CN106952278A - A kind of automatic division method in dynamic outdoor environment based on super-pixel - Google Patents
A kind of automatic division method in dynamic outdoor environment based on super-pixel Download PDFInfo
- Publication number
- CN106952278A CN106952278A CN201710216762.9A CN201710216762A CN106952278A CN 106952278 A CN106952278 A CN 106952278A CN 201710216762 A CN201710216762 A CN 201710216762A CN 106952278 A CN106952278 A CN 106952278A
- Authority
- CN
- China
- Prior art keywords
- pixel
- super
- image
- background
- color
- 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.)
- Withdrawn
Links
Abstract
Automatic division method in a kind of dynamic outdoor environment based on super-pixel proposed in the present invention, its main contents include：Calculate low texture super-pixel, threshold value tone images are generated using maximum variance between clusters, estimate background distributions, pixel tag is distributed, create mask, its process is, first convert the image into tone intensity value color space, and it is divided into super-pixel, super-pixel is calculated using the energy driving method of sampling, reuse maximum variance between clusters generation threshold value tone images, the probability distribution of background pixel is estimated followed by super-pixel group, tone images are compared with distribution, generate pixel tag, finally create the mask image in the region for including object of interest and background object.The present invention utilizes low texture super-pixel, generates threshold value tone images using maximum variance between clusters so that also can accurately split under conditions of lighting condition is unstable, and algorithm is simple, and the degree of accuracy also increases.
Description
Technical field
The present invention relates to image segmentation field, more particularly, in a kind of dynamic outdoor environment based on super-pixel from
Dynamic dividing method.
Background technology
In the research and application of image, people are often interested in some of image part, these parts interested
Corresponding to the specific, region with special nature in image, we term it target or prospect, other parts turn into image
Background.In order to recognize and resolution target, it would be desirable to which target is isolated out, here it is image segmentation from image.The technology
It agriculturally can be used for many aspects such as resource investigation, agricultural projects, crop yield estimation, pest and disease damage detection, forest survey；
Medically, the segmentation to focus is can be used for again to extract, so that for clinical research and treatment.But in existing technology
In, because data acquisition is frequently in outdoor progress, lighting condition is unstable, therefore image can be affected by it, so as to segmentation
Cause certain difficulty.
The present invention proposes the automatic division method in a kind of dynamic outdoor environment based on super-pixel, first changes image
For tone intensity value color space, and super-pixel is divided into, calculates super-pixel using the energy driving method of sampling, then make
Threshold value tone images are generated with maximum variance between clusters, the probability distribution of background pixel are estimated followed by super-pixel group, by color
Change the line map as being compared with distribution, generate pixel tag, finally establishment includes covering for the region of object of interest and background object
Code image.The present invention utilizes low texture super-pixel, and threshold value tone images are generated using maximum variance between clusters so that in illumination bar
Also can accurately it split under conditions of part is unstable, and algorithm is simple, and the degree of accuracy also increases.
The content of the invention
For lighting condition it is unstable the problem of, it is an object of the invention to provide a kind of dynamic based on super-pixel is outdoor
Automatic division method in environment, first converts the image into tone intensity value color space, and is divided into super-pixel, makes
Super-pixel is calculated with the energy driving method of sampling, maximum variance between clusters generation threshold value tone images are reused, followed by super
Pixel groups estimate the probability distribution of background pixel, and tone images are compared with distribution, generate pixel tag, finally create bag
Include the mask image in the region of object of interest and background object.
To solve the above problems, the present invention provides the automatic segmentation side in a kind of dynamic outdoor environment based on super-pixel
Method, its main contents include：
(1) low texture super-pixel is calculated；
(2) threshold value tone images are generated using maximum variance between clusters；
(3) background distributions are estimated；
(4) pixel tag is distributed；
(5) mask is created.
Wherein, described automatic division method, tone intensity value (HSV) color space of input picture, output segmentation
Image；Algorithm steps are as follows：
(1) one group of super-pixel is calculated using energy driving sampling (SEEDS) methodAnd find subsetLow-quality sense
Super-pixel；
(2) threshold process, generation binary picture T are carried out to form and aspect passage using maximum variance between clusters；
(3) it is based onDetermine the distribution of background；
(4) basisLabel is distributed for each output pixel；
(5) mask is created to eliminate the region outside background object.
Wherein, the low texture super-pixel of described calculating, converts the image into HSV color spaces, and be divided into super picture
Element；Super-pixel is calculated using SEEDS methods；In the super-pixel method, image is divided into lattice, surpasses as initial
Pixel is distributed；Super-pixel is improved by its border of iterative modifications to distribute.
Further, described SEEDS algorithms, set the parameter of SEEDS algorithms, low texture region is had and original net
Lattice distribute shape identical super-pixel, evenThe one group of super-pixel generated for SEEDS, then have one group of super-pixelShape
For rectangle, it corresponds to the constant super-pixel of low texture region；Corresponding to the smallest blocks of SEEDS algorithms, it can be based on figure
The size of picture and determined for the SEEDS parameters selected.
Wherein, described use maximum variance between clusters generation threshold value tone images, using maximum variance between clusters, pass through
The tone passages of thresholding HSV images generates binary picture T；In binary picture, value is black higher than the pixel of threshold value
Color, rest of pixels is white；Image T be used to producing low texture region in image it is assumed that because the face of low-quality sense background object
Color relative constancy；In the application, the color value of blue background object is than other common colors (such as brown, grey or green)
Color value it is low；As long as the tone value of most of pixel is different from the color-values of foreground object in background, then multistage can be used
Thresholding algorithm generates T.
Wherein, described estimation background distributions, using algorithm determine T withIn the overlapping region of super-pixel, i.e. algorithm time
Go throughIn super-pixel；If threshold binary image T-phase answers the percentage exceedance ζ of the white pixel in region, super-pixel is added
It is added to setIn；SetThe low texture region and all super-pixel of its color constancy determined by SEEDS algorithms is constituted；
This group of super-pixelIn all pixels position be used to estimate the probability distribution of background pixel, as in color-phase diagram picture
The value of ith pixel；Assuming that image pixel intensities according toNormal distribution, i.e. its probability density byProvide；Distributed constant μ and σ fromSample mean and pixel sample variance in obtain；Will
The minimum value of sample variance is set to σmin=4.
Further, described algorithm, it is used for determining background pixel, and low texture super-pixel is set during inputAnd threshold value
Tone images T；It is output as setting background super-pixelFor each super-pixelIn riT in the T of regioniRepresent white
Pixel count；If ti/area(ri)>ζ, then
Wherein, described pixel tag distribution, by tone images and distributionIt is compared, generates pixel tag
L image；Make FiTo indicate stochastic variable, if the ith pixel in color-phase diagram picture belongs to object of interest, for 0, otherwise
For 0；According to p (Fi|hi)=φ, condition is hiStochastic variable FiIt is bernoulli formula distribution, wherein parameterThe label of image L ith pixel is by 255 × p (Fi|hi) provide.
Wherein, described establishment mask, creates mask image M, including object of interest and the region of background object,
Them labeled as white, other all zone markers are black；Initially,In all pixels and pixel i marked in M
For white, wherein p (hi)>pm；The connection component of threshold value less than M is removed from M；Then, continuous dilation procedure is carried out to M, directly
Component is connected to only one of which；The quantity of required dilation procedure is expressed as d；Profile, convex hull and convex defect are from connection
Calculated in component；Fill more than threshold value (∈d) depth convex defect (solstics i.e. in convex defect and convex hull it
Between distance) and all in-profiles；Finally, d erosion of mask is returned into original size.
Further, described mask, pixel tag image L is applied to by mask, should if M ith pixel is 0
Pixel is outside area-of-interest, and the respective pixel in label L；In this application, mark value is 0 pixel, meanwhile,
Pixel in region interested is marked as 255 × p (Fi|hi)。
Brief description of the drawings
Fig. 1 is the system flow chart of the automatic division method in a kind of dynamic outdoor environment based on super-pixel of the present invention.
Fig. 2 is that the low texture of calculating of the automatic division method in a kind of dynamic outdoor environment based on super-pixel of the present invention surpasses
Pixel.
Fig. 3 is the generation threshold value tone of the automatic division method in a kind of dynamic outdoor environment based on super-pixel of the present invention
Image.
Fig. 4 is the estimation background point of the automatic division method in a kind of dynamic outdoor environment based on super-pixel of the present invention
Cloth.
Fig. 5 is the pixel tag distribution of the automatic division method in a kind of dynamic outdoor environment based on super-pixel of the present invention
With establishment mask.
Embodiment
It should be noted that in the case where not conflicting, the feature in embodiment and embodiment in the application can phase
Mutually combine, the present invention is described in further detail with specific embodiment below in conjunction with the accompanying drawings.
Fig. 1 is the system flow chart of the automatic division method in a kind of dynamic outdoor environment based on super-pixel of the present invention.
It is main to include calculating low texture super-pixel, threshold value tone images are generated using maximum variance between clusters, background distributions, pixel are estimated
Label distributes and created mask.
Automatic division method, tone intensity value (HSV) color space of input picture exports segmentation figure picture；Algorithm steps
It is rapid as follows：
(1) one group of super-pixel is calculated using energy driving sampling (SEEDS) methodAnd find subsetLow-quality sense
Super-pixel；
(2) threshold process, generation binary picture T are carried out to form and aspect passage using maximum variance between clusters；
(3) it is based onDetermine the distribution of background；
(4) basisLabel is distributed for each output pixel；
(5) mask is created to eliminate the region outside background object.
Fig. 2 is that the low texture of calculating of the automatic division method in a kind of dynamic outdoor environment based on super-pixel of the present invention surpasses
Pixel.HSV color spaces are converted the image into, and are divided into super-pixel；Super-pixel is calculated using SEEDS methods；At this
In super-pixel method, image is divided into lattice, is distributed as initial super-pixel；Improved by its border of iterative modifications
Super-pixel is distributed.
The parameter of SEEDS algorithms is set, there is low texture region and distribute shape identical super-pixel with initial mesh, i.e.,
OrderThe one group of super-pixel generated for SEEDS, then have one group of super-pixelBe shaped as rectangle, its correspond to low texture area
The constant super-pixel in domain；Corresponding to the smallest blocks of SEEDS algorithms, size and be SEEDS selections that it can be based on image
Parameter is determined.
Fig. 3 is the generation threshold value tone of the automatic division method in a kind of dynamic outdoor environment based on super-pixel of the present invention
Image.Using maximum variance between clusters, binary picture T is generated by the tone passage of thresholding HSV images；In binary system
In image, value is black higher than the pixel of threshold value, and rest of pixels is white；Image T is used to produce low texture region in image
It is assumed that because the color relative constancy of low-quality sense background object；In the application, the color value of blue background object is than other
The color value of common colors (such as brown, grey or green) is low；As long as the tone value and foreground object of most of pixel in background
Color-values it is different, then multilevel threshold algorithm can be used to generate T.
Fig. 4 is the estimation background point of the automatic division method in a kind of dynamic outdoor environment based on super-pixel of the present invention
Cloth.Using algorithm determine T withIn the overlapping region of super-pixel, i.e. algorithm travels throughIn super-pixel；If threshold binary image T
The percentage exceedance ζ of white pixel in respective regions, then super-pixel be added to setIn；SetBy SEEDS algorithms
The low texture region and all super-pixel of its color constancy determined is constituted；
This group of super-pixelIn all pixels position be used to estimate the probability distribution of background pixel, as in color-phase diagram picture
The value of ith pixel；Assuming that image pixel intensities according toNormal distribution, i.e. its probability density byProvide；Distributed constant μ and σ fromSample mean and pixel sample variance in obtain；Will
The minimum value of sample variance is set to σmin=4.
Background pixel is determined with the algorithm, low texture super-pixel is set during inputWith threshold value tone images T；It is output as
Background super-pixel is setFor each super-pixelIn riT in the T of regioniRepresent white pixel number；If ti/
area(ri)>ζ, then
Fig. 5 is the pixel tag distribution of the automatic division method in a kind of dynamic outdoor environment based on super-pixel of the present invention
With establishment mask.By tone images and distributionIt is compared, generation pixel tag L image；Make FiIt is random to indicate
Variable, is otherwise 0 for 0 if the ith pixel in color-phase diagram picture belongs to object of interest；According to p (Fi|hi)=φ, bar
Part is hiStochastic variable FiIt is bernoulli formula distribution, wherein parameterImage L ith pixel
Label by 255 × p (Fi|hi) provide.
Mask image M is created, including object of interest and the region of background object, they are labeled as white, its
Its all zone marker is black；Initially,In all pixels and pixel i marked in M and be, wherein p (hi)>pm；
The connection component of threshold value less than M is removed from M；Then, continuous dilation procedure is carried out to M, until only one of which connects component；
The quantity of required dilation procedure is expressed as d；Profile, convex hull and convex defect are calculated from the component of connection；Fill out
Fill more than threshold value (∈d) depth convex defect the distance between (solstics i.e. in convex defect with convex hull) and it is all in
Contouring；Finally, d erosion of mask is returned into original size.
By mask be applied to pixel tag image L, if M ith pixel be 0, the pixel outside area-of-interest,
And the respective pixel in label L；In this application, mark value is 0 pixel, meanwhile, the pixel in region interested
It is marked as 255 × p (Fi|hi)。
For those skilled in the art, the present invention is not restricted to the details of above-described embodiment, in the essence without departing substantially from the present invention
In the case of refreshing and scope, the present invention can be realized with other concrete forms.In addition, those skilled in the art can be to this hair
Bright to carry out various changes and modification without departing from the spirit and scope of the present invention, these are improved and modification also should be regarded as the present invention's
Protection domain.Therefore, appended claims are intended to be construed to include preferred embodiment and fall into all changes of the scope of the invention
More and modification.
Claims (10)
1. the automatic division method in a kind of dynamic outdoor environment based on super-pixel, it is characterised in that main to include calculating low
Texture super-pixel (one)；Use maximum variance between clusters generation threshold value tone images (two)；Estimate background distributions (three)；Pixel mark
Label distribution (four)；Create mask (five).
2. based on the automatic division method described in claims 1, it is characterised in that the tone intensity value of input picture
(HSV) color space, exports segmentation figure picture；Algorithm steps are as follows：
(1) one group of super-pixel is calculated using energy driving sampling (SEEDS) methodAnd find subsetThe super picture of low-quality sense
Element；
(2) threshold process, generation binary picture T are carried out to form and aspect passage using maximum variance between clusters；
(3) it is based onDetermine the distribution of background；
(4) basisLabel is distributed for each output pixel；
(5) mask is created to eliminate the region outside background object.
3. based on the low texture super-pixel (one) of calculating described in claims 1, it is characterised in that convert the image into HSV colors
Color space, and it is divided into super-pixel；Super-pixel is calculated using SEEDS methods；In the super-pixel method, image is divided
For lattice, distributed as initial super-pixel；Super-pixel is improved by its border of iterative modifications to distribute.
4. based on the SEEDS algorithms described in claims 3, it is characterised in that set the parameter of SEEDS algorithms, make low texture
Region has distributes shape identical super-pixel with initial mesh, evenThe one group of super-pixel generated for SEEDS, then have one group
Super-pixelBe shaped as rectangle, it corresponds to the constant super-pixel of low texture region；Corresponding to SEEDS algorithms most
Fritter, it can be determined based on the size of image and for the parameter of SEEDS selections.
5. based on the use maximum variance between clusters generation threshold value tone images (two) described in claims 1, it is characterised in that
Using maximum variance between clusters, binary picture T is generated by the tone passage of thresholding HSV images；In binary picture
In, value is black higher than the pixel of threshold value, and rest of pixels is white；Image T be used to producing low texture region in image it is assumed that
Because the color relative constancy of low-quality sense background object；In the application, the color value of blue background object is more common than other
The color value of color (such as brown, grey or green) is low；As long as the tone value and the color of foreground object of most of pixel in background
Coloured silk value is different, then multilevel threshold algorithm can be used to generate T.
6. based on the estimation background distributions (three) described in claims 1, it is characterised in that using algorithm determine T withIn it is super
The overlapping region of pixel, i.e. algorithm are traveled throughIn super-pixel；If threshold binary image T-phase answers the percentage of the white pixel in region
Than exceedance ζ, then super-pixel is added to setIn；SetThe low texture region determined by SEEDS algorithms and its constant face
All super-pixel composition of color；
This group of super-pixelIn all pixels position be used to estimate the probability distribution of background pixel, as i-th in color-phase diagram picture
The value of pixel；Assuming that image pixel intensities according toNormal distribution, i.e. its probability density by
Provide；Distributed constant μ and σ fromSample mean and pixel sample variance in obtain；The minimum value of sample variance is set
It is set to σmin=4.
7. based on the algorithm described in claims 6, it is characterised in that it is used for determining background pixel, and low line is set during input
Manage super-pixelWith threshold value tone images T；It is output as setting background super-pixelFor each super-pixelIn riArea
T in the T of domainiRepresent white pixel number；If ti/area(ri)>ζ, then
8. distribute (four) based on the pixel tag described in claims 1, it is characterised in that by tone images and distributionIt is compared, generation pixel tag L image；Make FiTo indicate stochastic variable, if i-th in color-phase diagram picture
Pixel belongs to object of interest, then is 0, is otherwise 0；According to p (Fi|hi)=φ, condition is hiStochastic variable FiIt is Bernoulli Jacob
Formula is distributed, wherein parameterThe label of image L ith pixel is by 255 × p (Fi|hi) provide.
9. it is emerging including feeling based on the establishment mask (five) described in claims 1, it is characterised in that create mask image M
The region of interesting object and background object, them labeled as white, other all zone markers are black；Initially,In institute
There are pixel and pixel i to mark in M to be, wherein p (hi)>pm；The connection component of threshold value less than M is removed from M；So
Afterwards, continuous dilation procedure is carried out to M, until only one of which connects component；The quantity of required dilation procedure is expressed as d；Profile,
Convex hull and convex defect are calculated from the component of connection；Fill more than threshold value (∈d) depth convex defect it is (i.e. convex
The distance between solstics and convex hull in shape defect) and all in-profiles；Finally, d erosion of mask is returned to
Size originally.
10. based on the mask described in claims 9, it is characterised in that mask is applied into pixel tag image L, if M
Ith pixel is 0, then the pixel is outside area-of-interest, and the respective pixel in label L；In this application, mark value
For 0 pixel, meanwhile, the pixel in region interested is marked as 255 × p (Fi|hi)。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710216762.9A CN106952278A (en) | 2017-04-05 | 2017-04-05 | A kind of automatic division method in dynamic outdoor environment based on super-pixel |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710216762.9A CN106952278A (en) | 2017-04-05 | 2017-04-05 | A kind of automatic division method in dynamic outdoor environment based on super-pixel |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106952278A true CN106952278A (en) | 2017-07-14 |
Family
ID=59475221
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710216762.9A Withdrawn CN106952278A (en) | 2017-04-05 | 2017-04-05 | A kind of automatic division method in dynamic outdoor environment based on super-pixel |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106952278A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108520524A (en) * | 2018-04-10 | 2018-09-11 | 深圳市唯特视科技有限公司 | A kind of Image Iterative over-segmentation method based on edge cluster |
| CN108961295A (en) * | 2018-07-27 | 2018-12-07 | 重庆师范大学 | Purple soil image segmentation extracting method based on normal distribution H threshold value |
| CN116912701A (en) * | 2023-09-14 | 2023-10-20 | 潍坊现代农业山东省实验室 | Weed identification method and device and weed spraying method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140267583A1 (en) * | 2013-03-13 | 2014-09-18 | Futurewei Technologies, Inc. | Augmented Video Calls on Mobile Devices |
| CN104915950A (en) * | 2015-04-29 | 2015-09-16 | 华南理工大学 | Region growth ultrasound image automation segmentation method based on energy constraint |
| CN106530319A (en) * | 2016-09-24 | 2017-03-22 | 上海大学 | Video object cooperative segmentation method based on track directed graph |
-
2017
- 2017-04-05 CN CN201710216762.9A patent/CN106952278A/en not_active Withdrawn
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140267583A1 (en) * | 2013-03-13 | 2014-09-18 | Futurewei Technologies, Inc. | Augmented Video Calls on Mobile Devices |
| CN104915950A (en) * | 2015-04-29 | 2015-09-16 | 华南理工大学 | Region growth ultrasound image automation segmentation method based on energy constraint |
| CN106530319A (en) * | 2016-09-24 | 2017-03-22 | 上海大学 | Video object cooperative segmentation method based on track directed graph |
Non-Patent Citations (1)
| Title |
|---|
| AMY TABB ET.AL: "AUTOMATIC SEGMENTATION IN DYNAMIC OUTDOOR ENVIRONMENTS", 《ARXIV:1702.07611V1 [CS.CV]》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108520524A (en) * | 2018-04-10 | 2018-09-11 | 深圳市唯特视科技有限公司 | A kind of Image Iterative over-segmentation method based on edge cluster |
| CN108961295A (en) * | 2018-07-27 | 2018-12-07 | 重庆师范大学 | Purple soil image segmentation extracting method based on normal distribution H threshold value |
| CN116912701A (en) * | 2023-09-14 | 2023-10-20 | 潍坊现代农业山东省实验室 | Weed identification method and device and weed spraying method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR102134405B1 (en) | System and Method for Improving Low Light Level Image Using Generative Adversarial Network | |
| CN106203430B (en) | A kind of conspicuousness object detecting method based on foreground focused degree and background priori | |
| CN103942751B (en) | A kind of video key frame extracting method | |
| CN108830870A (en) | Satellite image high-precision field boundary extracting method based on Multi-scale model study | |
| CN109584251A (en) | A kind of tongue body image partition method based on single goal region segmentation | |
| CN110084782B (en) | Full-reference image quality evaluation method based on image significance detection | |
| CN108388905B (en) | A kind of Illuminant estimation method based on convolutional neural networks and neighbourhood context | |
| CN103914699A (en) | Automatic lip gloss image enhancement method based on color space | |
| CN103971126A (en) | Method and device for identifying traffic signs | |
| US10945657B2 (en) | Automated surface area assessment for dermatologic lesions | |
| CN101479762A (en) | A method, a system and a computer program for determining a threshold in an image comprising image values | |
| CN103617596A (en) | Image color style transformation method based on flow pattern transition | |
| CN106952278A (en) | A kind of automatic division method in dynamic outdoor environment based on super-pixel | |
| CN109766828A (en) | A kind of vehicle target dividing method, device and communication equipment | |
| CN110276764A (en) | K-Means underwater picture background segment innovatory algorithm based on the estimation of K value | |
| CN103218833A (en) | Edge-reinforced color space maximally stable extremal region detection method | |
| CN109447111A (en) | A kind of remote sensing supervised classification method based on subclass training sample | |
| CN103500451B (en) | A kind of independent floating ice extracting method for satellite data | |
| CN105225238A (en) | A kind of gray space division methods of the Image semantic classification based on mean filter | |
| Tyagi et al. | Performance comparison and analysis of medical image segmentation techniques | |
| CN106846343A (en) | A kind of pathological image feature extracting method based on cluster super-pixel segmentation | |
| CN103946868A (en) | Processing method and system for medical images | |
| CN104866853A (en) | Method for extracting behavior characteristics of multiple athletes in football match video | |
| Delibasis et al. | A novel robust approach for handling illumination changes in video segmentation | |
| CN108898601A (en) | Femoral head image segmentation device and dividing method based on random forest |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20170714 |
|
| WW01 | Invention patent application withdrawn after publication |