CN106981066B - A kind of interior face image dividing method based on the colour of skin - Google Patents
A kind of interior face image dividing method based on the colour of skin Download PDFInfo
- Publication number
- CN106981066B CN106981066B CN201710126258.XA CN201710126258A CN106981066B CN 106981066 B CN106981066 B CN 106981066B CN 201710126258 A CN201710126258 A CN 201710126258A CN 106981066 B CN106981066 B CN 106981066B
- Authority
- CN
- China
- Prior art keywords
- skin
- colour
- cut zone
- exposure mask
- constriction coefficient
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
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
- G06T2207/10024—Color image
Landscapes
- Image Analysis (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to a kind of interior face image dividing method based on the colour of skin, comprising: the conversion of RGB to YCrBr color space is carried out to face picture positive under white light, while the model of ellipse for constructing colour of skin cluster is filtered candidate image, obtains colour of skin exposure mask；Boundary rectangle is made in the candidate region that face is obtained based on colour of skin exposure mask, and generates preliminary oval cut zone and be split to original image；It constructs shrink space and determines exposure mask, and carry out logical operation with the skin distribution figure in elliptic region, obtain the non-skin pixel number of four direction remnants, and in this, as adaptive shortening coefficient；Oval cut zone is updated according to constriction coefficient；Iteration, until the number of iterations reaches within the scope of the upper limit or contraction factor arrival specification error, stopping iteration exporting target image.The present invention does not depend on any library file, and strong real-time, accuracy of identification is high, shrinks time-consuming less, and interior face region fitting is accurate.
Description
Technical field
The present invention relates to technical field of image processing, specifically a kind of interior face image dividing method based on the colour of skin.
Background technique
Currently, there is many skin detection equipment in the market, the facial image clarity of shooting is high, occupy capacity and
Bandwidth is high, brings very big pressure to the resource of system；Meanwhile when detecting all kinds of indexs, due to background and face edge picture
The interference of element distortion, the accuracy of detection can be disturbed greatly.
The skin quality detector of part mainstream carries out circle Selected Inspection manually after using shooting image such as the VISIA system in the U.S.
The mode in region is surveyed to shield to inactive area, while reducing the processing difficulty of detection algorithm, but its operation is sufficiently complex,
The positioning for requiring to carry out multiple (such as forehead, cheek) regions dozens of point after shooting every time is adjusted, in user or behaviour
User experience is significantly reduced from the perspective of work person.
Therefore, in view of above series of the problem of, for the automation of skin quality testing process, the simplification of detection algorithm,
Image is propagated and the lightweight of storage, the invention proposes a kind of interior face image dividing method based on the colour of skin.
Summary of the invention
The present invention provides a kind of interior face image based on the colour of skin point to overcome drawbacks described above existing in the prior art
Segmentation method is carried out the inactive pixels of smooth segmentation, shielding face week and background to face part in face based on area of skin color, reduced
Capacity shared by image reduces algorithm detection difficulty.
To solve the above problems, the interior face image dividing method proposed by the present invention based on the colour of skin, comprising the following steps:
Step 1: carrying out the conversion of RGB to YCrBr color space to face picture positive under white light, while constructing colour of skin cluster
Model of ellipse candidate image is filtered, obtain colour of skin exposure mask；
Step 2: obtaining the candidate region of face based on colour of skin exposure mask, make boundary rectangle, and generates preliminary oval segmentation
Region is split original image, by the logical operation of colour of skin exposure mask and oval cut zone, obtains in oval cut zone
Skin distribution figure, and calculate the (skin surface in skin accounting=ellipse cut zone of the skin accounting in oval cut zone
Product/oval cut zone the gross area), using it as one of end determination flag of iteration；
Step 3: sentencing respectively in four top of oval cut zone, bottom, left part, right part placement configurations shrink spaces
Determine exposure mask, and carry out logical operation with the skin distribution figure in oval cut zone, obtains the non-skin picture of four direction remnants
Vegetarian refreshments number determines the contraction speed of four direction using the non-skin pixel number of four direction remnants as adaptive shortening coefficient
Rate；
Step 4: adjusting elliptical central point and transverse and longitudinal axis radius according to the constriction coefficient of four direction, oval point is updated
Cut region；
Step 5: step 3 and step 4 are repeated, until the number of iterations reaches the upper limit or the arrival of four direction constriction coefficient
Within the scope of specification error, stop iteration, exports target image.
In above-mentioned technical proposal, elliptical central point abscissa is determined by the difference of left and right constriction coefficient in the step 4,
Ordinate is determined by the difference for pushing up bottom constriction coefficient, and left and right, top bottom constriction coefficient is made to tend to be equal respectively after update；Elliptical horizontal axis
Radius is determined that longitudinal axis radius is determined by pushing up bottom constriction coefficient mean value by left and right constriction coefficient mean value, and transverse and longitudinal axis radius is made after update
Shorten, inside face is shunk.
The principle of interior face image dividing method proposed by the present invention based on the colour of skin is as follows:
Segmented shape: in view of the interior face shape of face of people and effective coverage, this method uses ellipse for main segmented shape.It is ellipse
Circular top part is face hair line, bottom in view of the pad interference below capture apparatus, segmentation to lower lip and chin lower edge it
Between, specific location is related with the illumination brightness of chin, and the right and left does not occur subject to ear on the inside of two ears.Elliptical shape
Changed by central point and transverse and longitudinal radius, by calculating adjustment elliptic region to optimum range.
Skin cluster: the colour of skin of people can tend to a lesser region after cluster in different color spaces.There is reality
It tests and shows that distribution of the colour of skin in YCrBr color space is similar to an ellipse, therefore can be with the region threshold in the color space
Value to the skin area in image extracts calculating.
Shrink at edge: the cut zone obtained due to relying on Face Detection for the first time cannot be fitted face edge well,
Exist and do not reject complete background residual, therefore this method increases the shrink space detection of four direction, and carries out adaptive
The contraction iteration of stepping is to make cut zone be fitted interior face edge as far as possible.
The present invention has the advantages that compared with prior art and advantage:
1) the present invention is based on the model of ellipse of YCrBr color space to carry out Face Detection, do not depend on any library file, real
Shi Xingqiang, accuracy of identification are high；
2) present invention is fitted interior face cut zone using elliptical shape, and process successive ignition from four direction inwardly into
Row adaptive shortening shrinks time-consuming less, and interior face region fitting is accurate.
Specific embodiment
Below in conjunction with specific embodiment, the present invention is described in further detail:
In the present embodiment, the interior face image dividing method proposed by the present invention based on the colour of skin, comprising the following steps:
Step 1: carrying out the conversion of RGB to YCrBr color space to face picture positive under white light, while constructing colour of skin cluster
Model of ellipse candidate image is filtered, obtain colour of skin exposure mask, oval top is face hair line, and bottom is in view of shooting
Pad interference below equipment, divides between lower lip and chin lower edge, and specific location is related with the illumination brightness of chin, left
Right both sides do not occur subject to ear on the inside of two ears；
Step 2: obtaining the candidate region of face based on colour of skin exposure mask, make boundary rectangle, and generates preliminary oval segmentation
Region is split original image, by the logical operation of colour of skin exposure mask and oval cut zone, obtains in oval cut zone
Skin distribution figure, and calculate the (skin surface in skin accounting=ellipse cut zone of the skin accounting in oval cut zone
Product/oval cut zone the gross area), using it as one of end determination flag of iteration；
Step 3: being covered respectively in four top of elliptic region, bottom, left part, right part placement configurations shrink space judgements
Film, and logical operation is carried out with the skin distribution figure in oval cut zone, obtain the non-skin pixel of four direction remnants
Number, the contraction rate of four direction is determined using the non-skin pixel number of four direction remnants as adaptive shortening coefficient；
Step 4: adjusting elliptical central point and transverse and longitudinal axis radius according to the constriction coefficient of four direction, oval point is updated
Region is cut, wherein elliptical central point abscissa is determined by the difference of left and right constriction coefficient, ordinate is by pushing up the difference of bottom constriction coefficient
It determines, left and right, top bottom constriction coefficient is made to tend to be equal respectively after update；Elliptical horizontal axis radius is determined by left and right constriction coefficient mean value
Fixed, longitudinal axis radius is determined by pushing up bottom constriction coefficient mean value, shortens transverse and longitudinal axis radius after update, and inside face is shunk；
Step 5: step 3 and step 4 are repeated, until the number of iterations reaches the upper limit or the arrival of four direction constriction coefficient
Within the scope of specification error, stop iteration, exports target image.
Finally, it is stated that the above examples are only used to illustrate the technical scheme of the present invention and are not limiting, although referring to compared with
Good embodiment describes the invention in detail, those skilled in the art should understand that, it can be to skill of the invention
Art scheme is modified or replaced equivalently, and without departing from the objective and range of technical solution of the present invention, should all be covered at this
In the scope of the claims of invention.
Claims (2)
1. a kind of interior face image dividing method based on the colour of skin, which comprises the following steps:
Step 1: carrying out the conversion of RGB to YCrBr color space to face picture positive under white light, while constructing the ellipse of colour of skin cluster
Circle model is filtered candidate image, obtains colour of skin exposure mask；
Step 2: obtaining the candidate region of face based on colour of skin exposure mask, make boundary rectangle, and join according to the center of rectangle and length and width
Number generates preliminary oval cut zone and is split to original image, is transported by the logical AND of colour of skin exposure mask and oval cut zone
It calculates, obtains the skin distribution figure in oval cut zone, and calculate the skin accounting in oval cut zone, change as algorithm
In generation, terminates one of Judging index, i.e., when skin accounting reaches preset threshold value, expression meets segmentation demand, exits algorithm iteration
Process, skin area/ellipse cut zone gross area in the skin accounting=ellipse cut zone；
Step 3: respectively in 1/3 region of the top of oval cut zone, 1/3 region of bottom, 1/6 region of left part, 1/6th area of right part
Four, domain placement configurations shrink space determines exposure mask, and carries out logical operation with the skin distribution figure in oval cut zone, obtains
To the non-skin pixel number of four direction remnants, using the non-skin pixel number of four direction remnants as adaptive shortening system
Count the contraction rate to determine four direction；
Step 4: adjusting elliptical central point and transverse and longitudinal axis radius according to the constriction coefficient of four direction, oval cut section is updated
Domain；
Step 5: step 3 and step 4 are repeated, until the constriction coefficient arrival that the number of iterations reaches the upper limit or four direction refers to
Determine in error range, stop iteration, exports target image.
2. the interior face image dividing method according to claim 1 based on the colour of skin, which is characterized in that ellipse in the step 4
Round central point abscissa is determined that ordinate is determined by the difference for pushing up bottom constriction coefficient by the difference of left and right constriction coefficient, is made after update
Left and right, top bottom constriction coefficient tend to be equal respectively；Elliptical horizontal axis radius determines by left and right constriction coefficient mean value, longitudinal axis radius by
It pushes up bottom constriction coefficient mean value to determine, shortens transverse and longitudinal axis radius after update, inside face is shunk.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710126258.XA CN106981066B (en) | 2017-03-06 | 2017-03-06 | A kind of interior face image dividing method based on the colour of skin |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710126258.XA CN106981066B (en) | 2017-03-06 | 2017-03-06 | A kind of interior face image dividing method based on the colour of skin |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106981066A CN106981066A (en) | 2017-07-25 |
| CN106981066B true CN106981066B (en) | 2019-07-12 |
Family
ID=59338658
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710126258.XA Active CN106981066B (en) | 2017-03-06 | 2017-03-06 | A kind of interior face image dividing method based on the colour of skin |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106981066B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110378327B (en) * | 2019-07-09 | 2021-05-18 | 浙江大学 | Target detection device and method with auxiliary significant features added |
| CN111259806B (en) * | 2020-01-16 | 2023-11-14 | 广州杰赛科技股份有限公司 | Face area identification method, device and storage medium |
| CN111507944B (en) * | 2020-03-31 | 2023-07-04 | 北京百度网讯科技有限公司 | Determination method and device for skin smoothness and electronic equipment |
| CN112381046B (en) * | 2020-11-30 | 2023-02-14 | 华南理工大学 | Multitask posture-invariant face recognition method, system, device and storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2010102725A (en) * | 2002-08-22 | 2010-05-06 | Eastman Kodak Co | Method for detecting object in digital image |
| WO2010082942A9 (en) * | 2008-02-01 | 2011-01-20 | Canfield Scientific, Incorporated | Automatic mask design and registration and feature detection for computer-aided skin analysis |
| CN102270303A (en) * | 2011-07-27 | 2011-12-07 | 重庆大学 | Joint detection method for sensitive image |
| CN104715244A (en) * | 2015-04-01 | 2015-06-17 | 华中科技大学 | Multi-viewing-angle face detection method based on skin color segmentation and machine learning |
| CN105205480A (en) * | 2015-10-31 | 2015-12-30 | 潍坊学院 | Complex scene human eye locating method and system |
| CN105224917A (en) * | 2015-09-10 | 2016-01-06 | 成都品果科技有限公司 | A kind of method and system utilizing color space to create skin color probability map |
| CN105809089A (en) * | 2014-12-29 | 2016-07-27 | 中国科学院深圳先进技术研究院 | Multi-face detection method and device under complex background |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7116820B2 (en) * | 2003-04-28 | 2006-10-03 | Hewlett-Packard Development Company, Lp. | Detecting and correcting red-eye in a digital image |
-
2017
- 2017-03-06 CN CN201710126258.XA patent/CN106981066B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2010102725A (en) * | 2002-08-22 | 2010-05-06 | Eastman Kodak Co | Method for detecting object in digital image |
| EP1394723A3 (en) * | 2002-08-22 | 2011-04-13 | Eastman Kodak Company | Method for detecting red-eye effect in digital images |
| WO2010082942A9 (en) * | 2008-02-01 | 2011-01-20 | Canfield Scientific, Incorporated | Automatic mask design and registration and feature detection for computer-aided skin analysis |
| CN102270303A (en) * | 2011-07-27 | 2011-12-07 | 重庆大学 | Joint detection method for sensitive image |
| CN105809089A (en) * | 2014-12-29 | 2016-07-27 | 中国科学院深圳先进技术研究院 | Multi-face detection method and device under complex background |
| CN104715244A (en) * | 2015-04-01 | 2015-06-17 | 华中科技大学 | Multi-viewing-angle face detection method based on skin color segmentation and machine learning |
| CN105224917A (en) * | 2015-09-10 | 2016-01-06 | 成都品果科技有限公司 | A kind of method and system utilizing color space to create skin color probability map |
| CN105205480A (en) * | 2015-10-31 | 2015-12-30 | 潍坊学院 | Complex scene human eye locating method and system |
Non-Patent Citations (4)
| Title |
|---|
| Improved Skin Detection based on Dynamic Threshold using Multi-Colour Space;Mohd. Zamri Osman等;《IEEE》;20141231;全文 |
| Skin Segmentation Based on Human Face Illumination Feature;Pan Ng等;《ACM》;20121231;全文 |
| 一种基于肤色模型和眼睛定位的人脸检测方法;俞王新等;《计算机工程与应用》;20081231;第44卷(第26期);全文 |
| 基于肤色和器官定位的实时人脸检测;黄禹馨等;《计算机工程与科学》;20140531;第36卷(第5期);全文 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106981066A (en) | 2017-07-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106981066B (en) | A kind of interior face image dividing method based on the colour of skin | |
| CN107506693B (en) | Distort face image correcting method, device, computer equipment and storage medium | |
| CN106657946B (en) | System and method are eliminated in image purple boundary | |
| CN103914699B (en) | A kind of method of the image enhaucament of the automatic lip gloss based on color space | |
| CN109636732A (en) | A kind of empty restorative procedure and image processing apparatus of depth image | |
| US20080278516A1 (en) | System and method for adjusting perceived eye rotation in image of face | |
| CN108876797B (en) | Image segmentation system and method based on Spiking-SOM neural network clustering | |
| WO2019165604A1 (en) | Hairline contour locating method and device, and computer readable storage medium | |
| CN115409745B (en) | CT image enhancement method applied to radiotherapy preparation | |
| CN105045968B (en) | A kind of hair styling method and system | |
| CN111340734B (en) | Image purple boundary correction method using convolutional neural network model | |
| CN102982524B (en) | Splicing method for corn ear order images | |
| CN109857255A (en) | A kind of display parameter regulation method, device and wear display equipment | |
| CN108710832B (en) | Reference-free iris image definition detection method | |
| TWI810595B (en) | Parameter adjustment method, device and equipment of vibrating mirror | |
| WO2020216810A1 (en) | Method and device for measuring the local refractive power and/or the refractive power distribution of a spectacle lens | |
| CN110390657A (en) | A kind of image interfusion method | |
| CN109542145B (en) | Automatic light path collimation method of high-power laser device | |
| CN109145865A (en) | Face standard level calculating method and device | |
| CN110335332B (en) | Automatic drawing method of human face cartoon | |
| CN109461123B (en) | A kind of geometric distortion automatic analysis method for picture quality objective evaluating point diagram | |
| CN114333073B (en) | Intelligent table lamp auxiliary adjusting method and system based on visual perception | |
| CN113567058B (en) | Light source parameter adjusting method based on artificial intelligence and visual perception | |
| US10257405B2 (en) | Automatic focusing method and automatic focusing system | |
| CN111784760B (en) | Method for correcting radar machine learning extrapolation result by radar linear optical flow extrapolation result |
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 |