CN103473532B - The quick pedestrian detecting system of airborne platform and method - Google Patents
The quick pedestrian detecting system of airborne platform and method Download PDFInfo
- Publication number
- CN103473532B CN103473532B CN201310400824.3A CN201310400824A CN103473532B CN 103473532 B CN103473532 B CN 103473532B CN 201310400824 A CN201310400824 A CN 201310400824A CN 103473532 B CN103473532 B CN 103473532B
- Authority
- CN
- China
- Prior art keywords
- gradient
- image
- airborne
- pedestrian
- value
- 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
- 238000000034 method Methods 0.000 title claims abstract description 17
- 238000001514 detection method Methods 0.000 claims abstract description 38
- 235000018734 Sambucus australis Nutrition 0.000 claims abstract description 24
- 244000180577 Sambucus australis Species 0.000 claims abstract description 24
- OGFXBIXJCWAUCH-UHFFFAOYSA-N meso-secoisolariciresinol Natural products C1=2C=C(O)C(OC)=CC=2CC(CO)C(CO)C1C1=CC=C(O)C(OC)=C1 OGFXBIXJCWAUCH-UHFFFAOYSA-N 0.000 claims abstract description 24
- 238000012545 processing Methods 0.000 claims abstract description 23
- HPTJABJPZMULFH-UHFFFAOYSA-N 12-[(Cyclohexylcarbamoyl)amino]dodecanoic acid Chemical compound OC(=O)CCCCCCCCCCCNC(=O)NC1CCCCC1 HPTJABJPZMULFH-UHFFFAOYSA-N 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000012937 correction Methods 0.000 claims description 5
- 230000005055 memory storage Effects 0.000 claims 1
- 230000000875 corresponding effect Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000012544 monitoring process Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 230000006399 behavior Effects 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Landscapes
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of quick pedestrian detecting system of airborne platform and method.Including high-definition camera, it is connected to an airborne quick processing module SECO CARMA DevKit；After high-definition camera takes in high-definition image, by USB interface, image is passed to airborne quick processing module SECO CARMA DevKit, the high-definition image of Real-time Collection is carried out real-time pedestrian detection.The method of the present invention is the characteristic vector based on gradient orientation histogram first having to calculate the pedestrian in picture, including color space standards, calculates gradient, the gradient statistics in space and direction, and the contrast standardization in overlapping block, characteristic vector generates；It is then based on linear SVM characteristic vector to be classified, differentiates in picture containing pedestrian further according to classification results.The pedestrian detection that embodiments of the invention are mainly used in image calculates, and the particularly pedestrian detection in mobile robot embedded device calculates.
Description
Technical field
The invention discloses a kind of quick pedestrian detecting system of airborne platform and method, relate to robot vision, pattern is known
Other technology and CUDA parallel computation field.
Background technology
It is that necessary to intelligent vehicle DAS (Driver Assistant System) in the future, it can assist in urban district effectively that pedestrian carries out detecting
The driver's environment the most to external world driving vehicle in environment is made a response, it is to avoid collision pedestrian, reduces sending out of vehicle accident
Raw, particularly nighttime driving person visual field finite sum fatigue etc. reason.In addition, human detection can be also used for video monitoring
Real-time security system analyze from continuous print video clips and detect the behavior of intrusion, ensure individual and public the person property
Safety.
Pedestrian detection is widely used in public safety, video monitoring, and intelligence auxiliary driving technology and traffic monitoring etc. are
System.At present in terms of pedestrian detection research, gradient orientation histogram HOG algorithm has and surpassingly shows, and it can extract exactly
Pedestrian target in picture and video.Good for pedestrian detection robustness and uniqueness based on gradient orientation histogram, but
Join speed slow.
At present, the image algorithm that direct computing is complicated on airborne platform, now airborne platform needs to carry exclusive
FPGA hardware, and algorithm is solidificated on relevant hardware, and carry out the optimization being correlated with, reduce economy；Or move robot
In the image servo of airborne platform, then image information is processed to upper server machine by network delivery, so
After will process after result be transferred on airborne processor, timeliness reduce.
The resolution great majority making image procossing on airborne platform are 640 × 480, cannot be carried out in real time for high-definition image
Process.Due to computationally intensive, data are many, and the industrial control computer carried on the mobile man-machine carrying platform of machine cannot meet directly
Connect the calculating making pedestrian detection.
Summary of the invention
In order to overcome above-mentioned the deficiencies in the prior art, the invention provides a kind of quick pedestrian detecting system of airborne platform and
Method, solves the problem that existing high definition real time imaging pedestrian detection processing speed is slow or needs dedicated hardware equipment.
In order to achieve the above object, insight of the invention is that the image first being gathered high definition by image input system；Then
Image is sent to SECO CARMA DevKit embedded type C UDA hardware and software platform the high-definition image of Real-time Collection is carried out in real time
Pedestrian detection.
The quick pedestrian detecting system of airborne platform of the present invention includes:
(1) high-definition image input, ARTAM-1400MI-USB3 high-definition camera is passed on processor by USB interface；
(2) fast image processing system: by SECO CARMA DevKit embedded type C UDA hardware and software platform parallel computation
Technology, carries out real-time pedestrian detection to the high-definition image of Real-time Collection.
Conceiving according to foregoing invention, the present invention uses following technical proposals:
A kind of quick pedestrian detecting system of airborne platform, including high-definition camera, it is characterised in that: described high-definition camera
It is connected to an airborne quick processing module SECO CARMA DevKit；After described high-definition camera takes in high-definition image, pass through
Image is passed to airborne quick processing module SECO CARMA DevKit, described quick processing module SECO by USB interface
CARMA DevKit carries out real-time pedestrian detection to the high-definition image of Real-time Collection.
A kind of airborne platform rapid pedestrian detection method, uses the quick pedestrian of airborne platform according to claim 1
Detecting system detects, it is characterised in that detecting step is as follows:
Step 1: by real-time incoming airborne quick processing module SECO CARMA after high-definition camera picked-up image
DevKit；
Step 2: input picture is handled as follows by airborne quick processing module SECO CARMA DevKit:
Color space standards: the transfer function that the picture read in from photographic head directly utilizes CUDA is passed directly to SECO
On CARMA DevKit embedded gpu and process realization, it is simply that read the value of each pixel of data in picture, obtain pixel
Value after square root just can be utilized to carry out Gamma correction, carry out the most as follows:,
WhereinRepresent the image pixel value after Gamma correction,) represent the value of non-timing；
Calculate gradient: this is in the kernel function started on GPU, and coloured image to enter on tri-passages of G, B respectively at R
Row calculates, and can obtain the amplitude of the gradient of three passages and three deflections, now takes the channel value that gradient magnitude is maximum
With corresponding deflection.Each thread block comprises 256 × 1 × 1 thread, and each thread calculates a gradient, and mould and direction, with height
The information of shared drive (Shared Memory) storage pixel on speed sheet.When coloured image calculates, distribute to each thread block
(256+2) × 3 bytes sizes shared EMS memory spaces, numerical value 2 mainly considers that border issue, numerical value 3 are to be able to store three
The result of calculation of passage, specific formula for calculation is as follows:
Wherein,Represent the gradient in x, y direction respectively,Represent gradient direction；
The gradient statistics in space and direction: utilizeThe gradient that result is asked for, each thread calculates alone rectangular histogram, knot
Fruit is stored in shared memory cell, gradient obtain gradient orientation histogram, following steps generation histograms of oriented gradients: 1) really
FixedGrad distribution space；2) value in incremental gradient Distribution value space；3) 1 is repeated) and 2)；
Contrast standardization in overlapping block: rectangular histogram one large-scale Nogata of composition that all thread block are generated
Figure, is normalized the most in the following manner, and each thread calculates corresponding HOG block, calculates and carries out as follows:, wherein v represents vector,Represent the norm of vector v,Represent the least value；
Characteristic vector generates: in the detection window of 128 × 64 pixels, it has been found that best result be use 16 ×
16 block of pixels, 128 × 64 pixels are formed by 15 × 7 pieces.Each piece contains 4 sub-blocks, and it is mono-that each sub-block comprises again a Cell
Unit, and a Cell unit is made up of 8 × 8 pixels.Each Cell unit can generate 9bins, each piece will generate 4 ×
9 bins.So the dimension of characteristic vector is 3780 in final 128 × 64 detection windows；
Based on linear vector machine SVM, characteristic vector is classified.
Step 3: by airborne quick processing module SECO CARMA DevKit real-time output detections pedestrian's result.
Step in described step 2Based on linear SVM by the method that characteristic vector carries out classifying it is: based on line
Property support vector machine characteristic vector is classified: detection window has 64 × 128 pixels, is formed by 7 × 15 pieces, by figure
The width of picture, highly, the stride of window and block calculates the number of the block of CUDA, and each detection window is mapped to a thread block
In.Each histogrammic value is multiplied by SVM weight, then adds weighted value with parallel reducing method, then deduct the biasing of hyperplane,
Finally result of calculation is write back in the global memory of GPU.
The present invention compared with prior art, has following obvious prominent substantive distinguishing features and remarkable advantage: this
Bright on airborne platform can to input high-definition image be quickly detected from pedestrian.So there is no need to airborne exclusive figure
As processing hardware and the hardware algorithm of optimization, and quickly image pedestrian's detecting system is deployed on mobile robot platform.
Accompanying drawing explanation
Fig. 1 is the software and hardware frame diagram of the system of the present invention；
Fig. 2 is system hardware figure；
Fig. 3 is pedestrian detection block diagram；
Fig. 4 is Detection results figure.
Detailed description of the invention
Below in conjunction with the accompanying drawings the preferred embodiment in the present invention is clearly and completely described, it is clear that described reality
Execute a part of embodiment that example is only the present invention.
Embodiment one:
Seeing Fig. 1 and Fig. 2, the quick pedestrian detecting system of this airborne platform, including high-definition camera (1), it is characterised in that:
Described high-definition camera (1) is connected to an airborne quick processing module SECO CARMA DevKit(2)；Described high-definition camera
Machine (1) is taken in after high-definition image, by USB interface, image is passed to airborne quick processing module SECO CARMA DevKit
(2), described quick processing module SECO CARMA DevKit(2) high-definition image of Real-time Collection is carried out real-time pedestrian's inspection
Survey.
Embodiment two:
See Fig. 1, Fig. 3 and Fig. 4, this airborne platform rapid pedestrian detection method, use said system to detect, detection
Step is as follows:
Step 1: after being taken in image by high-definition camera (1), inputs airborne quick processing module SECO CARMA in real time
DevKit(2);
Step 2: by airborne quick processing module SECO CARMA DevKit(2) input picture is handled as follows:
Color space standards: the transfer function that the picture read in from photographic head directly utilizes CUDA is passed directly to
On SECO CARMA DevKit embedded gpu and process realization, it is simply that read the value of each pixel of data in picture, obtain
Square root just can be utilized after the value of capture vegetarian refreshments to carry out Gamma correction, carry out the most as follows:, whereinRepresent the image pixel value after Gamma correction,Represent not
The value of timing；
Calculate gradient: this is in the kernel function started on GPU, and coloured image to enter on tri-passages of G, B respectively at R
Row calculates, and can obtain the amplitude of the gradient of three passages and three deflections, now takes the channel value that gradient magnitude is maximum
With corresponding deflection.Each thread block comprises 256 × 1 × 1 thread, and each thread calculates a gradient, and mould and direction, with height
The information of shared drive (Shared Memory) storage pixel on speed sheet.When coloured image calculates, distribute to each thread block
(256+2) × 3 bytes sizes shared EMS memory spaces, numerical value 2 mainly considers that border issue, numerical value 3 are to be able to store three
The result of calculation of passage, specific formula for calculation is as follows:
Wherein,Represent the gradient in x, y direction respectively,Represent gradient direction；
The gradient statistics in space and direction: utilizeThe gradient that result is asked for, each thread calculates alone rectangular histogram, knot
Fruit is stored in shared memory cell, gradient obtain gradient orientation histogram, following steps generation histograms of oriented gradients: 1) really
FixedGrad distribution space；2) value in incremental gradient Distribution value space；3) 1 is repeated) and 2)；
Contrast standardization in overlapping block: rectangular histogram one large-scale Nogata of composition that all thread block are generated
Figure, is normalized the most in the following manner, and each thread calculates corresponding HOG block, calculates and carries out as follows:, wherein v represents vector,Represent the norm of vector v,Represent the least value；
Characteristic vector generates: in the detection window of 128 × 64 pixels, it has been found that best result be use 16 ×
16 block of pixels, 128 × 64 pixels are formed by 15 × 7 pieces.Each piece contains 4 sub-blocks, and it is mono-that each sub-block comprises again a Cell
Unit, and a Cell unit is made up of 8 × 8 pixels.Each Cell unit can generate 9bins, each piece will generate 4 ×
9 bins.So the dimension of characteristic vector is 3780 in final 128 × 64 detection windows；
Based on linear SVM, characteristic vector is classified.
Step 3: by airborne quick processing module SECO CARMA DevKit(2) real-time output detections pedestrian's result.
Step in described step 2Based on linear SVM by the method that characteristic vector carries out classifying it is: detection window
Having 64 × 128 pixels in Kou, formed by 7 × 15 pieces, by the width of image, highly, the stride of window and block calculates CUDA
The number of block, each detection window is mapped in a thread block.Each histogrammic value is multiplied by SVM weight, then
Add weighted value with parallel reducing method, then deduct the biasing of hyperplane, finally result of calculation is write back to the global memory of GPU
In.According to the position at the pedestrian place in picture, use square frame labelling.So far, the present frame figure to video camera input is just completed
The pedestrian detection of sheet.
The pedestrian detection that embodiments of the invention are mainly used in image calculates, the particularly pedestrian in embedded device
Detection calculates.
The above, the only detailed description of the invention of the present invention, but protection scope of the present invention is not only limited to this, and any
Those of ordinary skill in the art in the technical scope that the invention discloses, the variations and alternatives that can readily occur in, all should contain
Within protection scope of the present invention.Therefore, protection scope of the present invention should be described and is as the criterion with scope of the claims.
Claims (2)
1. an airborne platform rapid pedestrian detection method, the quick pedestrian detecting system of airborne platform of employing detects, should
Detecting system includes that high-definition camera (1), described high-definition camera (1) are connected to airborne quick processing module SECO
CARMA DevKit(2)；Described high-definition camera (1) is taken in after high-definition image, by USB interface image is passed to airborne soon
Speed processing module SECO CARMA DevKit(2), described quick processing module SECO CARMA DevKit(2) to adopting in real time
The high-definition image of collection carries out real-time pedestrian detection；It is characterized in that, detecting step is as follows:
Step 1: by real-time incoming airborne quick processing module SECO CARMA DevKit after high-definition camera (1) picked-up image
(2)；
Step 2: by airborne quick processing module SECO CARMA DevKit(2) input picture is handled as follows:
Color space standards: picture is passed directly to parallelization on CUDA and processes realization, it is simply that obtain each of image data
The value of pixel, just can utilize square root Gamma to correct, carry out the most as follows after getting the value of pixel:, whereinRepresent the image pixel value after Gamma correction,) table
Show the value of non-timing；
Calculate gradient: utilizing each thread block of CUDA to comprise 256 × 1 × 1 thread, each thread calculates a gradient, mould and side
To, by the information of shared drive Shared Memory storage pixel on high-speed chip, so it is calculated the gradient of three passages
Amplitude and three deflections, now take the maximum channel value of gradient magnitude and corresponding deflection, and concrete calculating is by following public
Formula is carried out:
Wherein,Represent the gradient in x, y direction respectively,Represent gradient direction；
The gradient statistics in space and direction: utilize stepResult asks for gradient, and each thread calculates alone rectangular histogram, result
It is stored in shared memory cell, gradient obtains gradient orientation histogram, following steps generation histograms of oriented gradients: 1) determine
θ (x, y) Grad distribution space；2) value in incremental gradient Distribution value space；3) step 1) and step 2 are repeated)；
Contrast standardization in overlapping block: rectangular histogram one large-scale rectangular histogram of composition that all thread block are generated, afterwards
Being normalized in the following manner, each thread calculates corresponding HOG block, calculates and carries out as follows:, wherein v represents vector,Represent the norm of vector v,Represent the least value；
Characteristic vector generates: in the detection window of 128 × 64 pixels, best result is use 16 × 16 block of pixels, and 128
× 64 pixels are formed by 15 × 7 pieces；Each piece contains 4 sub-blocks, and each sub-block comprises again a Cell unit, and a Cell
Unit is made up of 8 × 8 pixels；Each Cell unit can generate 9bins, each piece will generate 4 × 9 bins；So
In final 128 × 64 detection windows, the dimension of characteristic vector is 3780；
Based on linear vector machine SVM, characteristic vector is classified；
Step 3: by airborne quick processing module SECO CARMA DevKit(2) real-time output detections pedestrian's result.
Airborne platform rapid pedestrian detection method the most according to claim 1, it is characterised in that described stepIn based on
The method that characteristic vector carries out classifying is by linear vector machine SVM: have 64 × 128 pixels in detection window, by 7 × 15 pieces of groups
Becoming, by the width of image, highly, the stride of window and block calculates the number of the block of CUDA, is mapped to by each detection window
In one thread block；Each histogrammic value is multiplied by SVM weight, then adds weighted value with parallel reducing method, then deduct super
The biasing of plane, finally writes back to result of calculation in the global memory of GPU.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310400824.3A CN103473532B (en) | 2013-09-06 | 2013-09-06 | The quick pedestrian detecting system of airborne platform and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310400824.3A CN103473532B (en) | 2013-09-06 | 2013-09-06 | The quick pedestrian detecting system of airborne platform and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103473532A CN103473532A (en) | 2013-12-25 |
| CN103473532B true CN103473532B (en) | 2016-11-23 |
Family
ID=49798377
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310400824.3A Expired - Fee Related CN103473532B (en) | 2013-09-06 | 2013-09-06 | The quick pedestrian detecting system of airborne platform and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103473532B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103903279B (en) * | 2014-03-21 | 2017-07-25 | 上海大学 | Parallel Tracking System and method for based on bionic binocular vision airborne platform |
| CN103996181B (en) * | 2014-05-12 | 2017-06-23 | 上海大学 | A kind of big view field image splicing system and method based on bionical eyes |
| CN104036477A (en) * | 2014-06-06 | 2014-09-10 | 上海大学 | Large-view-field image splicing device and method based on two biomimetic eyes |
| CN105574488B (en) * | 2015-12-07 | 2019-03-22 | 北京航空航天大学 | It is a kind of to be taken photo by plane the pedestrian detection method of infrared image based on low latitude |
| CN107316320A (en) * | 2017-06-19 | 2017-11-03 | 江西洪都航空工业集团有限责任公司 | The real-time pedestrian detecting system that a kind of use GPU accelerates |
| CN109063683B (en) * | 2018-08-28 | 2022-02-11 | 第六镜科技(北京)集团有限责任公司 | Face detection method and system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN202042051U (en) * | 2011-03-24 | 2011-11-16 | 浙江金之路信息科技有限公司 | Urban traffic intelligent supervision system |
| CN103177455A (en) * | 2013-03-20 | 2013-06-26 | 南京理工大学 | Method for realizing KLT (Karhunen Loeve Transform) moving target tracking algorithm based on multicore DSP (Digital Signal Processor) |
-
2013
- 2013-09-06 CN CN201310400824.3A patent/CN103473532B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN202042051U (en) * | 2011-03-24 | 2011-11-16 | 浙江金之路信息科技有限公司 | Urban traffic intelligent supervision system |
| CN103177455A (en) * | 2013-03-20 | 2013-06-26 | 南京理工大学 | Method for realizing KLT (Karhunen Loeve Transform) moving target tracking algorithm based on multicore DSP (Digital Signal Processor) |
Non-Patent Citations (3)
| Title |
|---|
| "HOG特征混合模型结合隐SVM的感兴趣目标检测定位算法";胡正平等;《信号处理》;20110831;第27卷(第8期);全文 * |
| "方向梯度直方图及其扩展";傅红普等;《计算机工程》;20130515;第39卷(第5期);全文 * |
| "目标检测的图像特征提取之（一）HOG 特征";zouxy9;《www.blog.csdn.net/zouxy9/article/details/7929348》;20120831;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103473532A (en) | 2013-12-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103473532B (en) | The quick pedestrian detecting system of airborne platform and method | |
| Zhou et al. | Safety helmet detection based on YOLOv5 | |
| Bauer et al. | FPGA-GPU architecture for kernel SVM pedestrian detection | |
| Hahnle et al. | FPGA-based real-time pedestrian detection on high-resolution images | |
| CN105426905B (en) | Robot barrier object recognition methods based on histogram of gradients and support vector machines | |
| CN102982341B (en) | Self-intended crowd density estimation method for camera capable of straddling | |
| US10096091B2 (en) | Image generating method and apparatus | |
| JP6398979B2 (en) | Video processing apparatus, video processing method, and video processing program | |
| Sakla et al. | Deep multi-modal vehicle detection in aerial ISR imagery | |
| CN103530640B (en) | Unlicensed vehicle checking method based on AdaBoost Yu SVM | |
| CN103345631B (en) | Image characteristics extraction, training, detection method and module, device, system | |
| Zhang et al. | A pedestrian detection method based on SVM classifier and optimized Histograms of Oriented Gradients feature | |
| CN109492609B (en) | Method for detecting lane line, vehicle and computing equipment | |
| CN103390164A (en) | Object detection method based on depth image and implementing device thereof | |
| CN105404857A (en) | Infrared-based night intelligent vehicle front pedestrian detection method | |
| CN109334563A (en) | A kind of anticollision method for early warning based on road ahead pedestrian and bicyclist | |
| EP2813973B1 (en) | Method and system for processing video image | |
| Kavitha et al. | Pothole and object detection for an autonomous vehicle using yolo | |
| CN105740751A (en) | Object detection and identification method and system | |
| CN110197104B (en) | Distance measurement method and device based on vehicle | |
| CN112101134B (en) | Object detection method and device, electronic equipment and storage medium | |
| Shi et al. | Cobev: Elevating roadside 3d object detection with depth and height complementarity | |
| Zhou et al. | A pipeline architecture for traffic sign classification on an FPGA | |
| CN117523612A (en) | Dense pedestrian detection method based on Yolov5 network | |
| CN116883870A (en) | Unmanned aerial vehicle falling stone target detection method and device based on deep 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 | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20161123 |