CN104484675B - A kind of iron spectrum abrasive particle texture feature extraction and the method for pattern-recognition - Google Patents
A kind of iron spectrum abrasive particle texture feature extraction and the method for pattern-recognition Download PDFInfo
- Publication number
- CN104484675B CN104484675B CN201410769766.6A CN201410769766A CN104484675B CN 104484675 B CN104484675 B CN 104484675B CN 201410769766 A CN201410769766 A CN 201410769766A CN 104484675 B CN104484675 B CN 104484675B
- Authority
- CN
- China
- Prior art keywords
- pixel
- msub
- sequence
- image
- picture
- 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.)
- Expired - Fee Related
Links
Classifications
-
- 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
- G06F18/2155—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the incorporation of unlabelled data, e.g. multiple instance learning [MIL], semi-supervised techniques using expectation-maximisation [EM] or naïve labelling
-
- 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
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/06—Recognition of objects for industrial automation
Abstract
The invention discloses a kind of iron spectrum abrasive particle texture feature extraction and the method for pattern-recognition, comprise the steps of:In standard iron spectrogram valut, multiple typical samples are respectively taken with the picture of equal amount respectively, and processing is amplified to each picture;The image of presetted pixel value is taken in each picture as master sample, all master sample composing training set;The characteristic vector of correspondence master sample is obtained after the generation of migration sequence and sequence diffusion that image is performed to each master sample;After all master samples have performed the generation of migration sequence and sequence diffusion of image, the characteristic vector set of training sample set is obtained;Optimal Separating Hyperplane is obtained using SVMs according to the characteristic vector set of training sample set, so as to realize grader;After the texture image that picture to be sorted is cut into presetted pixel value, it is put into grader and realizes classification.The present invention can realize the classification of these three particles of the tired particle of tired blocky-shaped particle, serious skimming wear particle, stratiform.
Description
Technical field
The present invention relates to mode identification technology, and in particular to a kind of iron spectrum abrasive particle texture feature extraction and pattern-recognition
Method, applied in the image identification of wear debris.
Background technology
Abrasive particle is important information carrier and abrasion mechanism criterion in friction, wear process.Abrasive particle is due to friction subtabulation
Face relative motion, and with the interaction of interfacial medium and ambiance, undergo a series of tribology processes, cause surface abrasion
The product of formation, containing it is abundant on material surface friction, abrasion information, its quantity, size, shape, color, pattern and
The abrasive manner of system mode and material when architectural feature etc. is produced with abrasive particle is closely related.
From damage mechanisms angle, wear mechanism can be divided mainly into adhesion(adhesion), grinding(abrasion), fatigue
(fatigue), tribochemistry(tribochemistry)Deng, corresponding wear particle can also be divided into normal skimming wear particle,
Abrasive wear particle, tired blocky-shaped particle, serious skimming wear particle, stratiform fatigue particle, oxide particle etc., wherein, it is tired
The tired particle of labor blocky-shaped particle, serious skimming wear particle, stratiform can not be identified with traditional method.
The content of the invention
It is an object of the invention to provide a kind of iron spectrum abrasive particle texture feature extraction and the method for pattern-recognition, it can realize
The classification of these three particles of the tired particle of tired blocky-shaped particle, serious skimming wear particle, stratiform, is provided for mechanical fault diagnosis
Foundation, and laid the first stone for the realization of mechanized classification instrument.
In order to achieve the above object, the present invention is achieved through the following technical solutions:A kind of iron composes abrasive particle texture feature extraction
With the method for pattern-recognition, it is characterized in, comprises the steps of:
S1, in standard iron spectrogram valut, multiple typical samples are respectively taken with the picture of equal amount respectively, and to each picture
It is amplified processing;
S2, the image of presetted pixel value is taken in each picture as master sample, all master sample composing training collection
Close;
S3, the migration sequence generation that image is performed to each master sample and sequence obtain correspondence master sample after spreading
Characteristic vector;
S4, after all master samples performed image migration sequence generation and sequence diffusion after, obtain training sample set
The characteristic vector set of conjunction;
S5, Optimal Separating Hyperplane using SVMs obtained according to the characteristic vector set of training sample set, so that real
Existing grader;
S6, picture to be sorted cut into after the texture image of presetted pixel value, be put into grader and realize classification.
Iron composes abrasive particle texture feature extraction and the method for pattern-recognition further includes step S7;
The step S7 is:
If S7.1, classification are correct, the classification of picture to be sorted is obtained;
If S7.2, classification error, picture to be sorted is put into the training set of grader.
Described multiple typical samples are tired blocky-shaped particle, serious skimming wear particle, the tired particle of stratiform.
The multiplication factor of picture is consistent in described step S1.
Migration sequence generation step is included in described step S3:
A1, using each pixel in image as starting point, according to be pre-configured with migration rule carry out migration;
A2, after all pixels have been performed both by migration in image, the pixel in image is divided into trip and returns arrangement set and non-
Arrangement set is returned in trip.
Described migration rule is:Setting memory μ size, selectes starting point pixel distance into its neighborhood territory pixel most short
Pixel is moved, wherein the pixel that μ steps have been passed by before being not included in.
The neighborhood territory pixel of a described pixel refers to that European geometric distance is less thanPixel constitute set;
Wherein,, i.e., one pixel positioned at image centre position, its
Neighborhood territory pixelIt is 8 pixels around it;Pixel positioned at image boundary position, its neighborhood territory pixelIt is around it
5 pixels;Positioned at the pixel of image Angle Position, its neighborhood territory pixelIt is 3 pixels around it.
Sequential diffusion steps are included in described step S3:
B1, the pixel returned according to all non-trips of relatedness computation formula calculating in sequence return the degree of correlation of sequence to trip;
B2, according to the degree of correlation by it is non-swim the pixel returned in sequence and be added to trip return in sequence, until to return sequential covering complete for trip
Portion's image.
Described relatedness computation formula is:
Wherein,It is pixelArrangement set is returned to tripDistance.
A kind of iron spectrum abrasive particle texture feature extraction of the present invention and the method for pattern-recognition have following compared with prior art
Advantage:Classification error rate can effectively be reduced;Quantity will be spread as characteristic vector to distinguishing tired blocky-shaped particle, seriously sliding
The tired particle of wear particle, stratiform has good effect, coordinates all kinds of mode identification methods all can be effective.
Brief description of the drawings
Fig. 1 is the texture picture of typical sample;
Fig. 2 is the diffusion profile of typical sample;
Fig. 3 is the block diagram that sequence spreads preceding 20 step diffusion;
Fig. 4 is a kind of iron spectrum abrasive particle texture feature extraction of the invention and the method flow diagram of pattern-recognition;
The sequence that Fig. 5 A are produced when being μ=2;
The sequence that Fig. 5 B are produced when being μ=3;
Fig. 6 returns the relatedness computation figure of sequence for non-trip.
Embodiment
Below in conjunction with accompanying drawing, by describing a preferably specific embodiment in detail, the present invention is further elaborated.
Typical sample as shown in Figure 1(The tired particle of tired blocky-shaped particle, serious skimming wear particle, stratiform)32 ×
32 texture tile.It may infer that in diffusion profile in Fig. 2:The texture of serious skimming wear particle is the most complicated, initial to expand
Laxity;The surface of stratiform fatigue particle is the most smooth, and initial propagations are fast;Tired blocky-shaped particle between serious skimming wear particle and
Between stratiform fatigue particle.From Fig. 3, we then infer, 20 layers of diffusion are slow before serious skimming wear particle, average each
15 pixels or so;10 layers of diffusion are exceedingly fast before stratiform fatigue particle, average each 40-50 pixel or so, afterwards diffusion velocity
It is rapid to decline;The scattering nature of tired blocky-shaped particle is between serious skimming wear particle and the tired particle of stratiform, more surely
Step, average 23-25 pixel.
By above-mentioned analysis, diffusion quantity as characteristic vector to distinguish tired blocky-shaped particle, serious skimming wear particle,
Stratiform fatigue particle has good effect, coordinates all kinds of mode identification methods all can be effective.
As shown in figure 4, specifically, sorting technique is comprised the steps of:
S1, in standard iron spectrogram valut, to multiple typical samples(Tired blocky-shaped particle, serious skimming wear particle, stratiform
Tired particle)The picture of equal amount is respectively taken respectively, and processing is amplified to each picture, and the multiplication factor of each picture is consistent;
S2, the image of presetted pixel value is taken in each picture as master sample, all master sample composing training collection
Close, wherein presetted pixel value can be any, but no more than 64 × 64, picture increase can increase amount of calculation, it is difficult to practical;
S3, the migration sequence generation that image is performed to each master sample and sequence obtain correspondence master sample after spreading
Characteristic vector;
S4, after all master samples performed image migration sequence generation and sequence diffusion after, obtain training sample set
The characteristic vector set of conjunction;
S5, Optimal Separating Hyperplane using SVMs obtained according to the characteristic vector set of training sample set, so that real
Existing grader;
S6, picture to be sorted cut into after the texture image of presetted pixel value, be put into grader and realize classification;
If S7.1, classification are correct, the classification of picture to be sorted is obtained;
If S7.2, classification error, picture to be sorted is put into the training set of grader.
Migration sequence diffusion method is the algorithm of a kind of generation based on pixel sequence and diffusion, due to different types of texture
Different diffusion characteristics can be produced, so as to for carrying out texture recognition.
Wherein, the step of migration sequence is generated is as follows:
A1, using each pixel in image as starting point, according to be pre-configured with migration rule carry out migration;
A2, after all pixels have been performed both by migration in image, the pixel in image is divided into trip and returns arrangement set and non-
Arrangement set is returned in trip.
Described migration rule is:Setting memory μ size, selectes starting point pixel distance into its neighborhood territory pixel most short
Pixel is moved, wherein the pixel that μ steps have been passed by before being not included in.
The neighborhood territory pixel of a described pixel refers to that European geometric distance is less thanPixel constitute set;
Wherein,, i.e., one pixel positioned at image centre position, its
Neighborhood territory pixelIt is 8 pixels around it;Pixel positioned at image boundary position, its neighborhood territory pixelIt is around it
5 pixels;Positioned at the pixel of image Angle Position, its neighborhood territory pixelIt is 3 pixels around it.
Assuming that image has N number of pixel, pixelGray value be, its scope is 0 to 255.Ash
Spend the absolute value for the gray value differences that distance in image is two adjacent pixels:
。
As shown in Figure 5A, the numeral in circle represents the gray scale of the pixel, and arrow represents the direction of migration, and migration is eventually
A circulation is absorbed in, this circulation returns sequence as trip.Memory μ size can be influenceed in sequence, Fig. 5 B, during memory μ=3, when walking
During to the pixel that gray scale is 18, there is the pixel that gray scale is 16 in memory, thus 16 that picture are gone to when can not be as memory μ=2
Element.Arrangement set writing is returned in trip, whereinFor starting point, μ is the size of memory.For in image, each pixel is
Point, performs walking process, can all ultimately form a trip and return sequence(It is possible to repeat).
Perform after walk process, image there are two parts, a part is that arrangement set is returned in trip, and another part is that sequence is returned in non-trip
Row set.The pixel that non-trip is mainly returned sequence by sequential diffusion steps is added to middle reaches and returns sequence.
Sequential diffusion steps are as follows:
B1, the pixel returned according to all non-trips of relatedness computation formula calculating in sequence return the degree of correlation of sequence to trip;
B2, according to the degree of correlation by it is non-swim the pixel returned in sequence and be added to trip return in sequence, until to return sequential covering complete for trip
Portion's image.
Non- trip is returned in sequencePixel returns sequence for streamThe degree of correlationIt is defined as nearest distance:
Wherein,It is pixelArrangement set is returned in convection currentDistance.
As shown in fig. 6, the circle of black, which represents trip, returns sequence pixel, the circle of grey represents the non-pixel swum and return sequence,
Numeral below in circle is the gray value of the pixel, and numeral above is then the degree of correlation calculated.Such as the first row first
The degree of correlation of the pixel of row is 7, because the distance that sequence is returned in its closest trip is 7, and the pixel phase that fifth line the 5th is arranged
Guan Du is 5().
Sequence pixel and the spatial distribution of other pixels are returned in the trip of this policy check.With the degree of correlation of each pixel, we
Just each trip can be returned sequence and expanded.The degree of correlation is arranged from small to large, in the expansion of each rank, we are just correlation
Degree identical pixel is added to trip and returned in sequence, and the quantity of every layer of expansion is constituted into characteristic vector。
Although present disclosure is discussed in detail by above preferred embodiment, but it should be appreciated that above-mentioned
Description is not considered as limitation of the present invention.After those skilled in the art have read the above, for the present invention's
A variety of modifications and substitutions all will be apparent.Therefore, protection scope of the present invention should be limited to the appended claims.
Claims (6)
1. a kind of iron spectrum abrasive particle texture feature extraction and the method for pattern-recognition, it is characterised in that comprise the steps of:
S1, in standard iron spectrogram valut, multiple typical samples are respectively taken with the picture of equal amount respectively, and each picture is carried out
Enhanced processing;
S2, the image of presetted pixel value is taken in each picture as master sample, all master sample composing training set;
S3, the migration sequence generation that image is performed to each master sample and sequence obtain the feature of correspondence master sample after spreading
Vector;
S4, after all master samples performed image migration sequence generation and sequence diffusion after, obtain training sample set
Characteristic vector set;
S5, Optimal Separating Hyperplane using SVMs obtained according to the characteristic vector set of training sample set, so as to realize point
Class device;
S6, picture to be sorted cut into after the texture image of presetted pixel value, be put into grader and realize classification;
Wherein, migration sequence generation step is included:
A1, using each pixel in image as starting point, according to be pre-configured with migration rule carry out migration;
A2, after all pixels have been performed both by migration in image, the pixel in image is divided into arrangement set is returned in trip and non-trip is returned
Arrangement set;
Described migration rule is:Setting memory μ size, selectes starting point pixel distance most short pixel into its neighborhood territory pixel
It is mobile, wherein the pixel that μ steps have been passed by before being not included in;
Wherein, sequential diffusion steps are included:
B1, the pixel returned according to all non-trips of relatedness computation formula calculating in sequence return the degree of correlation of sequence to trip;
B2, according to the degree of correlation by it is non-swim the pixel returned in sequence and be added to trip return in sequence, until sequential covering all figures are returned in trip
Picture;With the degree of correlation of each pixel, each trip is returned sequence and expanded, the degree of correlation is arranged from small to large, in every one-level
Degree of correlation identical pixel, is added to trip and returned in sequence by other expansion, and the quantity of every layer of expansion is constituted into characteristic vector.
2. the method as described in claim 1, it is characterised in that further comprising step S7;
The step S7 is:
If S7.1, classification are correct, the classification of picture to be sorted is obtained;
If S7.2, classification error, picture to be sorted is put into the training set of grader.
3. the method as described in claim 1, it is characterised in that described multiple typical samples are tired blocky-shaped particle, serious
The tired particle of skimming wear particle, stratiform.
4. the method as described in claim 1 or 3, it is characterised in that the multiplication factor of picture is consistent in described step S1.
5. the method as described in claim 1, it is characterised in that the neighborhood territory pixel of a described pixel refer to European geometry away from
From less thanPixel constitute set;
<mrow>
<mi>&eta;</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>p</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mo>{</mo>
<msub>
<mi>p</mi>
<mi>j</mi>
</msub>
<mo>|</mo>
<mi>d</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>p</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<msub>
<mi>p</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>&le;</mo>
<msqrt>
<mn>2</mn>
</msqrt>
<mo>}</mo>
</mrow>
Wherein,I.e. one pixel positioned at image centre position, its neighborhood picture
Plain η (pi) it is 8 pixels around it;Pixel positioned at image boundary position, its neighborhood territory pixel η (pi) it is 5 pictures around it
Element;Positioned at the pixel of image Angle Position, its neighborhood territory pixel η (pi) it is 3 pixels around it.
6. the method as described in claim 1, it is characterised in that described relatedness computation formula is:
<mrow>
<mi>&alpha;</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>p</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msub>
<mi>min</mi>
<msub>
<mi>p</mi>
<mi>i</mi>
</msub>
</msub>
<mo>{</mo>
<mi>&phi;</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>p</mi>
<mi>j</mi>
</msub>
<mo>,</mo>
<msubsup>
<mi>A</mi>
<mi>&mu;</mi>
<msub>
<mi>p</mi>
<mi>i</mi>
</msub>
</msubsup>
<mo>)</mo>
</mrow>
<mo>}</mo>
</mrow>
Wherein,It is pixel pjArrangement set is returned to tripDistance.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410769766.6A CN104484675B (en) | 2014-12-15 | 2014-12-15 | A kind of iron spectrum abrasive particle texture feature extraction and the method for pattern-recognition |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410769766.6A CN104484675B (en) | 2014-12-15 | 2014-12-15 | A kind of iron spectrum abrasive particle texture feature extraction and the method for pattern-recognition |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104484675A CN104484675A (en) | 2015-04-01 |
CN104484675B true CN104484675B (en) | 2017-10-31 |
Family
ID=52759216
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410769766.6A Expired - Fee Related CN104484675B (en) | 2014-12-15 | 2014-12-15 | A kind of iron spectrum abrasive particle texture feature extraction and the method for pattern-recognition |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104484675B (en) |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105631481B (en) * | 2016-01-07 | 2018-12-07 | 西安交通大学 | Iron based on genetic programming composes abrasive grain compound characteristics building method |
CN105701816A (en) * | 2016-01-13 | 2016-06-22 | 上海海事大学 | Automatic image segmentation method |
CN105547705B (en) * | 2016-03-10 | 2018-07-20 | 西安工业大学 | Engine performance degradation trend prediction technique |
CN108305259B (en) * | 2018-02-06 | 2020-03-24 | 西安交通大学 | Multi-texture feature fusion type automatic abrasive particle type identification method |
CN108446706B (en) * | 2018-02-27 | 2021-01-19 | 西安交通大学 | Automatic abrasive grain material identification method based on principal component extraction of colors |
US11842553B2 (en) * | 2018-12-06 | 2023-12-12 | ExxonMobil Technology and Engineering Company | Wear detection in mechanical equipment |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4717470A (en) * | 1983-03-24 | 1988-01-05 | Ivar Apeland | Method for classifying sand |
CN101706812A (en) * | 2009-11-24 | 2010-05-12 | 清华大学 | Method and device for searching documents |
CN102542285A (en) * | 2011-08-03 | 2012-07-04 | 清华大学 | Image collection scene sorting method and image collection scene sorting device based on spectrogram analysis |
CN102768730A (en) * | 2012-06-25 | 2012-11-07 | 中国人民解放军总参谋部陆航研究所 | Interactive wear particle image annotation method |
CN103886579A (en) * | 2013-12-11 | 2014-06-25 | 西安交通大学 | Abrasive particle chain self-adaptive segmentation method orienting online ferrographic image automatic identification |
-
2014
- 2014-12-15 CN CN201410769766.6A patent/CN104484675B/en not_active Expired - Fee Related
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4717470A (en) * | 1983-03-24 | 1988-01-05 | Ivar Apeland | Method for classifying sand |
CN101706812A (en) * | 2009-11-24 | 2010-05-12 | 清华大学 | Method and device for searching documents |
CN102542285A (en) * | 2011-08-03 | 2012-07-04 | 清华大学 | Image collection scene sorting method and image collection scene sorting device based on spectrogram analysis |
CN102768730A (en) * | 2012-06-25 | 2012-11-07 | 中国人民解放军总参谋部陆航研究所 | Interactive wear particle image annotation method |
CN103886579A (en) * | 2013-12-11 | 2014-06-25 | 西安交通大学 | Abrasive particle chain self-adaptive segmentation method orienting online ferrographic image automatic identification |
Non-Patent Citations (1)
Title |
---|
图像颜色特征提取在铁谱图像分类及磨粒识别中的应用研究;陈桂明,谢友柏,江良洲;《中国机械工程》;20060831;第17卷(第15期);第1576-1580页 * |
Also Published As
Publication number | Publication date |
---|---|
CN104484675A (en) | 2015-04-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104484675B (en) | A kind of iron spectrum abrasive particle texture feature extraction and the method for pattern-recognition | |
CN110176024B (en) | Method, device, equipment and storage medium for detecting target in video | |
CN102074020B (en) | Method for performing multi-body depth recovery and segmentation on video | |
CN103886325B (en) | Cyclic matrix video tracking method with partition | |
CN107424161B (en) | Coarse-to-fine indoor scene image layout estimation method | |
CN110751195B (en) | Fine-grained image classification method based on improved YOLOv3 | |
CN113033454B (en) | Method for detecting building change in urban video shooting | |
CN103903275A (en) | Method for improving image segmentation effects by using wavelet fusion algorithm | |
Lin et al. | Exploring the potential of image-based 3D geometry and appearance reasoning for automated construction progress monitoring | |
CN110570402B (en) | Binocular salient object detection method based on boundary perception neural network | |
CN107301417A (en) | A kind of method and device of the vehicle brand identification of unsupervised multilayer neural network | |
Mayr et al. | Self-supervised learning of the drivable area for autonomous vehicles | |
CN102289685B (en) | Behavior identification method for rank-1 tensor projection based on canonical return | |
CN113450579A (en) | Method, device, equipment and medium for acquiring speed information | |
CN103942793A (en) | Video consistent motion area detection method based on thermal diffusion | |
Song et al. | Robust single image reflection removal against adversarial attacks | |
CN114241522A (en) | Method, system, equipment and storage medium for field operation safety wearing identification | |
Kumar et al. | Semantic segmentation-based image inpainting detection | |
Ahsan et al. | Histogram of spatio temporal local binary patterns for human action recognition | |
JP5896661B2 (en) | Information processing apparatus, information processing apparatus control method, and program | |
Jahedsaravani et al. | Measurement of bubble size and froth velocity using convolutional neural networks | |
Li et al. | A novel game-theoretic model for content-adaptive image steganography | |
Fulir et al. | Synthetic data for defect segmentation on complex metal surfaces | |
CN103761513B (en) | A kind of face identification method based on mixed vector projection | |
Feng et al. | MT-ORL: multi-task occlusion relationship learning |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20171031 Termination date: 20201215 |