CN106651805A - Image watermark removing method based on machine learning - Google Patents
Image watermark removing method based on machine learning Download PDFInfo
- Publication number
- CN106651805A CN106651805A CN201611230137.1A CN201611230137A CN106651805A CN 106651805 A CN106651805 A CN 106651805A CN 201611230137 A CN201611230137 A CN 201611230137A CN 106651805 A CN106651805 A CN 106651805A
- Authority
- CN
- China
- Prior art keywords
- watermark
- image
- machine learning
- value
- transparency
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000010801 machine learning Methods 0.000 title claims abstract description 20
- 238000005516 engineering process Methods 0.000 claims description 10
- 230000006870 function Effects 0.000 claims description 8
- 238000010276 construction Methods 0.000 claims description 2
- 230000009466 transformation Effects 0.000 claims description 2
- 230000001131 transforming effect Effects 0.000 claims description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 2
- 230000000694 effects Effects 0.000 abstract description 9
- 230000001788 irregular Effects 0.000 abstract description 3
- 230000008569 process Effects 0.000 description 5
- 230000008859 change Effects 0.000 description 4
- 230000006835 compression Effects 0.000 description 3
- 238000007906 compression Methods 0.000 description 3
- 230000008439 repair process Effects 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 238000012952 Resampling Methods 0.000 description 1
- 241000679046 Teleas Species 0.000 description 1
- 230000001154 acute effect Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000007667 floating Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G06T5/77—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
Abstract
The invention discloses an image watermark removing method based on machine learning. According to the image watermark removing method, a way of the machine learning is adopted, and relatively accurate W and P are calculated and are obtained via an optimized algorithm, so that universal watermark removing parameters can be obtained by employing a few of a restored original image. The algorithm of the invention can adapt to a gray level and a color watermark as well as the watermark with an irregular pattern and a non-uniform transparency; and meanwhile, the method can automatically remove the watermark in batches and does not need too many interventions and operations. The method is simple and easy to operate, the cost is low, and the use effect is good.
Description
Technical field
The present invention relates to field of computer technology, especially a kind of image watermark minimizing technology based on machine learning.
Background technology
When copyright protection (or copyright notice) is carried out to image/video, the method for having a class common is to imaging importing
Upper visible watermark mark, and sometimes the user of image needs to remove watermark, using complete original image, it is necessary to carry out
Watermark is removed.
A kind of method is to use the method for image editing software such as PhotoShop to remove using artificial, and removal effect is relied on
The hight coordinate of human eye, human brain and drawing instrument coordinates, and can not batch and efficient process great amount of images.
Another method is, using the technology of image interpolation (inpaint), to carry out auto-mending to image, it is only necessary to thing
First mark watermark region, you can carry out batch and automatic watermark is removed.Comparative maturity is that Telea exists in inpaint methods
The An Image Inpainting Technique Based on the Fast Marching Method for delivering for 2004,
Abbreviation FMM methods.The method can utilize the color gradient information of image around region to be repaired, in addition directional weighting, distance
Weights, estimate point to be repaired, in the color of image change relatively steady and thinner situation of destroyed region lines, repair
Multiple effect is ideal.But the method does not utilize brightness and the structural information of the original image included in translucent watermark, to big
Estimate (reconstructed value) deviation in block region and acute variation region is larger.
The present invention exactly solves the method that visual watermark is removed using machine learning techniques.
The content of the invention
The purpose of the present invention is：There is provided a kind of image watermark minimizing technology based on machine learning, it can be using few
The artwork that amount is repaired, by machine learning method, obtains general watermark and removes parameter, and algorithm can adapt to gray scale and color water
Print, and the watermark that pattern is irregular, transparency is uneven, while automatically manually need not can too much intervene and operate
Batch de removes watermark, to overcome prior art not enough.
What the present invention was realized in：Based on the image watermark minimizing technology of machine learning, construction 100-1000 can be with
Then the estimation image of the watermarking images containing error and original image is solved as training set using the method for machine learning
Transforming function transformation function between watermarking images and original image respective pixel, inverts to this function and obtains watermark removal function, so as to containing
Watermark in the non-sample image of same watermark gives and removes.
Specifically include following steps；
1) learning of structure sample：From the 100-1000 image containing same watermark, 5-50 image is chosen；Selected
The watermark of image is the watermark that the mode of unbiased esti-mator can be utilized to be estimated, estimate is obtained after estimationSelected figure
As selecting readily estimated accurate sample, identical image as far as possible, if being unfamiliar with the people of image procossing, they can compare to do
It is difficult, it is possible to use training sample get just less, for the technical staff for being familiar with image procossing, this easily compares
Good to meet, training sample is taken more.
2) by machine learning, watermark value W and transparency P are obtained：According to watermark mixed model Y=W (1-P)+XP, B is made
=W (1-P), you can be transformed to Y=B+XP, directly solves the conclusion of slope and intercept using linear equation least square method, obtains
To P and B, and then obtain W；
3) watermark is removed：After obtaining the value of W and P, according to formula (1) reconstruction image X1：
X1=[Y-W × (1-P)]/P (1)
In formula, X represents original image, X1Reconstruction image is represented, Y represents the image containing watermark, and W represents watermark value, and P is represented
Lightness.
According to the principle of watermark mixing, original image X is superimposed after translucent image W, obtains the figure containing watermark
As Y, its conversion process can be expressed as Y=g (f (X, W)), and wherein f is watermark superposition conversion, and g is that resampling and image are damaged
Image change caused by compression.
For high-quality compression of images, we can assume that image change is negligible caused by g, then Y=f
(X,W)。
In order to process the truth that coloured image and watermark transparency can change in diverse location, we set difference
The coordinate of position is (x, y), and the numbering of RGB color image difference passage is c, then any one on non-watermarked coloured image X
Point any one Color Channel pixel value be X (x, y, c), watermarked W correspondence positions be W (x, y, c), mixed figure
As the correspondence position of Y is Y (x, y, c).
Common watermark see-through model is：Y (x, y, c)=W (x, y, c) × [1-P (x, y, c)]+X (x, y, c) × P (x,
y,c)。
Note：In order to describe simplicity, table below is up to omission (x, y, c).
According to above-mentioned principle, we will remove watermark, that is, estimate the process of X according to Y.
And X is obtained, W and P must be just obtained, but in most of application, we are not aware that in advance W and P,
It is exactly only to contain the image of watermark, but does not know that watermark value is how many, and does not know watermark is in what kind of ratio mixing
Go.
So, in order to obtain W and P, the present invention is processed using the method for machine learning.
Compared with prior art, the present invention calculates acquisition phase using the method for machine learning by the algorithm for optimizing
To accurate W and P such that it is able to using a small amount of artwork repaired, obtain general watermark and remove parameter, the algorithm energy of the present invention
Gray scale and color watermark, and the watermark that pattern is irregular, transparency is uneven are enough adapted to, while very important person can be automatically not required to
Work is intervened too much and the batch de of operation removes watermark.The present invention is simple, and with low cost, using effect is good.
Description of the drawings
Accompanying drawing 1 is without repairing very good study use-case in embodiments of the invention；
Accompanying drawing 2 is image of two width containing watermark in embodiments of the invention；
Accompanying drawing 3 is that embodiments of the invention image removed the effect of watermark using artificial 10 minutes editor's images；
Accompanying drawing 4 is the effect that embodiments of the invention remove watermark；
Accompanying drawing 5 is the watermark figure that embodiments of the invention learn.
Specific embodiment
Below in conjunction with the accompanying drawings the present invention is described in further detail, but not as any limitation of the invention
Embodiments of the invention：Based on the image watermark minimizing technology of machine learning, comprise the steps；
1) learning of structure sample：The readily estimated sample of a number of, background is collected as study material, that is, is taken
To individual Y；Depending on quantity visible image watermark estimation difficulty, if being difficult to estimate that the learning sample of reduction effect is more, need
Just than larger, our empirical value schemes this scope to quantity at 20-100；If the sample taken manually estimates that background is all missed
Difference is little, then the study of 5-10 sample graph can also make algorithm precise operation；
2) with FMM methods in inpaint algorithms, or other interpolated value methods of estimation (as Fig. 1 it is right), or even manual repair
Also allow (such as Fig. 3), estimate the situation of the no-watermark of artwork, so obtain the estimate of n X
3) the triple channel matrix W and P of watermark are set up, matrix size can be estimated, actual watermark size can be exceeded, nothing but
Minimal effect learns and removes watermark speed, and this step is not crucial：
4) using n to X and Y, substituting into machine learning method carries out parameter Estimation, for modal linear hybrid watermark mould
Type, is learnt using least square method, it is possible to use the formula of least square is obtained：
Note：Here with the replacement of above-mentioned B=(1-P) W, the expression of W is directly given, it is also the present invention's that this step is replaced
One of skill.
5th, artwork is recovered using X1=[Y-W × (1-P)]/P, reaches the purpose for removing watermark.
Wherein：What time have needs explanation：
1st, the statement of patent of the present invention carries out machine learning with the watermark of linear hybrid and least square method, but to non-thread
Property watermark mixing and other learning methods it is still effective.
2nd, estimate that accuracy requirements of the figure X in structure and texture be not high, but the hypothesis of zero-mean is important, that is,
Mean flow rate is consistent with artwork, and otherwise removing watermark figure can partially bright dark or colour cast partially.
3rd, Fig. 4 is examined, have part noise to repair even not as artificial reparation is schemed, its reason is that image is damaged
The caused picture breakdown of compression, not in the modeling of this retrieving algorithm and limit of consideration, we can not manually go to have repaired conduct
The result demonstration of this algorithm.Typically can be eliminated with the image filtering of a little yardstick, this elimination also has image side
The risk of edge destruction.
Left figure and right figure in Fig. 1 is respectively aqueous impression and simple and easy method removes the learning object that obtains after watermark, right
Y in arthmetic statement andCan seeOriginal background is estimated to be inaccurate, but it is this it is inaccurate can be with feeding device
Device learning algorithm is eliminated, and is the core skill of patent of the present invention.
Accompanying drawing 2 is that background behind their watermarks has certain complexity from the purpose of this two width figure, more difficult than learning sample
To estimate original graph.
Can see that, because process time is limited, many details of artificial treatment are difficult to the loyal fact, such as electric wire from accompanying drawing 3
It is erased with floating thing above, the position of shot-light and size are also incorrect.
And Fig. 4 is the effect that watermark is removed using the solution of the present invention, the X1 of correspondence formula is taken less than 1ms.
Fig. 5 is the watermark figure that the present invention learns, and is divided into color and transparency two parts, the W and P of correspondence formula.In Fig. 5
Left and right figure may each be coloured image, so that the watermarking images color and transparency that describe tri- Color Channels of R, G, B differ
The situation of cause.It is darker to represent opaquer for correspondence transparency image P, it is brighter to represent more transparent.So the watermark figure of example
Although black is very deep, not as white portion covers serious to artwork.
The above, is only the preferred embodiments of the present invention, and any pro forma restriction is not done to the present invention, any
Without departing from technical solution of the present invention content, above example is made any simple modification according to the technical spirit of the present invention,
Equivalent variations and modification, still fall within the range of technical solution of the present invention.
Claims (2)
1. a kind of image watermark minimizing technology based on machine learning, it is characterised in that：Construction 100-1000 can contain wrong
Then the estimation image of poor watermarking images and original image solves watermark figure as training set using the method for machine learning
Transforming function transformation function between picture and original image respective pixel, inverts to this function and obtains watermark removal function, so as to containing identical water
Watermark in the non-sample image of print gives and removes.
2. the image watermark minimizing technology based on machine learning according to claim 1, it is characterised in that specifically include as
Lower step；
1) learning of structure sample：From the 100-1000 image containing same watermark, 5-50 image is chosen；Selected image
Watermark be the watermark that be estimated of mode that can utilize unbiased esti-mator, estimate is obtained after estimation
2) by machine learning, watermark value W and transparency P are obtained：According to watermark mixed model Y=W (1-P)+XP, B=W is made
(1-P), you can be transformed to Y=B+XP, directly P is obtained using linear equation least square method solution slope and the conclusion of intercept
And B, and then obtain W；
3) watermark is removed：After obtaining the value of W and P, according to formula (1) reconstruction image X1：
X1=[Y-W × (1-P)]/P (1)
In formula, X represents original image, X1Reconstruction image is represented, Y represents the image containing watermark, and W represents watermark value, and P represents transparency.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611230137.1A CN106651805B (en) | 2016-12-28 | 2016-12-28 | Image watermark removing method based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611230137.1A CN106651805B (en) | 2016-12-28 | 2016-12-28 | Image watermark removing method based on machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106651805A true CN106651805A (en) | 2017-05-10 |
| CN106651805B CN106651805B (en) | 2020-08-25 |
Family
ID=58832899
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611230137.1A Active CN106651805B (en) | 2016-12-28 | 2016-12-28 | Image watermark removing method based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106651805B (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108805789A (en) * | 2018-05-29 | 2018-11-13 | 厦门市美亚柏科信息股份有限公司 | A kind of method, apparatus, equipment and readable medium removing watermark based on confrontation neural network |
| CN110084735A (en) * | 2019-04-26 | 2019-08-02 | 新华三云计算技术有限公司 | Watermark adding method, analytic method, device, electronic equipment and storage medium |
| CN110619597A (en) * | 2018-11-06 | 2019-12-27 | 北京时光荏苒科技有限公司 | Semitransparent watermark removing method and device, electronic equipment and storage medium |
| CN111192190A (en) * | 2019-12-31 | 2020-05-22 | 北京金山云网络技术有限公司 | Method and device for eliminating image watermark and electronic equipment |
| CN111798359A (en) * | 2020-05-19 | 2020-10-20 | 佛山市南海区广工大数控装备协同创新研究院 | Deep learning-based image watermark removing method |
| CN111932431A (en) * | 2020-07-07 | 2020-11-13 | 华中科技大学 | Visible watermark removing method based on watermark decomposition model and electronic equipment |
| CN112528245A (en) * | 2019-11-14 | 2021-03-19 | 百度（美国）有限责任公司 | Method for processing data by a data processing accelerator and data processing accelerator |
| CN113095988A (en) * | 2021-03-29 | 2021-07-09 | 贵州大学 | Dispersion tensor image robust zero watermarking method based on ORC sampling and QGPCE conversion |
| US11409843B2 (en) | 2019-10-10 | 2022-08-09 | Nxp B.V. | Method for protecting a software program from copying |
| US11409845B2 (en) | 2019-01-17 | 2022-08-09 | Nxp B.V. | Method for determining if a machine learning model has been copied |
| US11500970B2 (en) | 2019-08-02 | 2022-11-15 | Nxp B.V. | Machine learning model and method for determining if the machine learning model has been copied |
| US11586989B2 (en) | 2019-07-15 | 2023-02-21 | Nxp B.V. | Method for detecting if a machine learning model has been copied using intermediate outputs of the machine learning model |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2370437A (en) * | 2000-12-20 | 2002-06-26 | Motorola Inc | Method for watermarking a digital image |
| CN1945623A (en) * | 2006-10-26 | 2007-04-11 | 丹东东方测控技术有限公司 | Method for removing visual watermark of block local encrypted digital image |
| CN101142808A (en) * | 2005-01-04 | 2008-03-12 | 史密斯微软件公司 | Method and apparatus for creating a digital picture with a removable watermark |
| CN101661608A (en) * | 2009-08-31 | 2010-03-03 | 重庆大学 | Diginal image blind watermark adding, extracting and removing method suitable for internet |
| CN101889433A (en) * | 2007-12-13 | 2010-11-17 | 汤姆森许可贸易公司 | Method and apparatus for inserting a removable visible watermark in an image and method and apparatus for removing such watermarks |
| CN102930499A (en) * | 2012-11-16 | 2013-02-13 | 崔得龙 | Method for adding and removing digital watermark to and from digital image signal |
| US20150128290A1 (en) * | 2011-11-11 | 2015-05-07 | Optimark, Llc | Digital communications |
| CN105427233A (en) * | 2015-12-29 | 2016-03-23 | 小米科技有限责任公司 | Method and device for removing watermark |
| CN105898322A (en) * | 2015-07-24 | 2016-08-24 | 乐视云计算有限公司 | Video watermark removing method and device |
-
2016
- 2016-12-28 CN CN201611230137.1A patent/CN106651805B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2370437A (en) * | 2000-12-20 | 2002-06-26 | Motorola Inc | Method for watermarking a digital image |
| CN101142808A (en) * | 2005-01-04 | 2008-03-12 | 史密斯微软件公司 | Method and apparatus for creating a digital picture with a removable watermark |
| CN1945623A (en) * | 2006-10-26 | 2007-04-11 | 丹东东方测控技术有限公司 | Method for removing visual watermark of block local encrypted digital image |
| CN101889433A (en) * | 2007-12-13 | 2010-11-17 | 汤姆森许可贸易公司 | Method and apparatus for inserting a removable visible watermark in an image and method and apparatus for removing such watermarks |
| CN101661608A (en) * | 2009-08-31 | 2010-03-03 | 重庆大学 | Diginal image blind watermark adding, extracting and removing method suitable for internet |
| US20150128290A1 (en) * | 2011-11-11 | 2015-05-07 | Optimark, Llc | Digital communications |
| CN102930499A (en) * | 2012-11-16 | 2013-02-13 | 崔得龙 | Method for adding and removing digital watermark to and from digital image signal |
| CN105898322A (en) * | 2015-07-24 | 2016-08-24 | 乐视云计算有限公司 | Video watermark removing method and device |
| CN105427233A (en) * | 2015-12-29 | 2016-03-23 | 小米科技有限责任公司 | Method and device for removing watermark |
Non-Patent Citations (2)
| Title |
|---|
| ANDREAS WESTFELD 等: "A Regression-Based Restoration Technique for Automated Watermark Removal", 《MM&SEC"08 PROCEEDINGS OF THE 10TH ACM WORKSHOP ON MULTIMEDIA AND SECURITY》 * |
| 叶春: "图像数字水印及修复算法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108805789B (en) * | 2018-05-29 | 2022-06-03 | 厦门市美亚柏科信息股份有限公司 | Method, device and equipment for removing watermark based on antagonistic neural network and readable medium |
| CN108805789A (en) * | 2018-05-29 | 2018-11-13 | 厦门市美亚柏科信息股份有限公司 | A kind of method, apparatus, equipment and readable medium removing watermark based on confrontation neural network |
| CN110619597A (en) * | 2018-11-06 | 2019-12-27 | 北京时光荏苒科技有限公司 | Semitransparent watermark removing method and device, electronic equipment and storage medium |
| US11409845B2 (en) | 2019-01-17 | 2022-08-09 | Nxp B.V. | Method for determining if a machine learning model has been copied |
| CN110084735A (en) * | 2019-04-26 | 2019-08-02 | 新华三云计算技术有限公司 | Watermark adding method, analytic method, device, electronic equipment and storage medium |
| US11586989B2 (en) | 2019-07-15 | 2023-02-21 | Nxp B.V. | Method for detecting if a machine learning model has been copied using intermediate outputs of the machine learning model |
| US11500970B2 (en) | 2019-08-02 | 2022-11-15 | Nxp B.V. | Machine learning model and method for determining if the machine learning model has been copied |
| US11409843B2 (en) | 2019-10-10 | 2022-08-09 | Nxp B.V. | Method for protecting a software program from copying |
| CN112528245A (en) * | 2019-11-14 | 2021-03-19 | 百度（美国）有限责任公司 | Method for processing data by a data processing accelerator and data processing accelerator |
| CN111192190A (en) * | 2019-12-31 | 2020-05-22 | 北京金山云网络技术有限公司 | Method and device for eliminating image watermark and electronic equipment |
| CN111192190B (en) * | 2019-12-31 | 2023-05-12 | 北京金山云网络技术有限公司 | Method and device for eliminating image watermark and electronic equipment |
| CN111798359A (en) * | 2020-05-19 | 2020-10-20 | 佛山市南海区广工大数控装备协同创新研究院 | Deep learning-based image watermark removing method |
| CN111932431A (en) * | 2020-07-07 | 2020-11-13 | 华中科技大学 | Visible watermark removing method based on watermark decomposition model and electronic equipment |
| CN113095988B (en) * | 2021-03-29 | 2022-02-01 | 贵州大学 | Dispersion tensor image robust zero watermarking method based on ORC sampling and QGPCE conversion |
| CN113095988A (en) * | 2021-03-29 | 2021-07-09 | 贵州大学 | Dispersion tensor image robust zero watermarking method based on ORC sampling and QGPCE conversion |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106651805B (en) | 2020-08-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106651805A (en) | Image watermark removing method based on machine learning | |
| CN103971126B (en) | A kind of traffic sign recognition method and device | |
| CN101587586B (en) | Device and method for processing images | |
| CN105791803B (en) | A kind of display methods and system that two dimensional image is converted into multi-view image | |
| CN103295199B (en) | Intelligent repair assistance system for cracks of ancient wall murals | |
| CN103914699A (en) | Automatic lip gloss image enhancement method based on color space | |
| CN104809698A (en) | Kinect depth image inpainting method based on improved trilateral filtering | |
| US10620417B2 (en) | Method for generating a reflection-reduced contrast image and corresponding device | |
| CN108550159A (en) | A kind of flue gas concentration identification method based on the segmentation of three color of image | |
| CN109345480B (en) | Face automatic acne removing method based on image restoration model | |
| CN104134198A (en) | Method for carrying out local processing on image | |
| CN107689050A (en) | A kind of depth image top sampling method based on Color Image Edge guiding | |
| CN104103043A (en) | Spatial gamut mapping algorithm | |
| CN105338214B (en) | Image processing method and image processing apparatus | |
| CN111489322A (en) | Method and device for adding sky filter to static picture | |
| CN108257187A (en) | A kind of camera-projecting apparatus system scaling method | |
| CN105760869B (en) | Examination paper marking method based on image gray level recognition and processing | |
| CN104504661A (en) | Method for processing image to be in black and white effect | |
| CN102567969A (en) | Color image edge detection method | |
| CN111383352B (en) | Automatic color filling and abstraction method for third-order magic cube | |
| TWI462027B (en) | Image processing device and image processing method thereof | |
| CN111161228A (en) | Button surface defect detection method based on transfer learning | |
| US10175468B2 (en) | Method for generating a contrast image of an object structure and apparatuses relating thereto | |
| CN109191483A (en) | A kind of quick watershed detection method of helicopter blade Circle in Digital Images shape mark | |
| CN113284035A (en) | System and method for generating dynamic picture with two-dimensional code watermark |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20231220 Address after: Room 206, 2nd Floor, Building 13, Sun Garden, Dazhongsi East Road, Haidian District, Beijing, 100000 Patentee after: BEIJING NJA INFORMATION TECHNOLOGY CO.,LTD. Address before: 550025 science and technology office, north campus, Guizhou University, Huaxi District, Guiyang, Guizhou Patentee before: Guizhou University |