identify the skin complexion of facial image.
Fig. 6 is please referred to, is a kind of box signal for the skin complexion identification device 100 that preferred embodiments of the present invention provide
Figure.In embodiments of the present invention, the skin complexion identification device 100 includes skin identification module 110, region extraction module
120 and skin color model module 130.
The skin identification module 110 calculates each pixel in facial image to be identified for being based on partitioning into skin model
Face skin confidence level, and the face skin credible degree identification based on each pixel is obtained in the facial image to be identified
Skin area.
In the present embodiment, face skin credible degree identification of the skin identification module 110 based on each pixel obtains
The mode of skin area in the facial image to be identified includes:
The face skin confidence level of each pixel in the facial image to be identified is compared with the first confidence threshold value
Compared with obtaining the corresponding comparison result of each pixel;
Corresponding skin area is extracted from the facial image to be identified according to the comparison result, wherein the skin
The face skin confidence level of each pixel is more than first confidence threshold value in skin region.
Wherein, the skin identification module 110 can perform the sub-step shown in step S210 and Fig. 3 shown in Fig. 2
Rapid S211 and sub-step S212, specific implementation procedure are referred to above to step S210, sub-step S211 and sub-step
The detailed description of S212.
The region extraction module 120, for the face skin confidence level based on each pixel in the skin area
Target skin area is extracted from the skin area, and the target skin area extracted is transformed into LAB colors sky
In.
In the present embodiment, face skin of the region extraction module 120 based on each pixel in the skin area
The mode that skin confidence level extracts target skin area from the skin area includes:
The face skin confidence level of each pixel in the skin area is compared respectively with the second confidence threshold value
Compared with obtaining the corresponding comparison result of each pixel in the skin area;
Corresponding target skin area is extracted from the skin area according to the comparison result, wherein the target
The face skin confidence level of each pixel is more than second confidence threshold value in skin area.
Wherein, the region extraction module 120 can perform the sub-step shown in step S220 and Fig. 4 shown in Fig. 2
Rapid S221 and sub-step S222, specific implementation procedure are referred to above to step S220, sub-step S221 and sub-step
The detailed description of S222.
The skin color model module 130, for calculate the target skin area in LAB color spaces with the default colour of skin
Color similarity between criterion of identification, and the facial image to be identified is identified based on the color similarity being calculated
Skin complexion.
In the present embodiment, the skin color model module 130 calculate the target skin area in LAB color spaces with
The mode of color similarity between default skin color model standard includes:
Mean value computation is carried out to tri- components of L, A, B of each pixel in the target skin area respectively, is obtained described
L * component average value, A components average value and the B component average value of target skin area;
L * component, A components and B component based on each colour of skin in default skin color model standard in LAB color spaces, with institute
The L * component average value, A components average value and B component average value of target skin area are stated, calculating acquires the target skin area
With the aberration between each colour of skin in the default skin color model standard;
The target skin area and each skin are obtained based on the Colorimetry between the target skin area and each colour of skin
Color similarity between color.
In the present embodiment, the skin color model module 130 identifies described treat based on the color similarity being calculated
Identify that the mode of the skin complexion of facial image includes:
According to the color similarity between each colour of skin in the target skin area and the default skin color model standard, from
The colour of skin for the similarity maximum that gets colors in the corresponding each colour of skin of the default skin color model standard, as treating described in identifying
Identify the skin complexion of facial image.
Wherein, the skin color model module 130 can perform the sub-step shown in step S230 and Fig. 5 shown in Fig. 2
Rapid S231, sub-step S232 and sub-step S232, specific implementation procedure are referred to above to step S230, sub-step
The detailed description of S231, sub-step S232 and sub-step S233.
In conclusion in the skin complexion recognition methods provided in preferred embodiments of the present invention and device, the skin
Colour of skin method skin segmentation precision is high, and skin color model precision is high, can carry out accurate land for building to facial image to be identified
Skin region segmentation, side by side except interference of the environment disturbing factor to skin area skin color model process, so as to identify skin area
Corresponding high-precision Skin Color Information.First, the method is based on partitioning into skin model and calculates each pixel in facial image to be identified
The face skin confidence level of point, and the face skin credible degree identification based on each pixel is obtained in the facial image to be identified
Skin area;Then, face skin confidence level of the method based on each pixel in the skin area is from the skin
Target skin area is extracted in skin region, and the target skin area extracted is transformed into LAB color spaces;Most
Afterwards, the method is by calculating face of the target skin area in LAB color spaces between default skin color model standard
Color similarity, and identify based on the color similarity being calculated the skin complexion of the facial image to be identified.Wherein institute
It states partitioning into skin model to train to obtain based on a large amount of face skin area distributed datas, the method passes through the partitioning into skin mould
Type can carry out facial image to be identified accurately skin segmentation, and identification obtains the high skin area of precision;It is described
Target skin area is the region that face skin confidence level is higher in the skin area that identification obtains, and the method is by obtaining mesh
Skin area is marked, and Skin Color Information identification is carried out based on the target skin area, environmental disturbances factor can be excluded to the colour of skin
The interference of identification process, so as to obtain the higher Skin Color Information of precision.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this field
For art personnel, the invention may be variously modified and varied.All within the spirits and principles of the present invention, that is made any repaiies
Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (10)
1. a kind of skin complexion recognition methods, which is characterized in that the method includes:
The face skin confidence level of each pixel in facial image to be identified is calculated based on partitioning into skin model, and based on each pixel
The face skin credible degree identification of point obtains the skin area in the facial image to be identified;
Face skin confidence level based on each pixel in the skin area extracts purported skin from the skin area
Region, and the target skin area extracted is transformed into LAB color spaces;
Color similarity of the target skin area in LAB color spaces between default skin color model standard is calculated, and
The skin complexion of the facial image to be identified is identified based on the color similarity being calculated.
2. the according to the method described in claim 1, it is characterized in that, face skin credible degree identification based on each pixel
The step of obtaining the skin area in the facial image to be identified includes:
The face skin confidence level of each pixel in the facial image to be identified with the first confidence threshold value is compared, is obtained
To the corresponding comparison result of each pixel;
Corresponding skin area is extracted from the facial image to be identified according to the comparison result, wherein the skin region
The face skin confidence level of each pixel is more than first confidence threshold value in domain.
3. the according to the method described in claim 1, it is characterized in that, people of each pixel based in the skin area
Face skin confidence level is extracted the step of target skin area from the skin area and is included:
The face skin confidence level of each pixel in the skin area with the second confidence threshold value is compared respectively, is obtained
The corresponding comparison result of each pixel into the skin area;
Corresponding target skin area is extracted from the skin area according to the comparison result, wherein the purported skin
The face skin confidence level of each pixel is more than second confidence threshold value in region.
4. according to the method described in claim 1, it is characterized in that, described calculate the target skin area in LAB colors sky
Between in color similarity between default skin color model standard the step of include:
Mean value computation is carried out to tri- components of L, A, B of each pixel in the target skin area respectively, obtains the target
L * component average value, A components average value and the B component average value of skin area;
L * component, A components and B component based on each colour of skin in default skin color model standard in LAB color spaces, with the mesh
L * component average value, A components average value and the B component average value of skin area are marked, calculating acquires the target skin area and institute
State the aberration between each colour of skin in default skin color model standard;
Based on the Colorimetry between the target skin area and each colour of skin obtain the target skin area and each colour of skin it
Between color similarity.
5. according to the method described in any one in claim 1-4, which is characterized in that described based on the color phase being calculated
The step of identifying the skin complexion of the facial image to be identified like degree includes:
According to the color similarity between each colour of skin in the target skin area and the default skin color model standard, from described
The colour of skin for the similarity maximum that gets colors in the corresponding each colour of skin of skin color model standard is preset, it is described to be identified as what is identified
The skin complexion of facial image.
6. a kind of skin complexion identification device, which is characterized in that described device includes:
Skin identification module, the face skin for being calculated each pixel in facial image to be identified based on partitioning into skin model are put
Reliability, and the face skin credible degree identification based on each pixel obtains the skin area in the facial image to be identified;
Region extraction module, for the face skin confidence level based on each pixel in the skin area from the skin region
Target skin area is extracted in domain, and the target skin area extracted is transformed into LAB color spaces;
Skin color model module, for calculate the target skin area in LAB color spaces with default skin color model standard it
Between color similarity, and identify based on the color similarity being calculated the skin complexion of the facial image to be identified.
7. device according to claim 6, which is characterized in that face skin of the skin identification module based on each pixel
The mode that skin credible degree identification obtains the skin area in the facial image to be identified includes:
The face skin confidence level of each pixel in the facial image to be identified with the first confidence threshold value is compared, is obtained
To the corresponding comparison result of each pixel;
Corresponding skin area is extracted from the facial image to be identified according to the comparison result, wherein the skin region
The face skin confidence level of each pixel is more than first confidence threshold value in domain.
8. device according to claim 6, which is characterized in that the region extraction module is based in the skin area
The mode that the face skin confidence level of each pixel extracts target skin area from the skin area includes:
The face skin confidence level of each pixel in the skin area with the second confidence threshold value is compared respectively, is obtained
The corresponding comparison result of each pixel into the skin area;
Corresponding target skin area is extracted from the skin area according to the comparison result, wherein the purported skin
The face skin confidence level of each pixel is more than second confidence threshold value in region.
9. device according to claim 6, which is characterized in that the skin color model module calculates the target skin area
The mode of color similarity in LAB color spaces between default skin color model standard includes:
Mean value computation is carried out to tri- components of L, A, B of each pixel in the target skin area respectively, obtains the target
L * component average value, A components average value and the B component average value of skin area;
L * component, A components and B component based on each colour of skin in default skin color model standard in LAB color spaces, with the mesh
L * component average value, A components average value and the B component average value of skin area are marked, calculating acquires the target skin area and institute
State the aberration between each colour of skin in default skin color model standard;
Based on the Colorimetry between the target skin area and each colour of skin obtain the target skin area and each colour of skin it
Between color similarity.
10. according to the device described in any one in claim 6-9, which is characterized in that the skin color model module is based on meter
Obtained color similarity identifies that the mode of the skin complexion of the facial image to be identified includes:
According to the color similarity between each colour of skin in the target skin area and the default skin color model standard, from described
The colour of skin for the similarity maximum that gets colors in the corresponding each colour of skin of skin color model standard is preset, it is described to be identified as what is identified
The skin complexion of facial image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810053781.9A CN108269290A (en) | 2018-01-19 | 2018-01-19 | Skin complexion recognition methods and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810053781.9A CN108269290A (en) | 2018-01-19 | 2018-01-19 | Skin complexion recognition methods and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108269290A true CN108269290A (en) | 2018-07-10 |
Family
ID=62776101
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810053781.9A Pending CN108269290A (en) | 2018-01-19 | 2018-01-19 | Skin complexion recognition methods and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108269290A (en) |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108921128A (en) * | 2018-07-19 | 2018-11-30 | 厦门美图之家科技有限公司 | Cheek sensitivity flesh recognition methods and device |
CN109308456A (en) * | 2018-08-31 | 2019-02-05 | 北京字节跳动网络技术有限公司 | The information of target object determines method, apparatus, equipment and storage medium |
CN109919030A (en) * | 2019-01-31 | 2019-06-21 | 深圳和而泰数据资源与云技术有限公司 | Black eye kind identification method, device, computer equipment and storage medium |
CN110211030A (en) * | 2019-06-04 | 2019-09-06 | 北京字节跳动网络技术有限公司 | Image generating method and device |
CN110490844A (en) * | 2019-07-24 | 2019-11-22 | 广州三得医疗科技有限公司 | A kind of recognition methods, system, device and the therapeutic equipment of electromagnetic therapeutic apparatus tank print |
WO2020015148A1 (en) * | 2018-07-16 | 2020-01-23 | 华为技术有限公司 | Skin spot detection method and electronic device |
CN111210440A (en) * | 2019-12-31 | 2020-05-29 | 联想(北京)有限公司 | Identification method and device of skin object and electronic equipment |
CN111429536A (en) * | 2020-03-13 | 2020-07-17 | 深圳市雄帝科技股份有限公司 | Method, system and storage medium for toning complexion in image |
CN111476849A (en) * | 2020-04-03 | 2020-07-31 | 腾讯科技(深圳)有限公司 | Object color recognition method and device, electronic equipment and storage medium |
CN111881789A (en) * | 2020-07-14 | 2020-11-03 | 深圳数联天下智能科技有限公司 | Skin color identification method and device, computing equipment and computer storage medium |
US20210343065A1 (en) * | 2020-08-20 | 2021-11-04 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Cartoonlization processing method for image, electronic device, and storage medium |
CN114582003A (en) * | 2022-04-24 | 2022-06-03 | 慕思健康睡眠股份有限公司 | Sleep health management system based on cloud computing service |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8055067B2 (en) * | 2007-01-18 | 2011-11-08 | DigitalOptics Corporation Europe Limited | Color segmentation |
CN103745193A (en) * | 2013-12-17 | 2014-04-23 | 小米科技有限责任公司 | Skin color detection method and skin color detection device |
CN106388781A (en) * | 2016-09-29 | 2017-02-15 | 深圳可思美科技有限公司 | Method for detecting skin colors and pigmentation situation of skin |
-
2018
- 2018-01-19 CN CN201810053781.9A patent/CN108269290A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8055067B2 (en) * | 2007-01-18 | 2011-11-08 | DigitalOptics Corporation Europe Limited | Color segmentation |
CN103745193A (en) * | 2013-12-17 | 2014-04-23 | 小米科技有限责任公司 | Skin color detection method and skin color detection device |
CN106388781A (en) * | 2016-09-29 | 2017-02-15 | 深圳可思美科技有限公司 | Method for detecting skin colors and pigmentation situation of skin |
Non-Patent Citations (1)
Title |
---|
李汝勤,宋钧才,黄新林: "《纤维和纺织品测试技术(4版)》", 31 March 2015, 东华大学出版社 * |
Cited By (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2020015148A1 (en) * | 2018-07-16 | 2020-01-23 | 华为技术有限公司 | Skin spot detection method and electronic device |
CN108921128B (en) * | 2018-07-19 | 2020-09-01 | 厦门美图之家科技有限公司 | Cheek sensitive muscle identification method and device |
CN108921128A (en) * | 2018-07-19 | 2018-11-30 | 厦门美图之家科技有限公司 | Cheek sensitivity flesh recognition methods and device |
CN109308456A (en) * | 2018-08-31 | 2019-02-05 | 北京字节跳动网络技术有限公司 | The information of target object determines method, apparatus, equipment and storage medium |
CN109919030A (en) * | 2019-01-31 | 2019-06-21 | 深圳和而泰数据资源与云技术有限公司 | Black eye kind identification method, device, computer equipment and storage medium |
CN109919030B (en) * | 2019-01-31 | 2021-07-13 | 深圳和而泰数据资源与云技术有限公司 | Black eye type identification method and device, computer equipment and storage medium |
CN110211030A (en) * | 2019-06-04 | 2019-09-06 | 北京字节跳动网络技术有限公司 | Image generating method and device |
CN110211030B (en) * | 2019-06-04 | 2023-10-17 | 北京字节跳动网络技术有限公司 | Image generation method and device |
CN110490844A (en) * | 2019-07-24 | 2019-11-22 | 广州三得医疗科技有限公司 | A kind of recognition methods, system, device and the therapeutic equipment of electromagnetic therapeutic apparatus tank print |
CN111210440A (en) * | 2019-12-31 | 2020-05-29 | 联想(北京)有限公司 | Identification method and device of skin object and electronic equipment |
CN111210440B (en) * | 2019-12-31 | 2023-12-22 | 联想(北京)有限公司 | Skin object identification method and device and electronic equipment |
CN111429536A (en) * | 2020-03-13 | 2020-07-17 | 深圳市雄帝科技股份有限公司 | Method, system and storage medium for toning complexion in image |
CN111429536B (en) * | 2020-03-13 | 2023-12-19 | 深圳市雄帝科技股份有限公司 | Method, system and storage medium for toning skin color in image |
CN111476849A (en) * | 2020-04-03 | 2020-07-31 | 腾讯科技(深圳)有限公司 | Object color recognition method and device, electronic equipment and storage medium |
CN111476849B (en) * | 2020-04-03 | 2023-08-29 | 腾讯科技(深圳)有限公司 | Object color recognition method, device, electronic equipment and storage medium |
CN111881789A (en) * | 2020-07-14 | 2020-11-03 | 深圳数联天下智能科技有限公司 | Skin color identification method and device, computing equipment and computer storage medium |
US11568590B2 (en) * | 2020-08-20 | 2023-01-31 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Cartoonlization processing method for image, electronic device, and storage medium |
US20210343065A1 (en) * | 2020-08-20 | 2021-11-04 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Cartoonlization processing method for image, electronic device, and storage medium |
CN114582003B (en) * | 2022-04-24 | 2022-07-29 | 慕思健康睡眠股份有限公司 | Sleep health management system based on cloud computing service |
CN114582003A (en) * | 2022-04-24 | 2022-06-03 | 慕思健康睡眠股份有限公司 | Sleep health management system based on cloud computing service |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108269290A (en) | Skin complexion recognition methods and device | |
CN111310808B (en) | Training method and device for picture recognition model, computer system and storage medium | |
CN111027378B (en) | Pedestrian re-identification method, device, terminal and storage medium | |
CN112016464A (en) | Method and device for detecting face shielding, electronic equipment and storage medium | |
CN104915972A (en) | Image processing apparatus, image processing method and program | |
CN108681692A (en) | Increase Building recognition method in a kind of remote sensing images based on deep learning newly | |
CN110648322A (en) | Method and system for detecting abnormal cervical cells | |
CN106295591A (en) | Gender identification method based on facial image and device | |
CN112419327A (en) | Image segmentation method, system and device based on generation countermeasure network | |
CN110689043A (en) | Vehicle fine granularity identification method and device based on multiple attention mechanism | |
CN111899515A (en) | Vehicle detection system based on wisdom road edge calculates gateway | |
CN108460346A (en) | Fingerprint identification method and device | |
CN112489143A (en) | Color identification method, device, equipment and storage medium | |
CN107256543A (en) | Image processing method, device, electronic equipment and storage medium | |
CN110555339A (en) | target detection method, system, device and storage medium | |
CN110378911A (en) | Weakly supervised image, semantic dividing method based on candidate region and neighborhood classification device | |
CN105405130A (en) | Cluster-based license image highlight detection method and device | |
CN112750162A (en) | Target identification positioning method and device | |
CN105631405B (en) | Traffic video intelligent recognition background modeling method based on Multilevel Block | |
CN113177451A (en) | Training method and device of image processing model, electronic equipment and storage medium | |
CN114187515A (en) | Image segmentation method and image segmentation device | |
CN113378790A (en) | Viewpoint positioning method, apparatus, electronic device and computer-readable storage medium | |
CN110135274A (en) | A kind of people flow rate statistical method based on recognition of face | |
CN111368838A (en) | Method and device for identifying reported screenshot | |
CN115359468A (en) | Target website identification method, device, equipment and medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180710 |
|
RJ01 | Rejection of invention patent application after publication |