CN109961426A - A kind of detection method of face skin skin quality - Google Patents
A kind of detection method of face skin skin quality Download PDFInfo
- Publication number
- CN109961426A CN109961426A CN201910181670.0A CN201910181670A CN109961426A CN 109961426 A CN109961426 A CN 109961426A CN 201910181670 A CN201910181670 A CN 201910181670A CN 109961426 A CN109961426 A CN 109961426A
- Authority
- CN
- China
- Prior art keywords
- small image
- image block
- skin
- colour
- face
- 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
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- 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
- G06T7/00—Image analysis
- G06T7/40—Analysis of texture
- G06T7/41—Analysis of texture based on statistical description of texture
-
- 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
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
- G06T2207/30004—Biomedical image processing
- G06T2207/30088—Skin; Dermal
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of detection methods of face skin skin quality.The realization step of method of the invention are as follows: (1) establish face dermatological specimens library；(2) training sample set and sample to be identified are obtained；(3) textural characteristics and RGB color multimedia message of colored small image block are extracted；(4) VGG16 classifier is trained using training sample set；(5) the small image block of colour concentrated to the sample identified is classified；(6) ratio that all kinds of classifications account for colored skin image is counted.The present invention improve to round skin problem complicated and changeable for example acne, spot, mole, pore detection accuracy, the severity of all kinds of skin problems can be analyzed, improve face skin skin quality analysis reliability.The present invention can be transplanted to mobile phone mobile terminal, the convenient intuitive skin condition for understanding skin.
Description
Technical field
The invention belongs to technical field of image processing, further relate to one of technical field of image detection face skin
The detection method of skin skin quality.The present invention can be contained by the pore in detection face skin image, acne, spot, wrinkle, mole
Amount, it is comprehensive to provide the analysis of skin skin quality as a result, achieving the purpose that intuitively to understand skin skin condition.
Background technique
In recent years, with the improvement of living standards, the rapid development of medical cosmetology, people increasingly start to focus on itself skin
The health status of skin.When being nursed to skin, it is necessary first to reasonably be evaluated skin of face.Face skin problem
It mainly include wrinkle, spot, acne, pore, mole etc., these skin problems are that people's concern is most, are medically had bright
True standard of perfection, can be by visually judging its type, but its severity is larger by subjective impact.It is both domestic and external at present
Face skin skin quality testing product is broadly divided into two classes: 1, the metallic test formula skin quality detection based on bio-electrical impedance measuring method
Instrument, such product price is cheap but detection function is single, precision is low；2, the mechanical detection instrument based on image procossing, this product function
Can be perfect, the disadvantage is that expensive, bulky, is not suitable for carrying.It to sum up analyzes, is evaluated using iconology analytic approach
Skin skin condition has feasibility.
A kind of patent document " facial skin quality evaluation method " (number of patent application of Northeastern University in its application
201810698035.5 publication number CN108932493 A) in disclose a kind of facial skin quality evaluation method.This method is first
Facial image is first acquired, gray processing is carried out to facial image and obtains face coordinate, and facial image is split；Then it uses
Uniform brightness, Gamma antidote are pre-processed, and the contrast of image is improved, special by LBP feature extraction dermatoglyph
Sign；Finally classified using support vector machines classifier to skin, facial ratio shared by statistics acne, wrinkle is divided
Analyse result.Although this method can acne to face and wrinkle problem detect, the deficiency that this method still has
Place is: this method is only extracted the textural characteristics of skin, lacks the extraction to color character, causes for circle complicated and changeable
The more difficult resolution of shape skin problem, such as spot, acne, pore, mole.
A kind of patent document " skin quality and skin problem recognition detection side based on facial image identification of the Wu Liang in its application
A kind of face skin skin quality and skin are disclosed in method " (number of patent application 201410537110.1, publication number CN104299011 A)
Skin problem identification detection method.This method inputs human face photo first and carries out recognition of face, and the mug shot of face is divided into
20 face-image blocks are based on local auto-adaptive threshold method to block facial each of after division and connected domain analysis method carry out
Hair and skin identification；Then the colour of skin and skin greasing degree are calculated in Lab color space, based on algorithm of co-matrix
The smooth degree value and skin problem of skin are calculated, skin problem includes small pox, red capillary, mole, freckle, coarse six class of pore；
Finally by support vector machines classifier in facial image block skin attribute and skin problem classify, according to
Classification type obtains the analysis result of skin quality and skin problem.Although this method can to face skin complexion, greasy degree with
And skin problem carries out recognition detection, but the shortcoming that this method still has is: face skin problem is complex, does not have
Have and the severity of skin problem is detected, only detects whether to lack flexibility and adaptability whether there is or not skin problem.
Summary of the invention
The purpose of the present invention is in view of the above shortcomings of the prior art, propose a kind of detection method of face skin skin quality.
The present invention improve to round skin problem complicated and changeable for example acne, spot, mole, pore detection accuracy, can
The severity of all kinds of skin problems is analyzed, the reliability of face skin skin quality analysis is improved.
Realizing the concrete thought of the object of the invention is: due to when dividing true value to whole face skin image, be easy by
It is influenced to subjective factor, it is difficult to embody the severity of skin problem, therefore first divide equally collected face skin image
At colored small image block, later use classifier is trained and detects to colored small image block, finally by all kinds of classes of statistics
The ratio of colored skin image is not accounted for, realizes the analysis to face skin image.In characteristic extraction procedure, by colored small image
Block is transformed into HSV color space from rgb color space, and therefrom isolates channel S, by carrying out wavelet decomposition to channel S, obtains
To the low frequency component of the small image block of channel S, the purpose of the contrast of the enhancing small image block of channel S is realized.Using side between maximum kind
Poor method to the small image block of low frequency carry out binaryzation, obtain the textural characteristics of skin image, by after binaryzation skin image with it is right
It answers RGB color image to be multiplied, obtains the RGB color multimedia message of corresponding texture region.In trained and detection-phase, VGG16 is used
Classifier is trained the texture extracted with color character, with trained VGG16 classifier to colored small image block
It is detected.
Step of the invention is as follows:
(1) face dermatological specimens library is established:
(1a) acquires the colored skin image of at least 100 people by high-definition camera, everyone acquires 5 portions of face
Image is opened in position, the acquisition 10 at each position；
Every colored skin image is divided into 100 small image blocks of colour by (1b)；
Skin condition is divided into 6 classes by (1c), and each small image block of colour corresponds to a kind of skin condition；
(1d) traverses each small image block of colour, determines that each small image block of colour corresponds to the type of skin condition；
All colored small image blocks and the corresponding type of each small image block of colour are formed face dermatological specimens by (1e)
Library；
(2) training sample set and sample to be identified are obtained:
(2a) randomly selects 70% colored small image block and its corresponding type composition instruction from face dermatological specimens library
Practice sample set；
The remaining institute small image block of chromatic colour in face dermatological specimens library is formed sample set to be identified by (2b)；
(3) textural characteristics and RGB color multimedia message of colored small image block are extracted:
(3a) utilizes conversion formula, and each small image block of colour is transformed into HSV color space from rgb color space, and
Channel S is isolated from HSV color space；
(3b) utilizes the wavelet decomposition formula of low-frequency image, carries out wavelet decomposition to channel S, obtains each small image of colour
The small image block of the corresponding low frequency of block；
(3c) utilizes maximum between-cluster variance formula, calculates the maximum between-cluster variance of the small image block of each low frequency；
(3d) is using the corresponding segmentation threshold of maximum between-cluster variance as optimal threshold；
(3e) judges whether the optimal threshold of the small image block of each low frequency is greater than the average gray value of the small image block of low frequency, if
It is then to determine to contain hair in the small image block of the low frequency, executes step (3f) and otherwise then determine in the small image block of the low frequency not
Contain hair；
After the optimal threshold of the small image block of each low frequency is added 40 by (3f), updated optimal threshold is obtained, update is used
Optimal threshold afterwards carries out binary conversion treatment to the small image block containing crinite low frequency, obtains the two-value of removal skin and hair interference
Small image block；
(3g) carries out binary conversion treatment without the small image block of crinite low frequency to each；
(3h) by the distribution of white pixel point in the small image block of each two-value, as the corresponding small image block of colour
Textural characteristics；
The small image block of the corresponding RGB color of small image block after each binaryzation is multiplied by (3i), obtains colored small
The RGB color multimedia message of image block；
(4) training VGG16 classifier:
(4a) is using the VGG16 disaggregated model of not top layer as classifier；
(4b) by training sample concentrate all small image blocks of colour textural characteristics and RGB color multimedia message and each colour it is small
The corresponding type of image block, while being input in classifier and being trained, obtain trained VGG16 classifier；
(5) the small image block of colour concentrated to the sample identified is classified:
The textural characteristics of the small image block of colour each in sample set to be identified and RGB color multimedia message are input to training simultaneously
Classify in good VGG16 classifier, obtains the classification results of each small image block of colour；
(6) ratio that all kinds of classifications account for colored skin image is counted:
The quantity for counting the small image block of colour under all kinds of classifications accounts for the ratio of colored small image block sum.
The present invention has the advantage that compared with prior art
First, the present invention is multiplied using the small image block of the corresponding RGB color of the small image block after binaryzation, obtains coloured silk
The RGB color multimedia message of the small image block of color, and the feature extracted is instructed using the VGG16 disaggregated model of convolutional neural networks
Practice, overcome cause in the prior art due to lacking to the extraction of color character for round skin problem complicated and changeable compared with
Indistinguishable disadvantage enables the invention to detection spot, acne, pore, mole.
Second, the present invention is divided into 100 small image blocks of colour using by every colored skin image, counts all kinds of classifications
Under the quantity of the small image block of colour account for the ratio of colored small image block sum, overcome and lack in the prior art to skin problem
The shortcomings that detection of severity, enables the invention to the skin condition that skin is intuitively presented.
Detailed description of the invention
Fig. 1 is flow chart of the invention；
Acquisition face skin area sample schematic diagram Fig. 2 of the invention；
The textural characteristics and rgb color information flow chart of the colored small image block of extraction Fig. 3 of the invention.
Specific implementation measure
Invention is described further with reference to the accompanying drawing.
It is described as follows in conjunction with specific steps of the attached drawing 1 to the method for the present invention:
Step 1, face dermatological specimens library is established.
The colored skin image of at least 100 people is acquired by high-definition camera.
In conjunction with the process that attached drawing 2 is acquired 5 positions in everyone face other than human face five-sense-organ do into
The description of one step.
Fig. 2 (a) is the overall schematic for acquiring 5 positions of face, and Fig. 2 (b) is that forehead, Fig. 2 (c) of acquisition face are
Acquire position on the upside of the right side cheek of face, Fig. 2 (d) is the left cheek upside position for acquiring face, Fig. 2 (e) is acquisition people
Position, Fig. 2 (f) are to acquire position on the downside of the left cheek of face on the downside of the right side cheek of face.
Image is opened in the acquisition 10 at each position, and collection process keeps light source consistent with physical distance, and camera is kept to stablize.
Every colored skin image is divided into 100 small image blocks of colour, and skin condition is divided into 6 classes: normal skin
Skin, pore be coarse, acne, spot, wrinkle, mole, and each small image block of colour corresponds to a kind of skin condition, and it is small to traverse each colour
Image block determines that each small image block of colour corresponds to the type of skin condition.
By all colored small image blocks and the corresponding type of each small image block of colour, face dermatological specimens library is formed.
Step 2, training sample set and sample to be identified are obtained.
From face dermatological specimens library, randomly selects 70% colored small image block and its corresponding type forms training sample
This collection, the remaining small image block of institute's chromatic colour form sample set to be identified.
Step 3, the textural characteristics and RGB color multimedia message of colored small image block are extracted.
In conjunction with attached drawing 3, to the textural characteristics of the colored small image block of extraction and the process of RGB color multimedia message of the invention do into
The description of one step.
Step 1: the small image block of input color.
Step 2: by conversion formula, being transformed into HSV color space from rgb color space for each small image block of colour,
And channel S is isolated from HSV color space.
Wherein, S (i, j) indicate abscissa be i, the saturation degree channel numerical value for the HSV image slices vegetarian refreshments that ordinate is j, max
Expression is maximized operation, r, g, and b respectively indicates the numerical value in three channels of red, green, blue of RGB color, r, g, b ∈ [0,
1 ..., 255], min expression is minimized operation.
Step 3: by the wavelet decomposition formula of low-frequency image, wavelet decomposition is carried out to channel S, obtains each small figure of colour
As the corresponding small image block of low frequency of block.
Dj=LrLcCj
Wherein, DjLow-frequency image after indicating jth time wavelet decomposition, L indicate one-dimensional low pass mirror image wavelet filtering operator, r
The row and column of channel S image, C are respectively indicated with cjChannel S image before indicating jth time wavelet decomposition.
Step 4: by maximum between-cluster variance formula, the maximum between-cluster variance of the small image block of each low frequency is calculated, and will most
The big corresponding segmentation threshold of inter-class variance is as optimal threshold.
Wherein, T indicates the segmentation threshold of image, segmentation threshold T being divided into the small image block of low frequency by image pixel size
Target and background two parts, gTIndicate that segmentation threshold is the inter-class variance of the small image block of low frequency of T, w0Indicate the small image block of low frequency
Object pixel account for the ratio of the small image block sum of all pixels of low frequency, u0Indicate that the average gray of target pixel points, u indicate that low frequency is small
The average gray of image block.
Step 5: judging whether the optimal threshold of the small image block of each low frequency is greater than the average gray of the small image block of low frequency, if
It is then to determine to contain hair in the small image block of the low frequency, executes step 6, otherwise, it is determined that is free of hairiness in the small image block of the low frequency
Hair executes step 7.
Step 6: the optimal threshold of the small image block of each low frequency is added 40, updated optimal threshold is obtained, uses update
Optimal threshold afterwards carries out binary conversion treatment to the small image block containing crinite low frequency, obtains the two-value of removal skin and hair interference
Small image block.
Step 7: binary conversion treatment is carried out to the small image block of each low frequency.
Step 8: the small image block of the corresponding RGB color of the small image block after each binaryzation is multiplied, colour is obtained
The RGB color multimedia message of small image block.
Small image block after binary conversion treatment is made of two kinds of colors of black and white, and the pixel value of black pixel point is 0,
The pixel value of white pixel point is 1, and the pixel value value range of the small image block of RGB color is 0 to 255.At each binaryzation
The small image block of the corresponding RGB color of small image block after reason is multiplied, and black pixel point and RGB are color in the small image block of binaryzation
Corresponding color pixel cell is multiplied to obtain black pixel point in the small image block of color, in the small image block of binaryzation white pixel point with
Corresponding color pixel cell is multiplied to obtain color pixel cell in the small image block of RGB color.
Step 4, VGG16 classifier is trained using training sample set.
Using the VGG16 disaggregated model of not top layer as classifier.The network structure of classifier is as shown in table 1.It will train
The textural characteristics of the small image block of institute's chromatic colour and RGB color multimedia message type corresponding with each small image block of colour in sample set,
It is input in classifier and is trained simultaneously, obtain trained VGG16 classifier.
Step 5, the small image block of colour concentrated to the sample identified is classified.
The textural characteristics of the small image block of colour each in sample set to be identified and RGB color multimedia message are input to training simultaneously
Classify in good VGG16 classifier, obtains the classification results of each small image block of colour.
Step 6, the ratio that all kinds of classifications account for colored skin image is counted.
The quantity for counting the small image block of colour under all kinds of classifications accounts for the ratio of colored small image block sum, each by analyzing
Ratio shared by class skin condition carries out comprehensive analysis to face skin skin quality, and normal skin state proportion is higher, explanation
Face skin skin quality is better, and classification shared by other class skin conditions is higher, more, illustrates that face skin problem is more serious.
The VGG16 network structure table of the not no top layer of table 1
| Network layer | Intrinsic dimensionality |
| Input layer (100 × 100RGB color image) | |
| Convolutional layer Block1Conv1 | 100×100×64 |
| Convolutional layer Block1Conv2 | 100×100×64 |
| Pond layer BlockPool | 50×50×64 |
| Convolutional layer Block2Conv1 | 50×50×128 |
| Convolutional layer Block2Conv2 | 50×50×128 |
| Pond layer Block2Pool | 25×25×128 |
| Convolutional layer Block3Conv1 | 25×25×256 |
| Convolutional layer Block3Conv2 | 25×25×256 |
| Convolutional layer Block3Conv3 | 25×25×256 |
| Pond layer Block3Pool | 13×13×256 |
| Convolutional layer Block4Conv1 | 13×13×256 |
| Convolutional layer Block4Conv2 | 13×13×256 |
| Convolutional layer Block4Conv3 | 13×13×256 |
| Pond layer Block4Pool | 7×7×512 |
| Convolutional layer Block5Conv1 | 7×7×512 |
| Convolutional layer Block5Conv2 | 7×7×512 |
| Convolutional layer Block5Conv3 | 7×7×512 |
| Pond layer Block5Pool | 4×4×512 |
Claims (6)
1. a kind of detection method of face skin skin quality, which is characterized in that extract the textural characteristics and RGB color of colored small image block
Multimedia message is trained VGG16 classifier using training sample set, and the specific steps of this method include the following:
(1) face dermatological specimens library is established:
(1a) acquires the colored skin image of at least 100 people by high-definition camera, everyone acquires 5 positions of face,
Image is opened in the acquisition 10 at each position；
Every colored skin image is divided into 100 small image blocks of colour by (1b)；
Skin condition is divided into 6 classes by (1c), and each small image block of colour corresponds to a kind of skin condition；
(1d) traverses each small image block of colour, determines that each small image block of colour corresponds to the type of skin condition；
All colored small image blocks and the corresponding type of each small image block of colour are formed face dermatological specimens library by (1e)；
(2) training sample set and sample to be identified are obtained:
(2a) randomly selects 70% colored small image block and its corresponding type forms training sample from face dermatological specimens library
This collection；
The remaining institute small image block of chromatic colour in face dermatological specimens library is formed sample set to be identified by (2b)；
(3) textural characteristics and RGB color multimedia message of colored small image block are extracted:
(3a) utilizes conversion formula, and each small image block of colour is transformed into HSV color space from rgb color space, and from HSV
Channel S is isolated in color space；
(3b) utilizes the wavelet decomposition formula of low-frequency image, carries out wavelet decomposition to channel S, obtains each small image block pair of colour
The small image block of the low frequency answered；
(3c) utilizes maximum between-cluster variance formula, calculates the maximum between-cluster variance of the small image block of each low frequency；
(3d) is using the corresponding segmentation threshold of maximum between-cluster variance as optimal threshold；
(3e) judges whether the optimal threshold of the small image block of each low frequency is greater than the average gray value of the small image block of low frequency, if so,
Then determine to contain hair in the small image block of the low frequency, executes step (3f) and otherwise then determine not containing in the small image block of the low frequency
Hair；
After the optimal threshold of the small image block of each low frequency is added 40 by (3f), updated optimal threshold is obtained, use is updated
Optimal threshold carries out binary conversion treatment to the small image block containing crinite low frequency, obtains the small figure of two-value of removal skin and hair interference
As block；
(3g) carries out binary conversion treatment without the small image block of crinite low frequency to each；
The texture of (3h) by the distribution of white pixel point in the small image block of each two-value, as the corresponding small image block of colour
Feature；
The small image block of the corresponding RGB color of small image block after each binaryzation is multiplied by (3i), obtains colored small image
The RGB color multimedia message of block；
(4) training VGG16 classifier:
(4a) is using the VGG16 disaggregated model of not top layer as classifier；
Training sample is concentrated the textural characteristics and RGB color multimedia message and each small image of colour of all small image blocks of colour by (4b)
The corresponding type of block, while being input in classifier and being trained, obtain trained VGG16 classifier；
(5) the small image block of colour concentrated to the sample identified is classified:
The textural characteristics of the small image block of colour each in sample set to be identified and RGB color multimedia message are input to simultaneously trained
Classify in VGG16 classifier, obtains the classification results of each small image block of colour；
(6) ratio that all kinds of classifications account for colored skin image is counted:
The quantity for counting the small image block of colour under all kinds of classifications accounts for the ratio of colored small image block sum.
2. a kind of detection method of face skin skin quality according to claim 1, which is characterized in that described in step (1a)
5 positions of face refer to: position above and below the forehead, left cheek other than human face five-sense-organ, position above and below the cheek of right side.
3. a kind of detection method of face skin skin quality according to claim 1, which is characterized in that described in step (1c)
Skin condition is divided into 6 classes to refer to: normal skin, pore be coarse, acne, spot, wrinkle, mole.
4. a kind of detection method of face skin skin quality according to claim 1, which is characterized in that described in step (3a)
Conversion formula it is as follows:
Wherein, S (i, j) indicate abscissa be i, the saturation degree channel numerical value for the HSV image slices vegetarian refreshments that ordinate is j, max indicate
It is maximized operation, r, g, b respectively indicates the numerical value in three channels of red, green, blue of RGB color, r, g, b ∈ [0,
1 ..., 255], min expression is minimized operation.
5. a kind of detection method of face skin skin quality according to claim 1, which is characterized in that described in step (3b)
The wavelet decomposition formula of low-frequency image is as follows:
Dj=LrLcCj
Wherein, DjLow-frequency image after indicating jth time wavelet decomposition, L indicate that one-dimensional low pass mirror image wavelet filtering operator, r and c divide
Not Biao Shi channel S image row and column, CjChannel S image before indicating jth time wavelet decomposition.
6. a kind of detection method of face skin skin quality according to claim 1, which is characterized in that described in step (3c)
Maximum between-cluster variance formula is as follows:
Wherein, T indicate image segmentation threshold, segmentation threshold T by image pixel size by the small image block of low frequency be divided into target and
Background two parts, gTIndicate that segmentation threshold is the inter-class variance of the small image block of low frequency of T, w0Indicate the target of the small image block of low frequency
Pixel accounts for the ratio of the small image block sum of all pixels of low frequency, u0Indicate that the average gray of target pixel points, u indicate the small image block of low frequency
Average gray.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910181670.0A CN109961426B (en) | 2019-03-11 | 2019-03-11 | Method for detecting skin of human face |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910181670.0A CN109961426B (en) | 2019-03-11 | 2019-03-11 | Method for detecting skin of human face |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109961426A true CN109961426A (en) | 2019-07-02 |
| CN109961426B CN109961426B (en) | 2021-07-06 |
Family
ID=67024131
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910181670.0A Active CN109961426B (en) | 2019-03-11 | 2019-03-11 | Method for detecting skin of human face |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109961426B (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110956623A (en) * | 2019-11-29 | 2020-04-03 | 深圳和而泰家居在线网络科技有限公司 | Wrinkle detection method, apparatus, device, and computer-readable storage medium |
| CN111401463A (en) * | 2020-03-25 | 2020-07-10 | 维沃移动通信有限公司 | Method for outputting detection result, electronic device, and medium |
| CN112053344A (en) * | 2020-09-02 | 2020-12-08 | 杨洋 | Skin detection method system and equipment based on big data algorithm |
| CN112396573A (en) * | 2019-07-30 | 2021-02-23 | 纵横在线(广州)网络科技有限公司 | Facial skin analysis method and system based on image recognition |
| CN112837304A (en) * | 2021-02-10 | 2021-05-25 | 姜京池 | Skin detection method, computer storage medium and computing device |
| CN113128375A (en) * | 2021-04-02 | 2021-07-16 | 西安融智芙科技有限责任公司 | Image recognition method, electronic device and computer-readable storage medium |
| CN113129250A (en) * | 2019-12-27 | 2021-07-16 | 华为技术有限公司 | Skin detection method and device, terminal equipment and computer storage medium |
| CN113554623A (en) * | 2021-07-23 | 2021-10-26 | 江苏医像信息技术有限公司 | Intelligent quantitative analysis method and system for human face skin |
| CN113723310A (en) * | 2021-08-31 | 2021-11-30 | 平安科技（深圳）有限公司 | Image identification method based on neural network and related device |
| CN115119897A (en) * | 2022-06-17 | 2022-09-30 | 上海食未生物科技有限公司 | 3D printing meat printing method and system |
| CN116993714A (en) * | 2023-08-30 | 2023-11-03 | 深圳伯德睿捷健康科技有限公司 | Skin detection method, system and computer readable storage medium |
| WO2024014853A1 (en) * | 2022-07-13 | 2024-01-18 | 주식회사 룰루랩 | Method and device for detecting facial wrinkles using deep learning-based wrinkle detection model trained according to semi-automatic labeling |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106407915A (en) * | 2016-08-31 | 2017-02-15 | 广州精点计算机科技有限公司 | SVM (support vector machine)-based face recognition method and device |
| US20170287134A1 (en) * | 2016-03-31 | 2017-10-05 | International Business Machines Corporation | Annotation of skin image using learned feature |
| CN107680128A (en) * | 2017-10-31 | 2018-02-09 | 广东欧珀移动通信有限公司 | Image processing method, device, electronic equipment and computer-readable recording medium |
| CN107862695A (en) * | 2017-12-06 | 2018-03-30 | 电子科技大学 | A kind of modified image segmentation training method based on full convolutional neural networks |
-
2019
- 2019-03-11 CN CN201910181670.0A patent/CN109961426B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170287134A1 (en) * | 2016-03-31 | 2017-10-05 | International Business Machines Corporation | Annotation of skin image using learned feature |
| CN106407915A (en) * | 2016-08-31 | 2017-02-15 | 广州精点计算机科技有限公司 | SVM (support vector machine)-based face recognition method and device |
| CN107680128A (en) * | 2017-10-31 | 2018-02-09 | 广东欧珀移动通信有限公司 | Image processing method, device, electronic equipment and computer-readable recording medium |
| CN107862695A (en) * | 2017-12-06 | 2018-03-30 | 电子科技大学 | A kind of modified image segmentation training method based on full convolutional neural networks |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112396573A (en) * | 2019-07-30 | 2021-02-23 | 纵横在线(广州)网络科技有限公司 | Facial skin analysis method and system based on image recognition |
| CN110956623B (en) * | 2019-11-29 | 2023-11-07 | 深圳数联天下智能科技有限公司 | Wrinkle detection method, wrinkle detection device, wrinkle detection equipment and computer-readable storage medium |
| CN110956623A (en) * | 2019-11-29 | 2020-04-03 | 深圳和而泰家居在线网络科技有限公司 | Wrinkle detection method, apparatus, device, and computer-readable storage medium |
| CN113129250A (en) * | 2019-12-27 | 2021-07-16 | 华为技术有限公司 | Skin detection method and device, terminal equipment and computer storage medium |
| CN111401463A (en) * | 2020-03-25 | 2020-07-10 | 维沃移动通信有限公司 | Method for outputting detection result, electronic device, and medium |
| CN111401463B (en) * | 2020-03-25 | 2024-04-30 | 维沃移动通信有限公司 | Method for outputting detection result, electronic equipment and medium |
| CN112053344A (en) * | 2020-09-02 | 2020-12-08 | 杨洋 | Skin detection method system and equipment based on big data algorithm |
| CN112837304A (en) * | 2021-02-10 | 2021-05-25 | 姜京池 | Skin detection method, computer storage medium and computing device |
| CN112837304B (en) * | 2021-02-10 | 2024-03-12 | 姜京池 | Skin detection method, computer storage medium and computing device |
| CN113128375A (en) * | 2021-04-02 | 2021-07-16 | 西安融智芙科技有限责任公司 | Image recognition method, electronic device and computer-readable storage medium |
| CN113128375B (en) * | 2021-04-02 | 2024-05-10 | 西安融智芙科技有限责任公司 | Image recognition method, electronic device, and computer-readable storage medium |
| CN113554623A (en) * | 2021-07-23 | 2021-10-26 | 江苏医像信息技术有限公司 | Intelligent quantitative analysis method and system for human face skin |
| CN113723310A (en) * | 2021-08-31 | 2021-11-30 | 平安科技（深圳）有限公司 | Image identification method based on neural network and related device |
| CN113723310B (en) * | 2021-08-31 | 2023-09-05 | 平安科技（深圳）有限公司 | Image recognition method and related device based on neural network |
| CN115119897A (en) * | 2022-06-17 | 2022-09-30 | 上海食未生物科技有限公司 | 3D printing meat printing method and system |
| WO2024014853A1 (en) * | 2022-07-13 | 2024-01-18 | 주식회사 룰루랩 | Method and device for detecting facial wrinkles using deep learning-based wrinkle detection model trained according to semi-automatic labeling |
| CN116993714A (en) * | 2023-08-30 | 2023-11-03 | 深圳伯德睿捷健康科技有限公司 | Skin detection method, system and computer readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109961426B (en) | 2021-07-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109961426A (en) | A kind of detection method of face skin skin quality | |
| CN106529429B (en) | A kind of skin of face analysis system based on image recognition | |
| CN104036278B (en) | The extracting method of face algorithm standard rules face image | |
| CN110363088B (en) | Self-adaptive skin inflammation area detection method based on multi-feature fusion | |
| CN100354875C (en) | Red eye moving method based on human face detection | |
| CN106682601B (en) | A kind of driver's violation call detection method based on multidimensional information Fusion Features | |
| CN103634680B (en) | The control method for playing back and device of a kind of intelligent television | |
| CN110097034A (en) | A kind of identification and appraisal procedure of Intelligent human-face health degree | |
| CN104484645B (en) | A kind of " 1 " gesture identification method and system towards man-machine interaction | |
| CN105139404A (en) | Identification camera capable of detecting photographing quality and photographing quality detecting method | |
| CN101251898A (en) | Skin color detection method and apparatus | |
| CN112396573A (en) | Facial skin analysis method and system based on image recognition | |
| CN106886216A (en) | Robot automatic tracking method and system based on RGBD Face datections | |
| CN108932493A (en) | A kind of facial skin quality evaluation method | |
| CN103366390B (en) | terminal and image processing method and device | |
| CN103077378B (en) | Contactless face recognition algorithms based on extension eight neighborhood Local textural feature and system of registering | |
| CN103034838A (en) | Special vehicle instrument type identification and calibration method based on image characteristics | |
| EP3249618A1 (en) | Banknote classification and identification method and device based on lab color space | |
| EP4083937A1 (en) | System and method for hair analysis of user | |
| CN109740572A (en) | A kind of human face in-vivo detection method based on partial color textural characteristics | |
| CN106067016B (en) | A kind of facial image eyeglass detection method and device | |
| CN107154058A (en) | A kind of method for guiding user to reduce magic square | |
| CN106650606A (en) | Matching and processing method of face image and face image model construction system | |
| CN109876416A (en) | A kind of rope skipping method of counting based on image information | |
| CN104000593A (en) | Color card and skin test method and system |
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 |