CN106651866A - Multi-beam water column target automatic segmentation method based on neural network - Google Patents
Multi-beam water column target automatic segmentation method based on neural network Download PDFInfo
- Publication number
- CN106651866A CN106651866A CN201611205431.7A CN201611205431A CN106651866A CN 106651866 A CN106651866 A CN 106651866A CN 201611205431 A CN201611205431 A CN 201611205431A CN 106651866 A CN106651866 A CN 106651866A
- Authority
- CN
- China
- Prior art keywords
- target
- sonar
- image
- neural network
- water column
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Landscapes
- Measurement Of Velocity Or Position Using Acoustic Or Ultrasonic Waves (AREA)
Abstract
The invention provides a multi-beam water column target automatic segmentation method based on a neural network. The method comprises the steps that a multi-beam original echo signal is converted into a sonar chart; the sonar chart is searched to determine a suspicious region where a target exists, and the suspicious region is separated out of the sonar chart; the suspicious region where the target exists is subjected to rough segmentation to obtain sonar images with the target and side lobes; some of the sonar images with the target and the side lobes are used as training samples, and the training samples are sent into the BP neural network for training; image processing is performed on the rest of the sonar images with the target and the side lobes through the trained BP neural network, so that the target is segmented out of the sonar images. Through the method, isolated noise points and side lobe jamming in the sonar chart where the target region is located can be effectively eliminated, the target can be extracted accurately, and the method is beneficial for recognizing and tracing the target in a water column.
Description
Technical field
The present invention relates to multi-beam water column process field, more particularly to a kind of multi-beam water column target based on neutral net
Automatic division method.
Background technology
Target recognition and tracking in water body is an important applied field of multibeam echosounder, can by water column analysis
The target informations such as the shoal of fish in obtain water body, submarine, bridge pier.But, because under water acoustic environment is complicated and changeable, noise is more,
And water column information is easily disturbed by multi-beam tunnel-effect, therefore, carry out the analysis of multi-beam water column and there is certain being stranded
It is difficult.
In view of this, it would be highly desirable to develop a kind of multi-beam water column analysis and processing method that can solve the problem that the problems referred to above.
The content of the invention
The purpose of the present invention aims to solve the problem that drawback, automatic so as to provide a kind of multi-beam water column target based on neutral net
Dividing method.
For achieving the above object, the invention provides a kind of multi-beam water column target based on neutral net side of segmentation automatically
Method.The method is comprised the following steps:
A, multi-beam original echoed signals are converted into into sonar chart;
B, the search sonar chart, it is determined that there is the suspicious region of target and separate from the sonar chart;
C, the suspicious region to there is target carry out coarse segmentation, to be partitioned into the sonar image with target and secondary lobe;
D, sonar image of the part with target and secondary lobe is sent into into BP neural network as training sample it is trained, the portion
The sonar image with target and secondary lobe is divided to be the image that can directly distinguish target and secondary lobe;
E, image procossing is carried out to sonar image of the remainder with target and secondary lobe by the BP neural network that trains, with
Target is partitioned into from sonar image, sonar image of the remainder with target and secondary lobe is not directly to distinguish
The image of target and secondary lobe.
Preferably, also include between step d and step e:
F, judge BP neural network training whether complete, when recognition correct rate exceed setting value when, judge to complete;Conversely, judging
Do not complete, continue to train.
Preferably, the multi-beam original echoed signals in step a are to realize turning for sonar chart by neighbor interpolation method
Change.
Preferably, step b is specifically included:Pixel value in sonar chart and given threshold are compared, will be above setting
Determine threshold region to be partitioned into, filter off less than the region of given threshold.
Preferably, step c is specifically included:
G, the suspicious region of presence target to isolating carry out histogram equalization process, to force down the pixel value of ambient noise,
Raise the pixel value of highlight bar;
H, by thresholding method by pixel value higher than setting value point and region segmentation go out;
Highlighted isolated noise in i, point and region by zone marker method by the pixel value being partitioned into higher than setting value is removed.
Preferably, step d is specifically included:
J, the characteristic quantity extracted in sonar image of the part with target and secondary lobe, the characteristic quantity includes:Part sonar image
In per width in the major axis of each sub-regions and ratio, the part sonar image of short axle per width in each sub-regions center to often
In the transverse and longitudinal coordinate difference and part sonar image of width picture centre per width in each sub-regions Hu squares front second moment;It is described
The arrange parameter of BP neural network includes:It is 5, output number of layers 2, middle number of layers 8 to choose number of layers 3, input number of layers;
K, using the characteristic quantity of each sub-regions for extracting as training sample send into BP neural network be trained.
The present invention can effectively eliminate target area institute based on the multi-beam water column target automatic division method of neutral net
Isolated noise and secondary lobe interference in sonar chart, accurately goes out Objective extraction, is conducive to entering the target in water column
Row identification and tracking.
Description of the drawings
Fig. 1 is flow process of the present invention based on one embodiment of the multi-beam water column target automatic division method of neutral net
Figure;
Fig. 2 is flow process of the present invention based on another embodiment of the multi-beam water column target automatic division method of neutral net
Figure;
Fig. 3 is present invention determine that there is the method flow diagram of the suspicious region of target;
Fig. 4 is the method flow diagram that suspicious region of the present invention to there is target carries out coarse segmentation;
Fig. 5 is the method flow diagram of present invention training BP neural network;
Fig. 6 is that the multi-beam echo-signal of the present invention is converted into the image schematic diagram after sonar chart;
Fig. 7 is the target area image Jing after histogram equalization and histogram distribution schematic diagram of the present invention;
Fig. 8 is the target and side lobe image schematic diagram after the coarse segmentation of the present invention;
Fig. 9 is the target image schematic diagram that the Jing BP neural networks of the present invention are partitioned into.
Specific embodiment
In order that the objects, technical solutions and advantages of the present invention become more apparent, below in conjunction with inventive embodiments
Accompanying drawing, the technical scheme in inventive embodiments is clearly and completely described, it is clear that the embodiments described below are only
Only it is a part of embodiment of invention, and not all embodiment.Based on the embodiment in invention, those of ordinary skill in the art exist
The all other embodiment obtained under the premise of creative work is not made, the scope of invention protection is belonged to.
Fig. 1 is referred to, the invention provides one of the multi-beam water column target automatic division method based on neutral net
Embodiment, comprises the following steps:
In a step 101, multi-beam original echoed signals are converted into into sonar chart, the method for conversion is neighbor interpolation method, specifically
Way is to solve 4 hithermost beam spots respectively to each picture position, is closed using the position between 4 points and picture position
System estimates the pixel value of the picture point, and to each location of pixels aforesaid operations are carried out, and finally gives the sonar chart shown in Fig. 6.
In a step 102, the sonar chart in search step 101, it is determined that there is the suspicious region of target and from sonar chart
Separate, target is the shoal of fish in water body, the information such as submarine or bridge pier.The schematic diagram isolated can be found in Fig. 7.
In step 103, the suspicious region to there is target in step 102 carries out coarse segmentation, to be partitioned into target
With the sonar image of secondary lobe, effect can be found in shown in Fig. 8.
At step 104, the sonar image in step 103 partly with target and secondary lobe is sent into into BP as training sample
Neutral net is trained, and sonar image of the which part with target and secondary lobe is the sound that can clearly distinguish target and secondary lobe
Receive image.Can be judged as clearly distinguishing the sonar of target and secondary lobe by experience and general knowledge than sonar image as shown in Figure 8
Image.
In step 105, by the BP neural network that trains in step 104 to remainder with target and secondary lobe
Sonar image carries out image procossing, and target is partitioned into from sonar image, and effect can be found in shown in Fig. 9.Wherein, remainder
The sonar image with target and secondary lobe is divided to be the image for not directly distinguishing target and secondary lobe.
It is more than a reality of the multi-beam water column target automatic division method based on neutral net that the present invention is provided
Apply example to be described in detail, below by the side of segmentation automatically of the multi-beam water column target based on neutral net for providing the present invention
Another embodiment of method is described in detail.
Refer to Fig. 2, the invention provides based on neutral net multi-beam water column target automatic division method it is another
Individual embodiment, comprises the following steps:
In step 201, multi-beam original echoed signals are converted into into sonar chart, the method for conversion is neighbor interpolation method, specifically
Way is to solve 4 hithermost beam spots respectively to each picture position, is closed using the position between 4 points and picture position
System estimates the pixel value of the picture point, and to each location of pixels aforesaid operations are carried out, and finally gives the sonar chart shown in Fig. 6.
In step 202., the sonar chart in search step 201, it is determined that there is the suspicious region of target and from sonar chart
Separate, target is the shoal of fish in water body, the information such as submarine or bridge pier.The schematic diagram isolated can be found in Fig. 7.
In step 203, the suspicious region to there is target in step 202 carries out coarse segmentation, to be partitioned into target
With the sonar image of secondary lobe, effect can be found in shown in Fig. 8.
In step 204, the sonar image in step 203 partly with target and secondary lobe is sent into into BP as training sample
Neutral net is trained, and sonar image of the which part with target and secondary lobe is the sound that can clearly distinguish target and secondary lobe
Image is received, than three strips that sonar image as shown in Figure 8 can be judged the upper left corner, the upper right corner and the lower right corner by experience and general knowledge
Shape subregion is secondary lobe interference, and middle subregion is target.
In step 205, judge whether the BP neural network training that sample training is sent in step 204 completes, work as identification
When accuracy exceedes setting value, judge to complete;Conversely, judging not completing, continue to train.Wherein setting value may be configured as 97%.Such as
Fruit sends into BP neural network and is trained using such sonar image of Fig. 8 as sample, and the result of training is to judge it is 3 mesh
Mark and 1 secondary lobe, then the accuracy of such result of determination is very low, hence it is evident that less than preset value, then BP neural network
Voluntarily adjusting parameter is accomplished by, continues to train, until accuracy reaches preset value, just judge that BP neural network training is completed.
In step 206, by the BP neural network that trains in step 205 to remainder with target and secondary lobe
Sonar image carries out image procossing, and target is partitioned into from sonar image, and effect can be found in shown in Fig. 9.White in Fig. 9
Region is target.Wherein sonar image of the remainder with target and secondary lobe is cannot clearly to distinguish target and secondary lobe
Sonar image.
Be more than to the present invention provide the multi-beam water column target automatic division method based on neutral net another
Embodiment is described in detail, below by an enforcement of the suspicious region method that there is target to the determination that the present invention is provided
Example is described in detail.
Fig. 3 is referred to, the invention provides determining the one embodiment for the suspicious region method that there is target, the embodiment
Mainly step 202 in step 102 in Fig. 1 and Fig. 2 is specifically described.
In step 301, by the pixel value and given threshold of the sonar chart in step 201 in step 101 in Fig. 1 or Fig. 2
Compare, judge sonar chart pixel value whether higher than given threshold.
In step 302, pixel value in step 301 is gone out higher than the region segmentation of given threshold, pixel value is less than and is set
The region for determining threshold value filters off.
It is that, because the pixel value of target area is general higher, and the pixel value of ambient noise is very low using above-mentioned steps, because
This can set a pixel value thresholding, will be above the region segmentation of the thresholding out, filter off less than the region of the thresholding.It is above-mentioned
Step primary concern is that elimination ambient noise and should not omit target area, therefore in actual set thresholding, the value is not
Can be too high, in case it is not very high target area to miss partial pixel value.
Be more than the determination suspicious region method that there is target that the present invention is provided one embodiment carry out it is detailed
Description, is below carried out one embodiment that rough segmentation segmentation method is carried out to the suspicious region to there is target that the present invention is provided in detail
Thin description.
Fig. 4 is referred to, the invention provides the suspicious region to there is target carries out one embodiment of rough segmentation segmentation method,
The embodiment is mainly specifically described to step 203 in step 103 in Fig. 1 and Fig. 2.
In step 401, by the suspicious region of the presence target isolated in step 202 in step 102 in Fig. 1 or Fig. 2
Histogram equalization process is carried out, to force down the pixel value of ambient noise, the pixel value of highlight bar is raised.
In step 402, by thresholding method by pixel value in step 401 higher than setting value point and region segmentation
Go out.
In step 403, by zone marker method by the pixel value being partitioned in step 402 higher than setting value Dian Hequ
Highlighted isolated noise in domain is removed.
It is more than that one embodiment that the suspicious region to there is target that the present invention is provided carries out rough segmentation segmentation method is entered
The detailed description of row, carries out detailed retouching by the one embodiment for the training BP neural network method provided the present invention below
State.
Fig. 5 is referred to, the invention provides one embodiment of training BP neural network method, the embodiment is mainly to figure
Step 204 is specifically described in step 104 and Fig. 2 in 1.
In step 501, the partly sonar with target and secondary lobe is extracted in Fig. 1 in step 103 or Fig. 2 in step 203
Characteristic quantity in image, this feature amount includes:In the sonar image of part per width in each sub-regions major axis and short axle ratio,
In the sonar image of part per width in each sub-regions center to each image center transverse and longitudinal coordinate difference and part sonar chart
As in per width in each sub-regions Hu squares front second moment;The arrange parameter of BP neural network includes:Choose number of layers 3, defeated
Enter number of layers for 5, output number of layers be 2, middle number of layers be 8.
Wherein, ratio of semi-minor axis length:Major axis is the oval long axis length for having identical standard second-order moment around mean with region.It is short
Axle is the oval minor axis length for having identical standard second-order moment around mean with region.Transverse and longitudinal coordinate of all subregion to picture centre
Difference:The center of each sub-regions is to the abscissa of the picture centre of view picture figure, the difference of ordinate.The upper left corner in such as Fig. 8
Stripe region to the difference of the transverse and longitudinal coordinate at view picture figure center is all greater than target to the difference of the transverse and longitudinal coordinate of picture centre.Hu squares
It is to be commonly used to calculate one group of image feature amount not bending moment, they are translated in image, when scaling and rotate, are worth holding not
Become, therefore be referred to as not bending moment, be used herein as its front second moment and characterize sub-district characteristic of field.The selection number of layers 3 of BP neural network is:It is defeated
Enter layer, intermediate layer and output layer.Be input into number of layers is for 5:One ratio of semi-minor axis length, two coordinate differences and two front second orders
Square.Exporting number of layers 2 is:As a result two kinds of forms, yes/no.Middle number of layers is 8, or is asked for middle number of layers
Other suitable natural numbers that empirical equation is obtained.
In step 502, the characteristic quantity of each sub-regions extracted in step 501 is sent into into BP as training sample refreshing
Jing networks are trained.
To sum up, the present invention can effectively eliminate target area based on the multi-beam water column target automatic division method of neutral net
Isolated noise and secondary lobe interference in the sonar chart of domain place, accurately goes out Objective extraction, is conducive to the mesh in water column
Mark is identified and follows the trail of.
Above-described specific embodiment, has been carried out further to the purpose of the present invention, technical scheme and beneficial effect
Describe in detail, should be understood that the specific embodiment that the foregoing is only the present invention, be not intended to limit the present invention
Protection domain, all any modification, equivalent substitution and improvements within the spirit and principles in the present invention, done etc. all should include
Within protection scope of the present invention.
Claims (6)
1. a kind of multi-beam water column target automatic division method based on neutral net, it is characterised in that comprise the following steps:
A, multi-beam original echoed signals are converted into into sonar chart;
B, the search sonar chart, it is determined that there is the suspicious region of target and separate from the sonar chart;
C, the suspicious region to there is target carry out coarse segmentation, to be partitioned into the sonar image with target and secondary lobe;
D, sonar image of the part with target and secondary lobe is sent into into BP neural network as training sample it is trained, the portion
The sonar image with target and secondary lobe is divided to be the image that can directly distinguish target and secondary lobe;
E, image procossing is carried out to sonar image of the remainder with target and secondary lobe by the BP neural network that trains, with
Target is partitioned into from sonar image, sonar image of the remainder with target and secondary lobe is not directly to distinguish
The image of target and secondary lobe.
2. a kind of multi-beam water column target automatic division method based on neutral net according to claim 1, its feature
It is also to include between step d and step e:
F, judge BP neural network training whether complete, when recognition correct rate exceed setting value when, judge to complete;Conversely, judging
Do not complete, continue to train.
3. a kind of multi-beam water column target automatic division method based on neutral net according to claim 1, its feature
It is that the multi-beam original echoed signals in step a are the conversions that sonar chart is realized by neighbor interpolation method.
4. a kind of multi-beam water column target automatic division method based on neutral net according to claim 1, its feature
It is that step b is specifically included:Pixel value in sonar chart and given threshold are compared, given threshold region is will be above
It is partitioned into, filters off less than the region of given threshold.
5. a kind of multi-beam water column target automatic division method based on neutral net according to claim 1, its feature
It is that step c is specifically included:
G, the suspicious region of presence target to isolating carry out histogram equalization process, to force down the pixel value of ambient noise,
Raise the pixel value of highlight bar;
H, by thresholding method by pixel value higher than setting value point and region segmentation go out;
Highlighted isolated noise in i, point and region by zone marker method by the pixel value being partitioned into higher than setting value is removed.
6. a kind of multi-beam water column target automatic division method based on neutral net according to claim 1, its feature
It is that step d is specifically included:
J, the characteristic quantity extracted in sonar image of the part with target and secondary lobe, the characteristic quantity includes:Part sonar image
In per width in the major axis of each sub-regions and ratio, the part sonar image of short axle per width in each sub-regions center to often
In the transverse and longitudinal coordinate difference and part sonar image of width picture centre per width in each sub-regions Hu squares front second moment;It is described
The arrange parameter of BP neural network includes:It is 5, output number of layers 2, middle number of layers 8 to choose number of layers 3, input number of layers;
K, using the characteristic quantity of each sub-regions for extracting as training sample send into BP neural network be trained.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611205431.7A CN106651866A (en) | 2016-12-23 | 2016-12-23 | Multi-beam water column target automatic segmentation method based on neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611205431.7A CN106651866A (en) | 2016-12-23 | 2016-12-23 | Multi-beam water column target automatic segmentation method based on neural network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106651866A true CN106651866A (en) | 2017-05-10 |
Family
ID=58826790
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611205431.7A Pending CN106651866A (en) | 2016-12-23 | 2016-12-23 | Multi-beam water column target automatic segmentation method based on neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106651866A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107564007A (en) * | 2017-08-02 | 2018-01-09 | 中国科学院计算技术研究所 | The scene cut modification method and system of amalgamation of global information |
CN108229402A (en) * | 2018-01-08 | 2018-06-29 | 哈尔滨工程大学 | Event detection system and detection method based on sound wave |
CN111028154A (en) * | 2019-11-18 | 2020-04-17 | 哈尔滨工程大学 | Rough-terrain seabed side-scan sonar image matching and splicing method |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20050270905A1 (en) * | 2004-04-06 | 2005-12-08 | College Of William & Mary | System and method for identification and quantification of sonar targets in a liquid medium |
CN103927534A (en) * | 2014-04-26 | 2014-07-16 | 无锡信捷电气股份有限公司 | Sprayed character online visual detection method based on convolutional neural network |
CN104732509A (en) * | 2013-12-18 | 2015-06-24 | 北京三星通信技术研究有限公司 | Self-adaptation image segmentation method and device |
CN105138008A (en) * | 2015-08-05 | 2015-12-09 | 魏芳 | Multifunctional submersible for underwater submerged reef detection |
CN105741284A (en) * | 2016-01-28 | 2016-07-06 | 中国船舶重工集团公司第七一〇研究所 | Multi-beam forward-looking sonar target detection method |
-
2016
- 2016-12-23 CN CN201611205431.7A patent/CN106651866A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20050270905A1 (en) * | 2004-04-06 | 2005-12-08 | College Of William & Mary | System and method for identification and quantification of sonar targets in a liquid medium |
CN104732509A (en) * | 2013-12-18 | 2015-06-24 | 北京三星通信技术研究有限公司 | Self-adaptation image segmentation method and device |
CN103927534A (en) * | 2014-04-26 | 2014-07-16 | 无锡信捷电气股份有限公司 | Sprayed character online visual detection method based on convolutional neural network |
CN105138008A (en) * | 2015-08-05 | 2015-12-09 | 魏芳 | Multifunctional submersible for underwater submerged reef detection |
CN105741284A (en) * | 2016-01-28 | 2016-07-06 | 中国船舶重工集团公司第七一〇研究所 | Multi-beam forward-looking sonar target detection method |
Non-Patent Citations (1)
Title |
---|
杨治明等: ""BP人工神经网络在图像分割中的应用"", 《计算机科学》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107564007A (en) * | 2017-08-02 | 2018-01-09 | 中国科学院计算技术研究所 | The scene cut modification method and system of amalgamation of global information |
CN107564007B (en) * | 2017-08-02 | 2020-09-11 | 中国科学院计算技术研究所 | Scene segmentation correction method and system fusing global information |
CN108229402A (en) * | 2018-01-08 | 2018-06-29 | 哈尔滨工程大学 | Event detection system and detection method based on sound wave |
CN111028154A (en) * | 2019-11-18 | 2020-04-17 | 哈尔滨工程大学 | Rough-terrain seabed side-scan sonar image matching and splicing method |
CN111028154B (en) * | 2019-11-18 | 2023-05-09 | 哈尔滨工程大学 | Side-scan sonar image matching and stitching method for rugged seafloor |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110929560B (en) | Video semi-automatic target labeling method integrating target detection and tracking | |
CN110544251B (en) | Dam crack detection method based on multi-migration learning model fusion | |
CN111325764B (en) | Fruit image contour recognition method | |
CN109191459B (en) | Automatic identification and rating method for continuous casting billet macrostructure center segregation defect | |
CN108509919B (en) | Method for detecting and identifying waterline in video or picture based on deep learning | |
CN107942329B (en) | Method for detecting sea surface ship target by maneuvering platform single-channel SAR | |
CN109977774B (en) | Rapid target detection method based on adaptive convolution | |
CN109242870A (en) | A kind of sea horizon detection method divided based on image with textural characteristics | |
CN107066952A (en) | A kind of method for detecting lane lines | |
CN104217215A (en) | Classification and identification method for foggy water surface image and clear water surface image | |
CN108009522B (en) | Road detection method, device and terminal | |
CN101915910B (en) | Method and system for identifying marine oil spill object by marine radar | |
CN109308713B (en) | Improved nuclear correlation filtering underwater target tracking method based on forward-looking sonar | |
CN106651866A (en) | Multi-beam water column target automatic segmentation method based on neural network | |
CN105182350A (en) | Multi-beam sonar target detection method by applying feature tracking | |
CN104021553A (en) | Sonar image object detection method based on pixel point layering | |
CN111968159A (en) | Simple and universal fish video image track tracking method | |
CN109255336A (en) | Arrester recognition methods based on crusing robot | |
CN111738114A (en) | Vehicle target detection method based on anchor-free accurate sampling remote sensing image | |
CN104391294A (en) | Connection domain characteristic and template matching based radar plot correlation method | |
CN108983194B (en) | Target extraction and condensation method based on ground monitoring radar system | |
CN112068085B (en) | Radar sea clutter original data rapid preprocessing method based on deep learning | |
CN107239761B (en) | Fruit tree branch pulling effect evaluation method based on skeleton angular point detection | |
CN104008365A (en) | Method for detecting sparse degree of fruit tree leaves based on image processing technology | |
CN109544574B (en) | Target extraction method based on all-solid-state VTS radar |
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: 20170510 |
|
RJ01 | Rejection of invention patent application after publication |