CN108985302A - A kind of skin lens image processing method, device and equipment - Google Patents
A kind of skin lens image processing method, device and equipment Download PDFInfo
- Publication number
- CN108985302A CN108985302A CN201810772239.9A CN201810772239A CN108985302A CN 108985302 A CN108985302 A CN 108985302A CN 201810772239 A CN201810772239 A CN 201810772239A CN 108985302 A CN108985302 A CN 108985302A
- Authority
- CN
- China
- Prior art keywords
- image
- lens image
- skin
- processed
- skin lens
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/30—Noise filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/03—Recognition of patterns in medical or anatomical images
Abstract
The application discloses a kind of skin lens image processing method, device and equipment, this method comprises: receiving skin lens image to be processed;Image characteristics extraction is carried out to the skin lens image to be processed, obtains the feature vector of the skin lens image to be processed;The first classification results are obtained after the processing of first disaggregated model using the feature vector of the skin lens image to be processed as the input parameter of trained first disaggregated model;According to first classification results, the cutaneous lesions result on the skin lens image to be processed is determined.The application using the first disaggregated model to skin lens image carry out cutaneous lesions identification, accurate cutaneous lesions can be obtained as a result, and recognition efficiency it is higher.
Description
Technical field
This application involves field of image processings, and in particular to a kind of skin lens image processing method, device and equipment.
Background technique
With the arrival of family doctor's concept, more and more families wish in the case where staying indoor can and doctor into
Row is linked up, to obtain the diagnostic result of some professions.For this purpose, some family's diagnostic equipments also come into being, such as household blood pressure
Instrument, domestic glucometer, household ophthalmoscope etc., skin mirror device is increasingly becoming each family as dermopathic diagnostic equipment
Requisite instrumentation.Skin mirror device is a kind of skin microscope that can amplify decades of times, is for observing skin-color disposition illness
Sharp weapon, using its obtain skin lens image can be used in diagnosing skin disease.
Currently, family doctor manually assesses cutaneous lesions by skin lens image, still, for general family
Doctor, the cutaneous lesions that accurate and self-confident determining whether there is needs skin biopsy or expert to change the place of examination still lack necessary training,
So needing a kind of effective ways that cutaneous lesions can be recognized accurately by skin lens image at present.
Summary of the invention
To solve the above problems, this application provides a kind of skin lens image processing method, device and equipment, particular technique
Scheme is as follows:
In a first aspect, this application provides a kind of skin lens image processing methods, which comprises
Receive skin lens image to be processed;
Image characteristics extraction is carried out to the skin lens image to be processed, obtains the feature of the skin lens image to be processed
Vector;
Using the feature vector of the skin lens image to be processed as the input parameter of trained first disaggregated model,
After the processing of first disaggregated model, the first classification results are obtained;
According to first classification results, the cutaneous lesions result on the skin lens image to be processed is determined.
Optionally, described according to first classification results, determine the cutaneous lesions on the skin lens image to be processed
As a result before, further includes:
Edge extracting processing is carried out to the skin lens image to be processed, obtains the edge of the skin lens image to be processed
Extract image;
Using the edge extracting image of the skin lens image to be processed as the input of trained second disaggregated model
Parameter obtains the second classification results after the processing of second disaggregated model;
Correspondingly, it is described according to first classification results, determine the cutaneous lesions on the skin lens image to be processed
As a result, specifically:
Comprehensive first classification results and second classification results, determine the skin on the skin lens image to be processed
Skin lesion result.
Optionally, described that image characteristics extraction is carried out to the skin lens image to be processed, obtain the skin to be processed
Before the feature vector of mirror image, further includes:
The processing of image enhanced fuzzy is carried out to the skin lens image to be processed.
Optionally, it is described the processing of image enhanced fuzzy carried out to the skin lens image to be processed before, further includes:
Denoising is filtered to the skin mirror image to be processed.
Optionally, described using the feature vector of the skin lens image to be processed as trained first disaggregated model
Input parameter, after the processing of first disaggregated model, before obtaining the first classification results, further includes:
The first training set of images is obtained, the first image training set includes several dermoscopies with cutaneous lesions label
Image;
Image characteristics extraction is carried out to each skin lens image in the first image training set respectively, obtains each skin
The feature vector of skin mirror image;
The first pre-generated disaggregated model is trained using the feature vector of each skin lens image, obtain by
The first trained disaggregated model.
Optionally, described using the edge extracting image of the skin lens image to be processed as trained second classification
The input parameter of model, after the processing of second disaggregated model, before obtaining the second classification results, further includes:
The second training set of images is obtained, second training set of images includes several dermoscopies with cutaneous lesions label
Image;
Edge extracting processing is carried out to each skin lens image in second training set of images respectively, obtains each skin
The edge extracting image of skin mirror image;
The second pre-generated disaggregated model is trained using the edge extracting image of each skin lens image, is obtained
Trained second disaggregated model.
Optionally, the pretreated skin lens image described to process respectively carries out image characteristics extraction or described
Respectively before the pretreated skin lens image progress edge extracting processing described to process, further includes:
The skin lens image with cutaneous lesions label is pre-processed respectively, the pretreatment includes that filtering is gone
Make an uproar processing and image enhanced fuzzy processing.
Optionally, it is described pretreatment further include predetermined angle rotation processing and or mirror image processing.
Second aspect, present invention also provides a kind of skin lens image processing unit, described device includes:
Receiving module, for receiving skin lens image to be processed;
First extraction module obtains described wait locate for carrying out image characteristics extraction to the skin lens image to be processed
Manage the feature vector of skin lens image;
First categorization module, for using the feature vector of the skin lens image to be processed as trained first point
The input parameter of class model obtains the first classification results after the processing of first disaggregated model;
Determining module, for determining the skin disease on the skin lens image to be processed according to first classification results
Become result.
Optionally, described device further include:
Second extraction module obtains described wait locate for carrying out edge extracting processing to the skin lens image to be processed
Manage the edge extracting image of skin lens image;
Second categorization module, for using the edge extracting image of the skin lens image to be processed as trained
The input parameter of two disaggregated models obtains the second classification results after the processing of second disaggregated model;
Correspondingly, the determining module, is specifically used for:
Comprehensive first classification results and second classification results, determine the skin on the skin lens image to be processed
Skin lesion result.
Optionally, described device further include:
First preprocessing module, for carrying out the processing of image enhanced fuzzy to the skin lens image to be processed.
Optionally, described device further include:
Second preprocessing module, for being filtered denoising to the skin mirror image to be processed.
Optionally, described device further include:
First obtains module, and for obtaining the first training set of images, the first image training set includes several with skin
The skin lens image of skin lesion label;
It is special to carry out image to each skin lens image in the first image training set for respectively for third extraction module
Sign is extracted, and the feature vector of each skin lens image is obtained;
First training module, for the feature vector using each skin lens image to the first pre-generated disaggregated model
It is trained, obtains trained first disaggregated model.
Optionally, described device further include:
Second obtains module, and for obtaining the second training set of images, second training set of images includes several with skin
The skin lens image of skin lesion label;
4th extraction module is mentioned for carrying out edge to each skin lens image in second training set of images respectively
Processing is taken, the edge extracting image of each skin lens image is obtained;
Second training module, for the edge extracting image using each skin lens image to the second pre-generated classification
Model is trained, and obtains trained second disaggregated model.
Optionally, described device further include:
Third preprocessing module, for being pre-processed respectively to the skin lens image with cutaneous lesions label,
The pretreatment includes filtering and noise reduction processing and the processing of image enhanced fuzzy.
Optionally, it is described pretreatment further include predetermined angle rotation processing and or mirror image processing.
The third aspect, this application provides a kind of skin lens image processing equipment, the equipment includes memory and processing
Device,
Said program code is transferred to the processor for storing program code by the memory;
The processor is used to execute described in any item skins of first aspect according to the instruction in said program code
Mirror image processing method.
In skin lens image processing method provided by the present application, after receiving skin lens image to be processed, to described wait locate
It manages skin lens image and carries out image characteristics extraction, obtain the feature vector of the skin lens image to be processed;It will be described to be processed
Input parameter of the feature vector of skin lens image as trained first disaggregated model, by first disaggregated model
Processing after, obtain the first classification results;Finally, according to first classification results, the skin lens image to be processed is determined
On cutaneous lesions result.The application carries out the identification of cutaneous lesions using the first disaggregated model to skin lens image, and existing
The manual evaluation cutaneous lesions mode for lacking necessary training in technology is compared, and the application utilizes first by great amount of samples training
Disaggregated model skin lens image is identified can obtain accurate cutaneous lesions as a result, and recognition efficiency it is higher.
In addition, the application can also be identified using edge extracting image of second disaggregated model to skin lens image,
The classification results of final comprehensive first disaggregated model and the second disaggregated model determine cutaneous lesions as a result, further improving skin
The accuracy of skin lesion result.
Detailed description of the invention
In order to more clearly explain the technical solutions in the embodiments of the present application, make required in being described below to embodiment
Attached drawing is briefly described, it should be apparent that, the drawings in the following description are only some examples of the present application, for
For those of ordinary skill in the art, without any creative labor, it can also be obtained according to these attached drawings
His attached drawing.
Fig. 1 is a kind of flow chart of skin lens image processing method provided by the embodiments of the present application;
Fig. 2 is the flow chart of another skin lens image processing method provided by the embodiments of the present application;
Fig. 3 is a kind of structural schematic diagram of skin lens image processing unit provided by the embodiments of the present application;
Fig. 4 is the structural schematic diagram of another skin lens image processing unit provided by the embodiments of the present application;
Fig. 5 provides a kind of structural schematic diagram of skin lens image processing equipment for the embodiment of the present application.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present application, technical solutions in the embodiments of the present application carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of embodiments of the present application, instead of all the embodiments.It is based on
Embodiment in the application, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall in the protection scope of this application.
Skin lens image is the image obtained using skin mirror device, since skin lens image is able to reflect cutaneous lesions feelings
Condition, so, doctor can diagnose skin disease based on skin lens image.But the diagnosis to skin disease, especially
The cutaneous lesions for needing skin biopsy or expert to change the place of examination are judged whether there is, the knowledge that comparison is professional, general family doctor are needed
Life may lack relevant professional knowledge, so accurate cutaneous lesions result can not be provided.
Based on this, this application provides a kind of skin lens image processing methods, based on by the training of a large amount of skin lens images
Disaggregated model, the identification of cutaneous lesions is carried out to skin lens image, so that it is determined that whether patient suffers from skin disease, and really
Patient is determined with any class skin disease etc..Specifically, the application is after receiving skin lens image to be processed, to the skin to be processed
Skin mirror image carries out image characteristics extraction, obtains the feature vector of the skin lens image to be processed.Using this feature vector as warp
The input parameter for crossing the first disaggregated model of training obtains the first classification results, most after the processing of first disaggregated model
The cutaneous lesions result on skin lens image is determined according to first classification results eventually.The application is utilized by great amount of samples training
Disaggregated model to skin lens image carry out automatic identification, accurate cutaneous lesions can be obtained as a result, and recognition efficiency
It is higher.
The embodiment for introducing a kind of skin lens image processing method provided by the present application in detail below is this Shen with reference to Fig. 1
Please a kind of flow chart of skin lens image processing method that provides of embodiment, this method specifically includes:
S101: skin lens image to be processed is received.
In the embodiment of the present application, skin lens image to be processed is the image obtained using skin mirror device, skin to be processed
Mirror image is able to reflect cutaneous lesions situation, can know whether patient suffers from cutaneous lesions based on skin lens image to be processed
Deng.
The embodiment of the present application is using skin lens image to be processed as process object, wherein skin lens image to be processed can be with
It is an image, is also possible to multiple images, that is to say, that the embodiment of the present application can be simultaneously to multiple dermoscopy figure to be processed
Identification as carrying out cutaneous lesions result, to further increase recognition efficiency.
S102: image characteristics extraction is carried out to the skin lens image to be processed, obtains the skin lens image to be processed
Feature vector.
In the embodiment of the present application, after receiving skin lens image to be processed, image characteristics extraction is carried out to it, to obtain
The feature vector of the skin lens image to be processed.Wherein, this feature vector can embody the feature of the skin lens image to be processed.
Scale invariant features transform (Scale-invariant feature transform, SIFT) calculation is generallyd use in practical application
Method carries out feature extraction to skin lens image, with no restrictions to specific features extracting method at this.
In order to enable the image effect that cutaneous lesions situation is able to reflect in skin lens image to be processed is apparent, the application
Embodiment first filters the skin lens image to be processed before carrying out image characteristics extraction to skin lens image to be processed
Wave denoising, to remove the noise on image.
In a kind of optional embodiment, it can use mean filter and skin lens image to be processed be filtered at denoising
Reason, this method are a kind of methods denoised using linear function.Specifically, assuming that f (x, y) is the original that a frame contains pollution noise
Beginning image, g (x, y) is the image handled by filtering and noise reduction, and in filtering and noise reduction treatment process, the value of g (x, y) is pixel
Point (x, y) is positioned adjacent to the average value of the grey scale pixel value of image-region, passes through filtering and noise reduction by what average value operation determined
Reduce pixel outstanding (i.e. noise spot) in the image of processing, to inhibit noise jamming, specifically, above-mentioned utilize mean value
The process that image denoising processing is completed in filtering can indicate that formula (1) is as follows with formula (1):
Wherein, the point centered on (x, y), the region having a size of m × n are adjacent image region.
In addition, the mode equivalence that formwork calculation also can be used in the functional operation in formula (1) replaces, wherein mean value filter
Wave template size is m × n.In practical application, each specific gravity power in mean filter template can be adjusted according to specific needs
Value, for example, the mean filter template that size is 3 × 3 below is common template, each specific gravity weight therein can be according to specific
It needs to be adjusted;Specifically, common template is as follows, specific gravity weight therein is respectively 1/9,1/10,1/16:
It is worth noting that, above mean filter is a kind of side for being filtered denoising to skin lens image
Formula, the embodiment of the present application can also realize filtering and noise reduction using other modes, no longer introduce one by one herein.
In addition, the skin lens image got due to skin mirror device is there may be unsharp situation, in order to by skin
Feature instantiation in mirror image becomes apparent from, to obtain more accurate cutaneous lesions recognition result, the embodiment of the present application is right
Before skin lens image to be processed carries out image characteristics extraction, image enhanced fuzzy processing can also be carried out to it.Specifically, this
Application can complete it is above-mentioned denoising is filtered to skin lens image to be processed after, to the skin lens image to be processed into
The processing of row image enhanced fuzzy.
In practical application, before carrying out the processing of image enhanced fuzzy to skin lens image to be processed, ash is carried out to it first
Degreeization processing carries out the processing of image enhanced fuzzy, a kind of image enhanced fuzzy processing introduced below based on the image after gray processing
Concrete mode:
Step 1), according to formula (2) by the data obfuscation of skin lens image to be processed;
Wherein, fijIndicate that the gray value of pixel (i, j), L indicate the skin lens image to be processed by gray processing processing
Gray level;μijIndicate the degree of membership of pixel (i, j);Work as fijWhen=0, μijFor minimum value 0, work as fij=L-1, μijFor maximum
Value 1, that is to say, that μijValue range be [0,1].
Specifically, skin lens image to be processed can be transformed into fuzzy space from data space using formula (2), determine
The fuzzy matrix of the skin lens image to be processed.It wherein, include the degree of membership of each pixel in fuzzy matrix.
Step 2) completes μijCalculating after, enhanced fuzzy fortune is implemented to the skin lens image to be processed using formula (3)
It calculates;
Wherein, μcExpression is getted over a little, μcValue need not be equal to 0.5, set generally according to demand;I(μij) indicate
To degree of membership μijImplement the membership values obtained after enhanced fuzzy operation;
Specifically, I (μij) for indicating that reduction (works as μij> μc) or increase (work as μij≤μc)μijValue;Work as μij≤μcWhen,
Nonlinear transformation operation can make μijValue increase, thus make the low ash angle value f of pixel (i, j)ijIncrease;In turn, work as μij>
μcWhen, nonlinear transformation operation can make μijValue reduce, therefore make the high gray value f of pixel (i, j)ijReduce.The application is real
The membership values of each pixel on skin lens image to be processed can be enhanced using above-mentioned enhanced fuzzy operation by applying example.
Step 3) after calculating enhancing by nonlinear function to skin lens image to be processed on fuzzy space, needs every
The degree of membership μ of a pixelijIt carries out taking inverse transformation, skin lens image to be processed is converted back into data space from fuzzy space, is obtained
The image that must enhance that treated;Wherein, for taking the formula (4) of inverse transformation as follows:
f′ij=(L-1) μij (4)
Wherein, f 'ijIndicate μijThe gray value obtained after taking inverse transformation.
Specifically, taking inverse transformation formula (4) can be by each pixel on enhancing treated image using above-mentioned
Degree of membership is converted to gray value, obtains skin lens image to be processed and implements the result after enhanced fuzzy operation.
Step 4), the adjustment to enhancing treated image degree of comparing obtained in step 3) obtain final complete
The image handled at image enhanced fuzzy;
Specifically, calculating the completion image enhanced fuzzy after setting contrast using formula (5), (6), (7), (8)
The gray value of each pixel on the image of processing;
Wherein, f indicates initial pictures, usually by filtering and noise reduction treated skin lens image;gkIt indicates as volume
The Gaussian function of product core;fkIndicate the image after two-dimensional discrete convolutional calculation;K is Gaussian function number;KkValue meet
Formula (6);It can choosekIt is 3, choosing σ is 5,20,100;
Specifically, do two-dimensional discrete convolutional calculation to initial pictures f using formula (5), the image f that obtains that treatedk;So
Afterwards, the image f that each process of convolution obtains is calculated separately using formula (7)kWith the ratio of initial pictures f:
Finally, carrying out linear weighted function calculating using formula (8), picture contrast is adjusted, processing result f " is obtained;
Wherein, wkFor weighting coefficient, usually taking 1/2, f " is the gray value of pixel in image f.
In the embodiment of the present application, by above-mentioned steps 1) to step 4) processing skin lens image feature instantiation it is brighter
It is aobvious, it ultimately helps to obtain more accurate cutaneous lesions recognition result.
In succession by the skin lens image to be processed of the processing of above-mentioned filtering and noise reduction, the processing of image enhanced fuzzy, spy is completed
After the extraction for levying vector, S103 is continued to execute.
S103: using the feature vector of the skin lens image to be processed as the input of trained first disaggregated model
Parameter obtains the first classification results after the processing of first disaggregated model.
In the embodiment of the present application, after the feature vector that S102 gets skin lens image to be processed, by this feature vector
As the input parameter of trained first disaggregated model, which is based on this feature vector and classifies, and obtains
To corresponding first classification results of the skin lens image to be processed.
It is right in advance before the first disaggregated model is used to carry out cutaneous lesions identification to skin lens image in practical application
First disaggregated model is trained, and specific training process is introduced subsequent.
S104: according to first classification results, the cutaneous lesions result on the skin lens image to be processed is determined.
In the embodiment of the present application, after the first disaggregated model exports the first classification results of the skin lens image to be processed,
According to first classification results, the cutaneous lesions result on the skin lens image to be processed is determined.
In a kind of optional embodiment, if the classification results of first disaggregated model are 0 and 1, and 0 represents skin disease
Become result and represents cutaneous lesions result into no lesion or without certain cutaneous lesions, 1 to have lesion or having certain cutaneous lesions;Then root
It can determine according to the first classification results with the presence or absence of cutaneous lesions on the skin lens image to be processed, or may determine whether to deposit
In certain cutaneous lesions, such as melanoma.
In another optional embodiment, if the classification results of first disaggregated model are 0,1,2 ..., and 0 generation
Table cutaneous lesions result is no lesion, and 1,2 ... respectively represent a kind of skin disease type (such as melanoma);Then according to
One classification results determination can determine the cutaneous lesions that whether there is on the skin lens image to be processed, and determine that there are skins
Specific skin disease type can be known when lesion.
Skin lens image processing method provided by the embodiments of the present application is based on using trained first disaggregated model
Image feature vector carries out the identification of cutaneous lesions to skin lens image, determines the cutaneous lesions result of patient.With the prior art
The middle manual evaluation cutaneous lesions mode for lacking necessary training is compared, and the embodiment of the present application utilizes the by great amount of samples training
One disaggregated model skin lens image is identified can obtain accurate cutaneous lesions as a result, and recognition efficiency it is higher.
In order to more accurately determine the cutaneous lesions on skin lens image as a result, the embodiment of the present application also provides a kind of skins
Skin mirror image processing method, on the basis of above method embodiment, the embodiment of the present application can also utilize trained the
Two disaggregated models classify to skin lens image to be processed, obtain the second classification results, final comprehensive first classification results and
Second classification results, determine the cutaneous lesions on skin lens image to be processed as a result, compared with above method embodiment, the application
Embodiment can obtain more accurate skin disease diagnostic result.
The embodiment for introducing another skin lens image processing method provided by the embodiments of the present application in detail below, with reference to figure
2, for the flow chart of another skin lens image processing method provided by the embodiments of the present application, this method is specifically included:
S201: skin lens image to be processed is received.
S202: image characteristics extraction is carried out to the skin lens image to be processed, obtains the skin lens image to be processed
Feature vector.
S203: using the feature vector of the skin lens image to be processed as the input of trained first disaggregated model
Parameter obtains the first classification results after the processing of first disaggregated model.
S201-S203 in the embodiment of the present application is identical as the S101-S103 in above method embodiment, can refer to reason
Solution, details are not described herein.
S204: edge extracting processing is carried out to the skin lens image to be processed, obtains the skin lens image to be processed
Edge extracting image.
Also influencing disease due to the edge shape (as the edge of melanoma is irregular) of the lesion region on skin
Diagnosis, detects the cutaneous lesions on skin lens image for this purpose, the embodiment of the present application is also based on edge extracting image.
Specifically, the embodiment of the present application can successively be filtered denoising, image enhanced fuzzy to skin lens image to be processed
After processing etc., edge extracting processing is carried out to the skin lens image to be processed, the edge for obtaining the skin lens image to be processed mentions
Take image, wherein edge extracting image can embody the edge feature of the skin lens image to be processed.
In practical application, Su Beier Sobel operator can be used by carrying out edge extracting processing to skin lens image, wherein
Sobel operator includes both direction template, the two direction templates are used to calculate the transverse edge and longitudinal edge of skin lens image
Edge, and calculated transverse edge and longitudinal edge carry out operation with image convolution respectively, finally obtain horizontal and vertical
The gradient approximation of straight edge.
It is illustrated below, it is assumed that A indicates initial skin mirror image, GxWith GyRespectively indicate the ladder at horizontal and vertical edge
Approximation is spent, calculation formula and horizontal vertical template are as follows:
Wherein, horizontal shuttering:Vertical formwork:
Then calculation formula:
In practical application, after obtaining the gradient approximation at horizontal and vertical edge by above-mentioned calculation formula, continue by
According to formula G=| H |+| V | carry out the approximate integral gradient value for seeking the skin lens image.The embodiment of the present application passes through above-mentioned Sobel
The skin lens image to be processed can be calculated by edge extracting treated edge extracting image in operator.
It is worth noting that, above-mentioned Sobel operator is only intended to realize the one way in which of Edge extraction, this Shen
Please embodiment other modes are not limited.
S205: using the edge extracting image of the skin lens image to be processed as trained second disaggregated model
Input parameter obtains the second classification results after the processing of second disaggregated model.
In the embodiment of the present application, after S204 gets the edge extracting image of skin lens image to be processed, by the edge
Input parameter of the image as trained second disaggregated model is extracted, which is based on the edge extracting image
Classify, obtains corresponding second classification results of the skin lens image to be processed.
It is right in advance before the second disaggregated model is used to carry out cutaneous lesions identification to skin lens image in practical application
Second disaggregated model is trained, and specific training process is introduced subsequent.
S206: comprehensive first classification results and second classification results determine the skin lens image to be processed
On cutaneous lesions result.
After obtaining the first classification results and the second classification results, in order to more accurately determine on skin lens image to be processed
Cutaneous lesions as a result, comprehensive first classification results of the embodiment of the present application and the second classification results, finally determine cutaneous lesions
As a result.
In a kind of optional embodiment, if the first classification results are no lesion, and the second classification results are no lesion,
It is then both comprehensive as a result, the cutaneous lesions result on final skin lens image to be processed can be disease-free change;If first point
Class result is to have a lesion, and the second classification results are to have lesion, then both comprehensive as a result, on final skin lens image to be processed
Cutaneous lesions result can be ill change;In addition, if only by one to have in the first classification results and the second classification results
Lesion, then it is both comprehensive as a result, the cutaneous lesions result on final skin lens image to be processed is uncertain.
In addition, for the case where classification results are specific skin disease type, only in the first classification results and second point
When class result is a certain skin disease type determined, the cutaneous lesions knot on skin lens image to be processed can be determined
Fruit is this kind of skin disease type, and otherwise cutaneous lesions result is uncertain.
The embodiment of the present application is using the first disaggregated model based on image feature vector to the cutaneous lesions on skin lens image
It is identified, and the cutaneous lesions on skin lens image is identified based on edge extracting image using the second disaggregated model, most
Both comprehensive classification results determine the cutaneous lesions on the skin lens image as a result, compared with above method embodiment eventually, this
Application embodiment further improves the accuracy of skin disease diagnosis.
In addition, the application is before handling skin lens image using the first disaggregated model and the second disaggregated model,
It is trained firstly the need of to pre-generated the first disaggregated model and the second disaggregated model, specifically, to the first disaggregated model
Training method it is as follows:
S1 obtains the first training set of images, and described image training set includes several dermoscopies with cutaneous lesions label
Image.
S2 carries out image characteristics extraction to each skin lens image in the first training set of images respectively, obtains each skin
The feature vector of skin mirror image;
S3 is trained the first pre-generated disaggregated model using the feature vector of each skin lens image, obtains
Trained first disaggregated model.
In the embodiment of the present application, several skin lens images with cutaneous lesions label are obtained as training sample and form the
One training set of images, for being trained to the first pre-generated disaggregated model.In practical application, there is cutaneous lesions label
Skin lens image can be the manual mark from professional skin disease doctor, be also possible to other modes and obtain, do not do herein
It limits.
Skin disease label can be to show whether corresponding skin lens image has the label of lesion, be also possible to show pair
The skin lens image answered has the label of certain skin disease type, is not defined herein to the form of label, the class of label
Type determines the identification degree of cutaneous lesions, if the label of training sample is whether to have lesion, final cutaneous lesions knot
Fruit is also whether to have lesion, if the label of training sample is with certain skin disease type, final cutaneous lesions
As a result or certain skin disease type is identified.
In order to more fully learn to the skin lens image with cutaneous lesions label, the embodiment of the present application can be right
Each skin lens image in first training set of images carries out the rotation processing or mirror image processing of predetermined angle, obtains angular transformation
Image, and angular transformation image is also added in the first training set of images, as training sample.By above-mentioned to skin lens image
Pretreatment, the study of all angles feature can be more fully carried out to existing skin lens image, while also further
The quantity of training sample is expanded.It is worth noting that, the angular transformation image of the first training set of images, which is added, also has correspondence
Skin lens image cutaneous lesions label.
In addition, before being trained using the training sample in the first training set of images to the first disaggregated model, first
Each training sample is pre-processed respectively, including successively carry out predetermined angle rotation processing and or mirror image processing, filtering
Denoising, the processing of image enhanced fuzzy, specific treatment process can refer to the description of preceding method embodiment, no longer superfluous herein
It states.
In the embodiment of the present application, the skin lens image in the first training set of images with cutaneous lesions label is subjected to image
After feature extraction, the feature vector of each skin lens image is obtained, using the feature vector of each skin lens image to pre- Mr.
At the first disaggregated model be trained, obtain trained first disaggregated model, for skin lens image carry out skin
The identification of lesion.
To the training method of the second disaggregated model with it is above-mentioned similar to the training method of the first disaggregated model, can refer to reason
Solution.Specifically, as follows to the training method of the second disaggregated model:
S11 obtains the second training set of images, and second training set of images includes several skins with cutaneous lesions label
Skin mirror image;
S12 carries out edge extracting processing to each skin lens image in the second training set of images respectively, obtains each skin
The edge extracting image of skin mirror image;
S13 is trained the second pre-generated disaggregated model using the edge extracting image of each skin lens image,
Obtain trained second disaggregated model.
Second training set of images and above-mentioned first training set of images can be the same training set of images, that is, include identical
Training sample is also possible to different training set of images, specifically, to having cutaneous lesions label in the second training set of images
The pretreatment of skin lens image can refer to the pre- place to the skin lens image in the first training set of images with cutaneous lesions label
Understood, details are not described herein.Wherein, pretreatment include predetermined angle rotation processing and or mirror image processing, filtering go
Make an uproar processing and image enhanced fuzzy processing etc..
In addition, during being trained to the second disaggregated model will there are cutaneous lesions in the second training set of images
After the skin lens image of label carries out edge extracting processing, the edge extracting image of each skin lens image is obtained, utilization is each
The edge extracting image of skin lens image is trained the second pre-generated disaggregated model, obtains trained second point
Class model, for carrying out the identification of cutaneous lesions to skin lens image.Specifically, carrying out edge extracting processing to skin lens image
Process can refer to the description of preceding method embodiment, details are not described herein.
In the embodiment of the present application, the first disaggregated model and the second disaggregated model are trained using a large amount of training sample
Afterwards, trained first disaggregated model and the second disaggregated model are obtained, subsequent while two kinds of models of utilization are to skin lens image
On cutaneous lesions identified that both comprehensive classification results more efficient can more accurately obtain skin disease diagnosis knot
Fruit.
Corresponding to the above method embodiment, present invention also provides a kind of skin lens image processing units, with reference to figure
3, it is a kind of structural schematic diagram of skin lens image processing unit provided by the embodiments of the present application, described device includes:
Receiving module 301, for receiving skin lens image to be processed;
First extraction module 302, for carrying out image characteristics extraction to the skin lens image to be processed, obtain it is described to
Handle the feature vector of skin lens image;
First categorization module 303, for using the feature vector of the skin lens image to be processed as trained
The input parameter of one disaggregated model obtains the first classification results after the processing of first disaggregated model;
Determining module 304, for determining the skin on the skin lens image to be processed according to first classification results
Lesion result.
The embodiment of the present application also provides a kind of skin lens image processing units to provide with reference to Fig. 4 for the embodiment of the present application
Another skin lens image processing unit structural schematic diagram, described device not only includes the modules in Fig. 1, can be with
Include:
Second extraction module 401, for carrying out edge extracting processing to the skin lens image to be processed, obtain it is described to
Handle the edge extracting image of skin lens image;
Second categorization module 402, for using the edge extracting image of the skin lens image to be processed as by training
The input parameter of the second disaggregated model obtain the second classification results after the processing of second disaggregated model;
Correspondingly, the determining module 304, is specifically used for:
Comprehensive first classification results and second classification results, determine the skin on the skin lens image to be processed
Skin lesion result.
Described device further include:
First preprocessing module, for carrying out the processing of image enhanced fuzzy to the skin lens image to be processed.
Described device further include:
Second preprocessing module, for being filtered denoising to the skin mirror image to be processed.
In order to be trained to the first disaggregated model, described device further include:
First obtains module, and for obtaining the first training set of images, the first image training set includes several with skin
The skin lens image of skin lesion label;
It is special to carry out image to each skin lens image in the first image training set for respectively for third extraction module
Sign is extracted, and the feature vector of each skin lens image is obtained;
First training module, for the feature vector using each skin lens image to the first pre-generated disaggregated model
It is trained, obtains trained first disaggregated model.
In order to be trained to the second disaggregated model, described device further include:
Second obtains module, and for obtaining the second training set of images, second training set of images includes several with skin
The skin lens image of skin lesion label;
4th extraction module is mentioned for carrying out edge to each skin lens image in second training set of images respectively
Processing is taken, the edge extracting image of each skin lens image is obtained;
Second training module, for the edge extracting image using each skin lens image to the second pre-generated classification
Model is trained, and obtains trained second disaggregated model.
In order to improve the accuracy to the first disaggregated model and the second disaggregated model training, described device further include:
Third preprocessing module, for being pre-processed respectively to the skin lens image with cutaneous lesions label,
The pretreatment includes filtering and noise reduction processing and the processing of image enhanced fuzzy.
In order to enrich training sample, the pretreatment further include predetermined angle rotation processing and or mirror image processing.
Skin lens image processing unit provided by the embodiments of the present application is based on using trained first disaggregated model
Image feature vector carries out the identification of cutaneous lesions to skin lens image, determines the cutaneous lesions result of patient.With the prior art
Compare, the embodiment of the present application can obtain accurate cutaneous lesions as a result, and recognition efficiency it is higher.
In addition, skin lens image processing unit provided by the embodiments of the present application, can also be based on using the second disaggregated model
Edge extracting image identifies that the classification results of final comprehensive the two determine the dermoscopy figure to the cutaneous lesions on skin lens image
As upper cutaneous lesions as a result, further improving the accuracy of skin disease diagnosis.
Correspondingly, the embodiment of the present invention also provides a kind of skin lens image processing equipment, it is shown in Figure 5, may include:
Processor 501, memory 502, input unit 503 and output device 504.Place in skin lens image processing equipment
The quantity for managing device 501 can be one or more, take a processor as an example in Fig. 5.In some embodiments of the invention, it handles
Device 501, memory 502, input unit 503 and output device 504 can be connected by bus or other means, wherein in Fig. 5 with
For being connected by bus.
Memory 502 can be used for storing software program and module, and processor 501 is stored in memory 502 by operation
Software program and module, thereby executing the various function application and data processing of skin lens image processing equipment.Storage
Device 502 can mainly include storing program area and storage data area, wherein storing program area can storage program area, at least one
Application program needed for function etc..In addition, memory 502 may include high-speed random access memory, it can also include non-easy
The property lost memory, a for example, at least disk memory, flush memory device or other volatile solid-state parts.Input dress
Setting 503 can be used for receiving the number or character information of input, and generate with the user setting of skin lens image processing equipment with
And the related signal input of function control.
Specifically in the present embodiment, processor 501 can be according to following instruction, by one or more application program
The corresponding executable file of process be loaded into memory 502, and run and be stored in memory 502 by processor 501
Application program, to realize the various functions in above-mentioned skin lens image processing method.
For device embodiment, since it corresponds essentially to embodiment of the method, so related place is referring to method reality
Apply the part explanation of example.The apparatus embodiments described above are merely exemplary, wherein described be used as separation unit
The unit of explanation may or may not be physically separated, and component shown as a unit can be or can also be with
It is not physical unit, it can it is in one place, or may be distributed over multiple network units.It can be according to actual
It needs that some or all of the modules therein is selected to achieve the purpose of the solution of this embodiment.Those of ordinary skill in the art are not
In the case where making the creative labor, it can understand and implement.
It should be noted that, in this document, relational terms such as first and second and the like are used merely to a reality
Body or operation are distinguished with another entity or operation, are deposited without necessarily requiring or implying between these entities or operation
In any actual relationship or order or sequence.Moreover, the terms "include", "comprise" or its any other variant are intended to
Non-exclusive inclusion, so that the process, method, article or equipment including a series of elements is not only wanted including those
Element, but also including other elements that are not explicitly listed, or further include for this process, method, article or equipment
Intrinsic element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that
There is also other identical elements in process, method, article or equipment including the element.
A kind of skin lens image processing method, device and equipment provided by the embodiment of the present application have been carried out in detail above
It introduces, specific examples are used herein to illustrate the principle and implementation manner of the present application, the explanation of above embodiments
It is merely used to help understand the present processes and its core concept;At the same time, for those skilled in the art, according to this
The thought of application, there will be changes in the specific implementation manner and application range, in conclusion the content of the present specification is not answered
It is interpreted as the limitation to the application.
Claims (10)
1. a kind of skin lens image processing method, which is characterized in that the described method includes:
Receive skin lens image to be processed;
Image characteristics extraction is carried out to the skin lens image to be processed, obtain the feature of the skin lens image to be processed to
Amount;
Using the feature vector of the skin lens image to be processed as the input parameter of trained first disaggregated model, pass through
After the processing of first disaggregated model, the first classification results are obtained;
According to first classification results, the cutaneous lesions result on the skin lens image to be processed is determined.
2. skin lens image processing method according to claim 1, which is characterized in that described to be tied according to first classification
Fruit, before determining the cutaneous lesions result on the skin lens image to be processed, further includes:
Edge extracting processing is carried out to the skin lens image to be processed, obtains the edge extracting of the skin lens image to be processed
Image;
Using the edge extracting image of the skin lens image to be processed as the input parameter of trained second disaggregated model,
After the processing of second disaggregated model, the second classification results are obtained;
Correspondingly, described according to first classification results, determine the cutaneous lesions on the skin lens image to be processed as a result,
Specifically:
Comprehensive first classification results and second classification results, determine the skin disease on the skin lens image to be processed
Become result.
3. skin lens image processing method according to claim 1 or 2, which is characterized in that described to the skin to be processed
Skin mirror image carries out image characteristics extraction, before obtaining the feature vector of the skin lens image to be processed, further includes:
The processing of image enhanced fuzzy is carried out to the skin lens image to be processed.
4. skin lens image processing method according to claim 3, which is characterized in that described to the dermoscopy to be processed
Image carries out before the processing of image enhanced fuzzy, further includes:
Denoising is filtered to the skin mirror image to be processed.
5. skin lens image processing method according to claim 1, which is characterized in that described by the dermoscopy to be processed
Input parameter of the feature vector of image as trained first disaggregated model, by the processing of first disaggregated model
Afterwards, before obtaining the first classification results, further includes:
The first training set of images is obtained, the first image training set includes several dermoscopy figures with cutaneous lesions label
Picture;
Image characteristics extraction is carried out to each skin lens image in the first image training set respectively, obtains each dermoscopy
The feature vector of image;
The first pre-generated disaggregated model is trained using the feature vector of each skin lens image, is obtained by training
The first disaggregated model.
6. skin lens image processing method according to claim 2, which is characterized in that described by the dermoscopy to be processed
Input parameter of the edge extracting image of image as trained second disaggregated model, by second disaggregated model
After processing, before obtaining the second classification results, further includes:
The second training set of images is obtained, second training set of images includes several dermoscopy figures with cutaneous lesions label
Picture;
Edge extracting processing is carried out to each skin lens image in second training set of images respectively, obtains each dermoscopy
The edge extracting image of image;
The second pre-generated disaggregated model is trained using the edge extracting image of each skin lens image, obtain by
The second trained disaggregated model.
7. skin lens image processing method according to claim 5 or 6, which is characterized in that described respectively described in process
Pretreated skin lens image carry out image characteristics extraction or it is described respectively to by the pretreated skin lens image into
Before the processing of row edge extracting, further includes:
The skin lens image with cutaneous lesions label is pre-processed respectively, the pretreatment includes at filtering and noise reduction
Reason and the processing of image enhanced fuzzy.
8. skin lens image processing method according to claim 7, which is characterized in that the pretreatment further includes preset angle
The rotation processing of degree and or mirror image processing.
9. a kind of skin lens image processing unit, which is characterized in that described device includes:
Receiving module, for receiving skin lens image to be processed;
First extraction module obtains the skin to be processed for carrying out image characteristics extraction to the skin lens image to be processed
The feature vector of skin mirror image;
First categorization module, for using the feature vector of the skin lens image to be processed as trained first classification mould
The input parameter of type obtains the first classification results after the processing of first disaggregated model;
Determining module, for determining the cutaneous lesions knot on the skin lens image to be processed according to first classification results
Fruit.
10. a kind of skin lens image processing equipment, which is characterized in that the equipment includes memory and processor,
Said program code is transferred to the processor for storing program code by the memory;
The processor is used for according to the instruction in said program code, and perform claim requires skin described in any one of 1-8
Mirror image processing method.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810772239.9A CN108985302A (en) | 2018-07-13 | 2018-07-13 | A kind of skin lens image processing method, device and equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810772239.9A CN108985302A (en) | 2018-07-13 | 2018-07-13 | A kind of skin lens image processing method, device and equipment |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108985302A true CN108985302A (en) | 2018-12-11 |
Family
ID=64537544
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810772239.9A Pending CN108985302A (en) | 2018-07-13 | 2018-07-13 | A kind of skin lens image processing method, device and equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108985302A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111772588A (en) * | 2020-07-29 | 2020-10-16 | 天津大学 | Classification method of skin mirror images based on neural network ensemble learning |
CN111797923A (en) * | 2020-07-03 | 2020-10-20 | 北京阅视智能技术有限责任公司 | Training method of image classification model, and image classification method and device |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN201806692U (en) * | 2009-12-31 | 2011-04-27 | 中国人民解放军空军总医院 | Multispectral dermoscopy image automatic analytical instrument for diagnosing malignant melanocyte tumour |
CN103646398A (en) * | 2013-12-04 | 2014-03-19 | 山西大学 | Demoscopy focus automatic segmentation method |
CN103778441A (en) * | 2014-02-26 | 2014-05-07 | 东南大学 | Dezert-Smaradache Theory (DSmT) and Hidden Markov Model (HMM) aircraft sequence target recognition method |
CN104680498A (en) * | 2015-03-24 | 2015-06-03 | 江南大学 | Medical image segmentation method based on improved gradient vector flow model |
CN106682435A (en) * | 2016-12-31 | 2017-05-17 | 西安百利信息科技有限公司 | System and method for automatically detecting lesions in medical image through multi-model fusion |
CN107203999A (en) * | 2017-04-28 | 2017-09-26 | 北京航空航天大学 | A kind of skin lens image automatic division method based on full convolutional neural networks |
CN107464230A (en) * | 2017-08-23 | 2017-12-12 | 京东方科技集团股份有限公司 | Image processing method and device |
CN107729948A (en) * | 2017-10-31 | 2018-02-23 | 京东方科技集团股份有限公司 | Image processing method and device, computer product and storage medium |
US20180130203A1 (en) * | 2016-11-06 | 2018-05-10 | International Business Machines Corporation | Automated skin lesion segmentation using deep side layers |
CN108198620A (en) * | 2018-01-12 | 2018-06-22 | 洛阳飞来石软件开发有限公司 | A kind of skin disease intelligent auxiliary diagnosis system based on deep learning |
-
2018
- 2018-07-13 CN CN201810772239.9A patent/CN108985302A/en active Pending
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN201806692U (en) * | 2009-12-31 | 2011-04-27 | 中国人民解放军空军总医院 | Multispectral dermoscopy image automatic analytical instrument for diagnosing malignant melanocyte tumour |
CN103646398A (en) * | 2013-12-04 | 2014-03-19 | 山西大学 | Demoscopy focus automatic segmentation method |
CN103778441A (en) * | 2014-02-26 | 2014-05-07 | 东南大学 | Dezert-Smaradache Theory (DSmT) and Hidden Markov Model (HMM) aircraft sequence target recognition method |
CN104680498A (en) * | 2015-03-24 | 2015-06-03 | 江南大学 | Medical image segmentation method based on improved gradient vector flow model |
US20180130203A1 (en) * | 2016-11-06 | 2018-05-10 | International Business Machines Corporation | Automated skin lesion segmentation using deep side layers |
CN106682435A (en) * | 2016-12-31 | 2017-05-17 | 西安百利信息科技有限公司 | System and method for automatically detecting lesions in medical image through multi-model fusion |
CN107203999A (en) * | 2017-04-28 | 2017-09-26 | 北京航空航天大学 | A kind of skin lens image automatic division method based on full convolutional neural networks |
CN107464230A (en) * | 2017-08-23 | 2017-12-12 | 京东方科技集团股份有限公司 | Image processing method and device |
CN107729948A (en) * | 2017-10-31 | 2018-02-23 | 京东方科技集团股份有限公司 | Image processing method and device, computer product and storage medium |
CN108198620A (en) * | 2018-01-12 | 2018-06-22 | 洛阳飞来石软件开发有限公司 | A kind of skin disease intelligent auxiliary diagnosis system based on deep learning |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111797923A (en) * | 2020-07-03 | 2020-10-20 | 北京阅视智能技术有限责任公司 | Training method of image classification model, and image classification method and device |
CN111772588A (en) * | 2020-07-29 | 2020-10-16 | 天津大学 | Classification method of skin mirror images based on neural network ensemble learning |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110428475B (en) | Medical image classification method, model training method and server | |
CN110232383B (en) | Focus image recognition method and focus image recognition system based on deep learning model | |
CN107895367B (en) | Bone age identification method and system and electronic equipment | |
CN106056595B (en) | Based on the pernicious assistant diagnosis system of depth convolutional neural networks automatic identification Benign Thyroid Nodules | |
Kumar et al. | Resnet-based approach for detection and classification of plant leaf diseases | |
CN109241967B (en) | Thyroid ultrasound image automatic identification system based on deep neural network, computer equipment and storage medium | |
CN112508850B (en) | Deep learning-based method for detecting malignant area of thyroid cell pathological section | |
CN108446621A (en) | Bank slip recognition method, server and computer readable storage medium | |
CN109389584A (en) | Multiple dimensioned rhinopharyngeal neoplasm dividing method based on CNN | |
CN109447981A (en) | Image-recognizing method and Related product | |
CN111291825A (en) | Focus classification model training method and device, computer equipment and storage medium | |
CN113012155A (en) | Bone segmentation method in hip image, electronic device, and storage medium | |
CN112233777A (en) | Gallstone automatic identification and segmentation system based on deep learning, computer equipment and storage medium | |
Zhao et al. | Fine-grained diabetic wound depth and granulation tissue amount assessment using bilinear convolutional neural network | |
CN108985302A (en) | A kind of skin lens image processing method, device and equipment | |
WO2024074921A1 (en) | Distinguishing a disease state from a non-disease state in an image | |
CN114757908A (en) | Image processing method, device and equipment based on CT image and storage medium | |
Hatano et al. | Classification of osteoporosis from phalanges CR images based on DCNN | |
CN110930373A (en) | Pneumonia recognition device based on neural network | |
CN113706514A (en) | Focus positioning method, device and equipment based on template image and storage medium | |
CN103268494A (en) | Parasite egg identifying method based on sparse representation | |
CN108876776A (en) | A kind of method of generating classification model, eye fundus image classification method and device | |
CN114010227B (en) | Right ventricle characteristic information identification method and device | |
CN113344911B (en) | Method and device for measuring size of calculus | |
CN115880358A (en) | Construction method of positioning model, positioning method of image mark points and electronic equipment |
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: 20181211 |
|
RJ01 | Rejection of invention patent application after publication |