CN107730568A - Color method and device based on weight study - Google Patents
Color method and device based on weight study Download PDFInfo
- Publication number
- CN107730568A CN107730568A CN201711048927.2A CN201711048927A CN107730568A CN 107730568 A CN107730568 A CN 107730568A CN 201711048927 A CN201711048927 A CN 201711048927A CN 107730568 A CN107730568 A CN 107730568A
- Authority
- CN
- China
- Prior art keywords
- mrow
- color
- pixel
- weight
- image
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T11/00—2D [Two Dimensional] image generation
- G06T11/001—Texturing; Colouring; Generation of texture or colour
Abstract
The invention discloses a kind of color method based on weight study, comprise the following steps:Several gray level images and corresponding coloured image are chosen, the feature difference between gray image and corresponding coloured image adjacent pixel are calculated respectively, as training dataset;Based on the training dataset, weight learning model is trained using random forests algorithm;The marker color on target gray image to be colored;The feature difference target gray image zooming-out adjacent pixel, as the input of weight learning model, optimal weights are obtained;Color transmission is carried out according to color mark and the optimal weights, obtains coloured image corresponding to target gray image.The color method of the present invention, obtains weight using the mode of study, can obtain the correlation between more excellent pixel, and obtain preferable coloring effect.
Description
Technical field
The present invention relates to a kind of computer assisted image rendering methods, more particularly to a kind of figure based on weight study
As color method.
Background technology
Image is as the true reflection that the carrier of information is to human visual perception, and color is it is appreciated that image is very heavy
The information wanted, it is one of most important attribute of image.People experienced the transition from black white image to coloured image, but
In early days, camera work at that time is limited, and can be only generated the photo and image of black and white, therefore add for these older pictures and image
Appropriate color, it is allowed to, with more sight, turn into a very important task.
Colour a word most to be proposed by Wilson Markle early in 1970, and be defined as one kind by computer to black and white
Or monochrome image and the process of video dyeing[1].The appearance of dye technology, the color of image can be reduced, strengthens or change,
Improve the visual effect of people, enable people to extract more accurate information from image, be advantageous to people's profound understanding figure
As content, so as to improve the use value of image.
In recent years, dye technology is quickly grown.And in early stage, people are caught full by way of painting by hand to image
The color of meaning, this task generally require professional to complete, and process is very time-consuming.With Digital Image Processing skill
The continuous development of art, it is desirable to handle the demand of this respect by computer help, the problem of digital picture coloring also meets the tendency of
And give birth to.
Currently used color method is broadly divided into two classes, color method based on color mark and based on reference picture picture
Color method.Color method based on reference picture picture does not need the interaction of user, but realizes color by means of reference picture picture
Migration.Method based on color mark needs user to draw colour-coded on gray level image, is then realized and marked using algorithm
Remember transmission of the color position to unknown color region.The advantages of this method is that people can be according to demand to the difference of image
Part is marked, so that the image dyed more meets the needs of to color.Wherein, Levin et al. proposes a kind of profit
With the optimized algorithm of the similarity relation (weight) of pixel between neighborhood.Weight, the phase of two neighboring pixel is represented in such method
Like relation, the color for being used to refer to be shown with how much amount during color transfer is transmitted to periphery.Weighted value is bigger, represents both
Between similitude it is bigger.Many methods are improved based on this, different weighting function defined in some methods, but these
The weighing computation method used in method is all pre-defined, and not clearly stating any weight can obtain preferably
As a result.Therefore, how weight is improved, is that those skilled in the art need urgent solve at present to improve coloring effect
Technical problem.
The content of the invention
To overcome above-mentioned the deficiencies in the prior art, the invention provides a kind of color method based on weight study, the party
Method mainly includes following two key components:Weight learns stage and tinting stage, and we are established from gray level image to colour
The weight learning model of image, and by the model learning to weight target image is coloured.The coloring side of the present invention
Method, weight is obtained using the mode of study, the correlation between more excellent pixel can be obtained, and obtain preferably coloring effect
Fruit.
To achieve the above object, the present invention adopts the following technical scheme that:
A kind of color method based on weight study, comprises the following steps:
Step 1:Several gray level images and corresponding coloured image are chosen, calculate gray image and corresponding coloured image phase respectively
Feature difference between adjacent pixel, as training dataset;
Step 2:Based on the training dataset, weight learning model is trained using random forests algorithm;
Step 3:The marker color on target gray image to be colored;
Step 4:The feature difference target gray image zooming-out adjacent pixel, as the input of weight learning model,
Obtain optimal weights;
Step 5:Color transmission is carried out according to color mark and the optimal weights, obtained color corresponding to target gray image
Color image.
Further, the feature difference between the gray image adjacent pixel is:
Frs=| | Fs-Fr||
Wherein, r is a certain pixel in gray image, and s is r neighborhood territory pixel, and F={ f1, f2 } then represents brightness and ladder
The characteristic vector of feature composition is spent, f1, f2 represent brightness and Gradient Features respectively.
Further, the feature difference between the coloured image adjacent pixel is color distortion:
Wherein, drs=| | Ls-Lr||2+||as-ar||2+||bs-br||2
R is a certain pixel in gray image, and s is r neighborhood territory pixel, and L represents monochrome information, and a, b represent two respectively
Color component, drsRepresent pixel s and r distance, var is threshold value, max (drs) represent to ask for d maximum in a neighborhoodrs。
Further, the weight learning model uses (Frs, Drs) it is used as training set.
Further, if the weight learning model learns to obtain Drs, pixel r and s optimal weights are:
Further, to pixel r labeled in image, pixel r and each pixel s in neighborhood weight are entered
Row normalized.
Further, during the color transfer, the majorized function of adjacent pixel color value is:
Wherein, C represents UV components, CrRepresent the color of center pixel, CsRepresent the color of field inner rim pixel.
According to the second object of the present invention, present invention also offers a kind of color applicator based on weight study, including deposit
Reservoir, processor and storage on a memory and the computer program that can run on a processor, described in the computing device
The above-mentioned color method based on weight study is realized during program.
According to the third object of the present invention, present invention also offers a kind of computer-readable recording medium, it is stored thereon with
Computer program, the program perform the above-mentioned color method based on weight study when being executed by processor.
Beneficial effects of the present invention
(1) present invention proposes a kind of mode of study and obtains weight, according to hypothesis:In coloured image, if pixel
Between color approach, then have very big similarity in corresponding gray level image, between pixel.I.e. in coloured image, adjacent picture
Correlation between plain color can more accurately express the information of weight.The color distance conduct of adjacent pixel in color space
The actual value of weight participates in training, establishes one from gray level image feature to the learning model of coloured image weight.It is given to appoint
Meaning gray scale target image, can be learnt to more excellent weight by the model, so as to obtain satisfied colouring results.
(2) present invention not only uses gray scale and represents to represent pixel, Gradient Features is also added into, the meter of relation pixel
Calculation adds more information, is then established with the mode of combinations of features and learns the association from gray level image to coloured image.
Experiment shows that brightness and the combination of gray feature can produce preferable colouring results.
(3) it is compared by the present invention with the result that Levin et al. methods proposed obtain, the knot that this method obtains
Fruit and original coloured image gap are smaller, and it has more preferable coloring effect.
Brief description of the drawings
The Figure of description for forming the part of the application is used for providing further understanding of the present application, and the application's shows
Meaning property embodiment and its illustrate be used for explain the application, do not form the improper restriction to the application.
Fig. 1 is the flow chart of color method of the present invention based on weight study;
Fig. 2 is present example design sketch, and 2 (a) and 2 (d) is tape label image;2 (b) and 2 (e) is what this method obtained
As a result;2 (c) and 2 (f) is original image.
Embodiment
It is noted that described further below is all exemplary, it is intended to provides further instruction to the application.It is unless another
Indicate, all technologies used herein and scientific terminology have to be led to the application person of an ordinary skill in the technical field
The identical meanings understood.
It should be noted that term used herein above is merely to describe embodiment, and be not intended to restricted root
According to the illustrative embodiments of the application.As used herein, unless the context clearly indicates otherwise, otherwise odd number shape
Formula is also intended to include plural form, additionally, it should be understood that, when in this manual use term "comprising" and/or
During " comprising ", it indicates existing characteristics, step, operation, device, component and/or combinations thereof.
In the case where not conflicting, the feature in embodiment and embodiment in the application can be mutually combined.
General thought proposed by the present invention:In order to obtain the correlation between more accurate pixel, obtain preferably
Color effect, the present invention propose a kind of mode of weight study and are used to colour.This method take into account image when obtaining weight
Colour information, the actual value of weight is calculated in original color image, office in gray level image is established by way of precondition
Relational model of portion's neighborhood into coloured image between optimal weights, corresponding to one target gray image of the model prediction
Accurate weight information, and using the weight by the color of mark to zone of ignorance transmission.
Embodiment one
Present embodiment discloses a kind of color method based on weight study, comprise the following steps:
Step 1:Several gray level images and corresponding coloured image are chosen, calculate gray image and corresponding coloured image phase respectively
Feature difference between adjacent pixel, as training dataset;
Training set includes two parts, first, the feature in gray level image between pixel is poor, second, the colour-difference in color space.
In ready gray level image and coloured image to upper, feature difference F is calculated respectivelyrsWith colour-difference Drs, FrsAnd DrsDo respectively to
The processing of quantization, FrsIt is the vector of bidimensional, represents brightness and the gradient two-dimensional vector of adjacent pixel, D respectivelyrsBe it is one-dimensional to
Amount, represent colour-difference.L represents the population size of relation pair between adjacent pixel.By (Frs, Drs) participate in training as training set.
For the feature difference between gray image adjacent pixel:
The combinations of features of gray scale and gradient that the present embodiment employs image represents each pixel in image, f1, f2
Brightness and Gradient Features are represented respectively, and F={ f1, f2 } then represents the characteristic vector of both compositions.Firstly, for gray level image
In each pixel, we extract f1, f2 features, form two characteristic patterns.For each pixel s, one is taken first
3*3 neighborhood, then calculate r and each adjacent pixel s in neighborhood feature difference Frs。
Frs=| | Fs-Fr||
For the feature difference between coloured image adjacent pixel:
The present embodiment makees the feature difference of coloured image using color distortion.Lab color spaces are used in calculating process, its
In, L represents monochrome information, and a, b represent two color components respectively.In three-dimensional color space, each color as one
Point, the similarity relation between two colors are represented by its distance.To each pixel s in coloured image, first using s as
Center takes 3*3 neighborhood, then calculates r and each adjacent pixel of neighborhood s distance.R and s distance is labeled as Drs,
Weight between r and s is designated as Wrs。
drs=| | Ls-Lr||2+||as-ar||2+||bs-br||2
Wherein, var is threshold value, and rule of thumb formula can be adjusted.max(drs) represent to ask in a neighborhood most
Big drs。
As can be seen from the above equation, DrsWhat is represented is distance between the two, and what weight represented is both similarity relations,
Both distances are bigger, and similarity is smaller, so the relation that distance and similarity are integrally inversely proportional.
Step 2:Based on the training dataset, weight learning model is trained using random forests algorithm;
Weight Learning Scheme proposed by the present invention is mainly to utilize the characteristic relation study in gray level image to arrive color space
In color between relation.Random forest builds a forest with random manner, has many independent decision trees in forest,
When there is new sample input, then decision-making is carried out with the tree built.For random forest grader, give training data X [X1,
X2 ..., XL] and mark Y (Y1, Y2 ..., YL) quantity of training data (wherein L be) corresponding to it, it will with training set (X,
Y learning model M) is built.The F wherein extracted in gray level imagersAs training data X, the D that is calculated in coloured imagers
As mark Y.By (Frs, Drs) as training set input random forest program, export a weight learning model.
Step 3:The marker color on target gray image to be colored;
For the ease of observing the coloring effect of this method, original color image is converted into gray level image by the present embodiment, is made
For target gray image to be colored, then according to the distribution situation of color in original image, drawn with paintbrush on gray level image
The line markings enameled, two images (being child and field respectively from top to bottom), mark image such as Fig. 2 are used in this example
Shown, the result (Fig. 2 (b)) finally obtained compares with original image (Fig. 2 (c)), the validity of verification method.
Step 4:The feature difference target gray image zooming-out adjacent pixel, as the input of weight learning model,
Obtain optimal weights;
Learn weight for target image:Weight learning model establishes the mapping between gray level image and coloured image, defeated
Enter the unknown image of any one color, then can learn to optimal weights.
In this example, as shown in Fig. 2 being processed for child and field two images.Respectively in child and the width mark of field two
Remember and gray scale and Gradient Features are extracted on image, the feature then calculated between pixel is poor, all saves as two-dimensional vector.By the vector
It is input to Random Forest model, output one-dimensional vector Drs.A kind of distance relation that the vector actually learns, Ran Houtong
Cross below equation and optimal weights are calculated
Step 5:Color transmission is carried out according to color mark and the optimal weights, obtained color corresponding to target gray image
Color image.
To pixel r labeled in image, 3*3 neighborhood is taken centered on r, each pixel s and center in neighborhood
The weight of pixel is known, it is necessary to make homogenization processing to these weights so that
Then the transmission for optimizing color method and realizing color is utilized.The process of color is transmitted using Levin et al. propositions
Optimize the method for coloring, communication process is carried out in YUV color spaces, and wherein Y is, it is known that recovering the UV of unknown color region
Value, is finally converted into rgb space by yuv space again.
The optimization color method is recovered not by reducing the gap of center pixel color value and neighborhood territory pixel color weight sum
Know field color.Its majorized function is:
This method uses YUV color spaces, and Y represents brightness, and UV represents two components respectively.Wherein UV points are represented with C
Amount, CrRepresent the color of center pixel, CsRepresent the color of field inner rim pixel.Obtained final result such as Fig. 2 (b) institutes
Show.
Except the method for above-mentioned Levin et al. optimization colorings proposed, other published face in the prior art can be also used
Color transmission method, such as the colour-difference of 4- neighborhood territory pixels by minimizing sub-pixel that Horiuchi is proposed carry out face
Color is propagated or using the color distortion minimum between neighborhood of pixels as optimization aim;Huang propose based on adaptive boundary
The color method of detection;Sapiro propose using brightness step come partial differential method for solving for being coloured etc..
Embodiment two
The purpose of the present embodiment is to provide a kind of computing device.
A kind of color applicator based on weight study, including memory, processor and storage on a memory and can located
The computer program that runs on reason device, following steps are realized during the computing device described program, including:
Step 1:Several gray level images and corresponding coloured image are chosen, calculate gray image and corresponding coloured image phase respectively
Feature difference between adjacent pixel, as training dataset;
Step 2:Based on the training dataset, weight learning model is trained using random forests algorithm;
Step 3:The marker color on target gray image to be colored;
Step 4:The feature difference target gray image zooming-out adjacent pixel, as the input of weight learning model,
Obtain optimal weights;
Step 5:Color transmission is carried out according to color mark and the optimal weights, obtained color corresponding to target gray image
Color image.
Embodiment three
The purpose of the present embodiment is to provide a kind of computer-readable recording medium.
A kind of computer-readable recording medium, is stored thereon with computer program, is coloured for gray level image, the program quilt
Following steps are performed during computing device:
Step 1:Several gray level images and corresponding coloured image are chosen, calculate gray image and corresponding coloured image phase respectively
Feature difference between adjacent pixel, as training dataset;
Step 2:Based on the training dataset, weight learning model is trained using random forests algorithm;
Step 3:The marker color on target gray image to be colored;
Step 4:The feature difference target gray image zooming-out adjacent pixel, as the input of weight learning model,
Obtain optimal weights;
Step 5:Color transmission is carried out according to color mark and the optimal weights, obtained color corresponding to target gray image
Color image.
Each step being related in the device of above example two and three is corresponding with embodiment of the method one, embodiment
Reference can be made to the related description part of embodiment one.Term " computer-readable recording medium " is construed as including one or more
The single medium or multiple media of individual instruction set;Any medium is should also be understood as including, any medium can be deposited
Store up, encode or carry for the instruction set by computing device and make the either method in the computing device present invention.
The present invention establishes the learning model from gray level image feature to coloured image weight.The power obtained based on the model
Color transfer is carried out again, and experiment proves there is preferable colouring results.
It will be understood by those skilled in the art that each module or each step of the invention described above can be filled with general computer
Put to realize, alternatively, they can be realized with the program code that computing device can perform, it is thus possible to which they are deposited
Storage performed in the storage device by computing device, either they are fabricated to respectively each integrated circuit modules or by it
In multiple modules or step be fabricated to single integrated circuit module to realize.The present invention is not restricted to any specific hard
The combination of part and software.
Although above-mentioned the embodiment of the present invention is described with reference to accompanying drawing, model not is protected to the present invention
The limitation enclosed, one of ordinary skill in the art should be understood that on the basis of technical scheme, those skilled in the art
Various modifications or deformation that creative work can make need not be paid still within protection scope of the present invention.
Claims (9)
1. a kind of color method based on weight study, it is characterised in that comprise the following steps:
Step 1:Several gray level images and corresponding coloured image are chosen, calculate gray image and the adjacent picture of corresponding coloured image respectively
Feature difference between element, as training dataset;
Step 2:Based on the training dataset, weight learning model is trained using random forests algorithm;
Step 3:The marker color on target gray image to be colored;
Step 4:The feature difference target gray image zooming-out adjacent pixel, as the input of weight learning model, obtain
Optimal weights;
Step 5:Color transmission is carried out according to color mark and the optimal weights, obtains cromogram corresponding to target gray image
Picture.
2. the color method as claimed in claim 1 based on weight study, it is characterised in that the gray image adjacent pixel
Between feature difference be:
Frs=| | Fs-Fr||
Wherein, r is a certain pixel in gray image, and s is r neighborhood territory pixel, and F={ f1, f2 } then represents that brightness and gradient are special
The characteristic vector of composition is levied, f1, f2 represent brightness and Gradient Features respectively.
3. the color method as claimed in claim 1 based on weight study, it is characterised in that the coloured image adjacent pixel
Between feature difference be color distortion:
<mrow>
<msub>
<mi>D</mi>
<mrow>
<mi>r</mi>
<mi>s</mi>
</mrow>
</msub>
<mo>=</mo>
<mfrac>
<msub>
<mi>d</mi>
<mrow>
<mi>r</mi>
<mi>s</mi>
</mrow>
</msub>
<mrow>
<mi>m</mi>
<mi>a</mi>
<mi>x</mi>
<mrow>
<mo>(</mo>
<mi>m</mi>
<mi>a</mi>
<mi>x</mi>
<mo>(</mo>
<msub>
<mi>d</mi>
<mrow>
<mi>r</mi>
<mi>s</mi>
</mrow>
</msub>
<mo>)</mo>
<mo>,</mo>
<mi>var</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
</mrow>
Wherein, drs=| | Ls-Lr||2+||as-ar||2+||bs-br||2
R is a certain pixel in gray image, and s is r neighborhood territory pixel, and L represents monochrome information, and a, b represent two colors respectively
Component, drsRepresent pixel s and r distance, var is threshold value, max (drs) represent to ask for d maximum in a neighborhoodrs。
4. the color method as claimed in claim 3 based on weight study, it is characterised in that the weight learning model uses
(Frs, Drs) it is used as training set.
5. the color method as claimed in claim 4 based on weight study, it is characterised in that set the weight learning model
Acquistion is to Drs, pixel r and s optimal weights are:
<mrow>
<msub>
<mi>W</mi>
<mrow>
<mi>r</mi>
<mi>s</mi>
<mo>,</mo>
<mi>s</mi>
<mo>&Element;</mo>
<mi>N</mi>
<mrow>
<mo>(</mo>
<mi>r</mi>
<mo>)</mo>
</mrow>
</mrow>
</msub>
<mo>=</mo>
<msup>
<mi>exp</mi>
<mrow>
<mo>-</mo>
<msub>
<mi>D</mi>
<mrow>
<mi>r</mi>
<mi>s</mi>
</mrow>
</msub>
</mrow>
</msup>
<mo>.</mo>
</mrow>
6. the color method as claimed in claim 5 based on weight study, it is characterised in that to picture labeled in image
Plain r, pixel r and each pixel s in neighborhood weight are normalized.
7. the color method as claimed in claim 6 based on weight study, it is characterised in that during the color transfer,
The majorized function of adjacent pixel color value is:
<mrow>
<mi>argmin</mi>
<munder>
<mo>&Sigma;</mo>
<mi>r</mi>
</munder>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>C</mi>
<mi>r</mi>
</msub>
<mo>-</mo>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mi>s</mi>
<mo>&Element;</mo>
<mi>N</mi>
<mrow>
<mo>(</mo>
<mi>r</mi>
<mo>)</mo>
</mrow>
</mrow>
</munder>
<msub>
<mi>W</mi>
<mrow>
<mi>r</mi>
<mi>s</mi>
</mrow>
</msub>
<msub>
<mi>C</mi>
<mi>s</mi>
</msub>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
Wherein, C represents UV components, CrRepresent the color of center pixel, CsRepresent the color of field inner rim pixel.
8. a kind of color applicator based on weight study, including memory, processor and storage on a memory and can handled
The computer program run on device, it is characterised in that realized during the computing device described program as claim 1-7 is any
The color method based on weight study described in.
9. a kind of computer-readable recording medium, is stored thereon with computer program, it is characterised in that the program is held by processor
The color method based on weight study as described in claim any one of 1-7 is performed during row.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711048927.2A CN107730568B (en) | 2017-10-31 | 2017-10-31 | Coloring method and device based on weight learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711048927.2A CN107730568B (en) | 2017-10-31 | 2017-10-31 | Coloring method and device based on weight learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107730568A true CN107730568A (en) | 2018-02-23 |
CN107730568B CN107730568B (en) | 2021-01-08 |
Family
ID=61202029
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711048927.2A Active CN107730568B (en) | 2017-10-31 | 2017-10-31 | Coloring method and device based on weight learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107730568B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108389168A (en) * | 2018-02-26 | 2018-08-10 | 上海工程技术大学 | A method of obtaining fixed area unmanned plane image |
CN108921932A (en) * | 2018-06-28 | 2018-11-30 | 福州大学 | A method of the black and white personage picture based on convolutional neural networks generates various reasonable coloring in real time |
CN112073596A (en) * | 2020-09-18 | 2020-12-11 | 青岛大学 | Simulated color processing method and system for specific black-and-white video signal |
CN112102438A (en) * | 2020-08-26 | 2020-12-18 | 东南大学 | Graph dyeing problem searching method based on elite solution driven multi-level tabu search |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103839079A (en) * | 2014-03-18 | 2014-06-04 | 浙江师范大学 | Similar image colorization algorithm based on classification learning |
CN104376529A (en) * | 2014-11-25 | 2015-02-25 | 深圳北航新兴产业技术研究院 | Gray level image colorization system and method based on GLCM |
CN104851074A (en) * | 2015-03-26 | 2015-08-19 | 温州大学 | Feature similarity-based non-local neighborhood gray level image colorization method |
-
2017
- 2017-10-31 CN CN201711048927.2A patent/CN107730568B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103839079A (en) * | 2014-03-18 | 2014-06-04 | 浙江师范大学 | Similar image colorization algorithm based on classification learning |
CN104376529A (en) * | 2014-11-25 | 2015-02-25 | 深圳北航新兴产业技术研究院 | Gray level image colorization system and method based on GLCM |
CN104851074A (en) * | 2015-03-26 | 2015-08-19 | 温州大学 | Feature similarity-based non-local neighborhood gray level image colorization method |
Non-Patent Citations (2)
Title |
---|
ANAT LEVIN,ET AL.: "Colorization using Optimization", 《ACM TRANSACTIONS ON GRAPHICS》 * |
朱黎博等: "基于色彩传递与扩展的图像着色算法", 《中国图象图形学报》 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108389168A (en) * | 2018-02-26 | 2018-08-10 | 上海工程技术大学 | A method of obtaining fixed area unmanned plane image |
CN108389168B (en) * | 2018-02-26 | 2022-01-28 | 上海工程技术大学 | Method for acquiring aerial image of unmanned aerial vehicle in fixed area |
CN108921932A (en) * | 2018-06-28 | 2018-11-30 | 福州大学 | A method of the black and white personage picture based on convolutional neural networks generates various reasonable coloring in real time |
CN108921932B (en) * | 2018-06-28 | 2022-09-23 | 福州大学 | Method for generating multiple reasonable colorings of black and white figure pictures based on convolutional neural network |
CN112102438A (en) * | 2020-08-26 | 2020-12-18 | 东南大学 | Graph dyeing problem searching method based on elite solution driven multi-level tabu search |
CN112102438B (en) * | 2020-08-26 | 2024-01-26 | 东南大学 | Graph dyeing problem searching method based on elite solution driven multi-level tabu search |
CN112073596A (en) * | 2020-09-18 | 2020-12-11 | 青岛大学 | Simulated color processing method and system for specific black-and-white video signal |
CN112073596B (en) * | 2020-09-18 | 2021-07-20 | 青岛大学 | Simulated color processing method and system for specific black-and-white video signal |
Also Published As
Publication number | Publication date |
---|---|
CN107730568B (en) | 2021-01-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107730568A (en) | Color method and device based on weight study | |
CN107833183B (en) | Method for simultaneously super-resolving and coloring satellite image based on multitask deep neural network | |
CN110378985A (en) | A kind of animation drawing auxiliary creative method based on GAN | |
US8508546B2 (en) | Image mask generation | |
CN108830913B (en) | Semantic level line draft coloring method based on user color guidance | |
Žeger et al. | Grayscale image colorization methods: Overview and evaluation | |
CN103914699A (en) | Automatic lip gloss image enhancement method based on color space | |
CN108830912A (en) | A kind of interactive grayscale image color method of depth characteristic confrontation type study | |
Hwang et al. | Image colorization with deep convolutional neural networks | |
CN114723643B (en) | Low-light image enhancement method based on reinforcement learning and aesthetic evaluation | |
CN108921932B (en) | Method for generating multiple reasonable colorings of black and white figure pictures based on convolutional neural network | |
CN112991371B (en) | Automatic image coloring method and system based on coloring overflow constraint | |
CN108388905A (en) | A kind of Illuminant estimation method based on convolutional neural networks and neighbourhood context | |
Zhang et al. | Inkthetics: a comprehensive computational model for aesthetic evaluation of Chinese ink paintings | |
CN114581356B (en) | Image enhancement model generalization method based on style migration data augmentation | |
Liu et al. | Very lightweight photo retouching network with conditional sequential modulation | |
CN112819096A (en) | Method for constructing fossil image classification model based on composite convolutional neural network | |
Gain et al. | A novel unbiased deep learning approach (dl-net) in feature space for converting gray to color image | |
CN102184403A (en) | Optimization-based intrinsic image extraction method | |
CN115018729B (en) | Content-oriented white box image enhancement method | |
Jimenez-Arredondo et al. | Multilevel color transfer on images for providing an artistic sight of the world | |
CN114299184B (en) | Hidden building colored drawing line manuscript painting method and device based on semantic matching | |
Halder et al. | Perceptual conditional generative adversarial networks for end-to-end image colourization | |
CN114917583A (en) | Animation style game background generation method and platform based on generation confrontation network | |
CN114187380A (en) | Color transfer method based on visual saliency and channel attention mechanism |
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 |