CN108399422A - A kind of image channel fusion method based on WGAN models - Google Patents
A kind of image channel fusion method based on WGAN models Download PDFInfo
- Publication number
- CN108399422A CN108399422A CN201810101708.4A CN201810101708A CN108399422A CN 108399422 A CN108399422 A CN 108399422A CN 201810101708 A CN201810101708 A CN 201810101708A CN 108399422 A CN108399422 A CN 108399422A
- Authority
- CN
- China
- Prior art keywords
- image
- channel
- convolution
- generator
- fusion method
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Abstract
The invention discloses a kind of image channel fusion methods based on WGAN models, belong to deep learning field of neural networks, include the following steps：S1, construction are originally generated confrontation network model；S2, construction Wo Sesitan distances, the judging quota as confrontation network model；S3, initialization random noise, input in generator；S4, in WGAN models using channel merge to image carry out convolution operation；S5, the loss function input generator that channel mixing operation obtains subsequently is trained.The WGAN models based on channel fusion of this method structure change the convolution mode after arbiter, generator reception picture, are all merged to the characteristic pattern in each channel of image after each convolution, so as to improve the robustness of whole network training pattern.
Description
Technical field
The present invention relates to deep learning neural networks, and in particular to a kind of image channel fusion side based on WGAN models
Method.
Background technology
It is by Goodfellow that production, which fights network (Generative Adversarial Network, abbreviation GAN),
The deep learning frame proposed in 2014, it is based on the thought of " game theory ", construction generator (generator) and arbiter
(discriminator) two kinds of models, the former generates image by the Uniform noise or gaussian random noise of input (0,1), after
Person differentiates the image of input, and determination is image from data set or the image generated by generator.
In traditional confrontation network model, picture quality is generated there is no unified judgment criteria for generator, because
This, needs those skilled in the art to choose a kind of rational parameter as the judging quota for generating confrontation network, can make whole
The training of a model can be toward progress be correctly oriented, simultaneously, it would be highly desirable to be proposed a kind of Feature fusion, be improved deep learning god
Robustness through network.
Invention content
The purpose of the present invention is to solve drawbacks described above in the prior art, provide a kind of image based on WGAN models
Channel fusion method, this method is using Wo Sesitan distances as the judging quota for generating confrontation network, to make entire model
Training can learn the method for characteristics of image furthermore with channel fusion toward being correctly oriented progresss, each layer of convolution it
All the characteristic pattern in each channel of image is merged afterwards, improves the robustness of whole network.
The purpose of the present invention can be reached by adopting the following technical scheme that：
A kind of image channel fusion method based on WGAN models, the image channel fusion method include following step
Suddenly：
S1, construction are originally generated confrontation network model, generate image by generator and are input to arbiter progress network
Training；
S2, construction Wo Sesitan distances, the judging quota as confrontation network model；
S3, initialization random noise, input in generator；
S4, in WGAN models using channel merge to image carry out convolution operation；
S5, the loss function input generator that channel mixing operation obtains subsequently is trained.
Further, the step S4 processes are as follows：
S41, by picture breakdown at multiple channels；
S42, each channel for image carry out convolution using different convolution kernels；
S43, after each layer of convolution is completed, the characteristic pattern in all channels is merged.
Further, the step S5 processes are as follows：
S51, by the characteristic pattern after convolution in step S4, input arbiter is differentiated；
S52, the loss function input generator that channel mixing operation obtains subsequently is trained.
Further, the expression formula of the loss function is：
Wherein, D (x) indicates that differentiation of the arbiter to image, pr indicate that the distribution of data images, pg indicate to generate image
Distribution, λ is hyper parameter,For gradient.
The present invention has the following advantages and effects with respect to the prior art：
1, in traditional model, the characteristic pattern in each channel of image after multilayer convolution by just being merged, we
The method that method uses channel fusion all merges the characteristic pattern in each channel of image after each layer of convolution terminates,
It is subsequently trained again.In this case, the robustness of whole network is further enhanced.
2, in the present invention, using Wo Sesitan distances as the judging quota for generating confrontation network, to make entire mould
The training of type past can be correctly oriented progress
Description of the drawings
Fig. 1 is the training flow chart of the image channel fusion method based on WGAN models in the present invention；
Fig. 2 is the schematic diagram merged into row of channels for specific convolutional layer.
Specific implementation mode
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art
The every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
Embodiment
As shown in Figure 1 and Figure 2.Present embodiment discloses a kind of image channel fusion method based on WGAN models,
Specifically include the following steps：
Step S1, construction is originally generated confrontation network model, generates image by generator and is input to arbiter progress
Network training.
Step S2, Wo Sesitan distances are constructed, the judging quota as confrontation network model；
Different convolution kernels is embodied in the difference of matrix numerical value, the difference of ranks number.
Multiple convolution kernels are constructed, during handling image, different convolution kernels is meant in network training
Different characteristic of the study to generation image in the process.
It is involved in the present invention to network model in, using Wo Sesitan distance as generate fight network judge refer to
Mark, to enable the training of entire model is past to be correctly oriented progress.
In traditional model, the characteristic pattern in each channel of image after multilayer convolution by just being merged, this method
The method for using channel fusion all merges the characteristic pattern in each channel of image after each layer of convolution terminates, then
Subsequently trained.In this case, the robustness of whole network is further enhanced.
In practical applications, it should which, according to the complexity of data images feature, the number of convolution kernel is set.
Step S3, random noise is initialized, is inputted in generator.
Step S4, it is merged using channel in WGAN models and convolution operation is carried out to image.
In a specific embodiment, the step is specific as follows：
S41, by picture breakdown at multiple channels；
S42, each channel for image carry out convolution using different convolution kernels；
S43, after each layer of convolution is completed, the characteristic pattern in all channels is merged, is subsequently trained.
Step S5, the loss function input generator that channel mixing operation obtains subsequently is trained.Detailed process is such as
Under：
S51, by the characteristic pattern after convolution in step S4, input arbiter is differentiated；
S52, the loss function input generator that channel mixing operation obtains subsequently is trained.
The effect of loss function is the ability weighed arbiter and judged generating image.The value of loss function is smaller, explanation
In current iteration, arbiter can have the generation image of preferable performance discrimination generator；Property that is on the contrary then illustrating arbiter
It can be poor.
The expression formula of loss function is：
Wherein, D (x) indicates that differentiation of the arbiter to image, pr indicate that the distribution of data images, pg indicate to generate image
Distribution, λ is hyper parameter,For gradient.
In conclusion present embodiment discloses a kind of image channel fusion methods based on WGAN models, by introducing
Wo Sesitan distances so that whole network has specific direction and judgment criteria during training.In traditional model
In, the characteristic pattern in each channel of image after multilayer convolution by just being merged, present invention employs the method for channel fusion,
After each layer of convolution terminates, all the characteristic pattern in each channel of image is merged, then is subsequently trained.In this feelings
Under condition, the robustness of whole network is further enhanced.
The above embodiment is a preferred embodiment of the present invention, but embodiments of the present invention are not by above-described embodiment
Limitation, it is other it is any without departing from the spirit and principles of the present invention made by changes, modifications, substitutions, combinations, simplifications,
Equivalent substitute mode is should be, is included within the scope of the present invention.
Claims (4)
1. a kind of image channel fusion method based on WGAN models, which is characterized in that the image channel fusion method packet
Include following steps：
S1, construction are originally generated confrontation network model, generate image by generator and are input to arbiter progress network training；
S2, construction Wo Sesitan distances, the judging quota as confrontation network model；
S3, initialization random noise, input in generator；
S4, in WGAN models using channel merge to image carry out convolution operation；
S5, the loss function input generator that channel mixing operation obtains subsequently is trained.
2. a kind of image channel fusion method based on WGAN models according to claim 1, which is characterized in that described
Step S4 processes are as follows：
S41, by picture breakdown at multiple channels；
S42, each channel for image carry out convolution using different convolution kernels；
S43, after each layer of convolution is completed, the characteristic pattern in all channels is merged.
3. a kind of image channel fusion method based on WGAN models according to claim 1, which is characterized in that described
Step S5 processes are as follows：
S51, by the characteristic pattern after convolution in step S4, input arbiter is differentiated；
S52, the loss function input generator that channel mixing operation obtains subsequently is trained.
4. a kind of image channel fusion method based on WGAN models according to claim 1, which is characterized in that described
The expression formula of loss function is：
Wherein, D (x) indicates that differentiation of the arbiter to image, pr indicate that the distribution of data images, pg indicate to generate point of image
Cloth, λ are hyper parameter,For gradient.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810101708.4A CN108399422A (en) | 2018-02-01 | 2018-02-01 | A kind of image channel fusion method based on WGAN models |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810101708.4A CN108399422A (en) | 2018-02-01 | 2018-02-01 | A kind of image channel fusion method based on WGAN models |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108399422A true CN108399422A (en) | 2018-08-14 |
Family
ID=63095727
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810101708.4A Pending CN108399422A (en) | 2018-02-01 | 2018-02-01 | A kind of image channel fusion method based on WGAN models |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108399422A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109118467A (en) * | 2018-08-31 | 2019-01-01 | 武汉大学 | Based on the infrared and visible light image fusion method for generating confrontation network |
| CN109636768A (en) * | 2018-12-12 | 2019-04-16 | 中国科学院深圳先进技术研究院 | Remote sensing image fusing method, device and electronic equipment |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1688872A3 (en) * | 2005-02-04 | 2009-12-30 | Bernard Angeniol | Informatics tool for prediction |
| CN107122826A (en) * | 2017-05-08 | 2017-09-01 | 京东方科技集团股份有限公司 | Processing method and system and storage medium for convolutional neural networks |
| CN107273936A (en) * | 2017-07-07 | 2017-10-20 | 广东工业大学 | A kind of GAN image processing methods and system |
| WO2017210690A1 (en) * | 2016-06-03 | 2017-12-07 | Lu Le | Spatial aggregation of holistically-nested convolutional neural networks for automated organ localization and segmentation in 3d medical scans |
| CN107563510A (en) * | 2017-08-14 | 2018-01-09 | 华南理工大学 | A kind of WGAN model methods based on depth convolutional neural networks |
| CN107563493A (en) * | 2017-07-17 | 2018-01-09 | 华南理工大学 | A kind of confrontation network algorithm of more maker convolution composographs |
| CN107590531A (en) * | 2017-08-14 | 2018-01-16 | 华南理工大学 | A kind of WGAN methods based on text generation |
-
2018
- 2018-02-01 CN CN201810101708.4A patent/CN108399422A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1688872A3 (en) * | 2005-02-04 | 2009-12-30 | Bernard Angeniol | Informatics tool for prediction |
| WO2017210690A1 (en) * | 2016-06-03 | 2017-12-07 | Lu Le | Spatial aggregation of holistically-nested convolutional neural networks for automated organ localization and segmentation in 3d medical scans |
| CN107122826A (en) * | 2017-05-08 | 2017-09-01 | 京东方科技集团股份有限公司 | Processing method and system and storage medium for convolutional neural networks |
| CN107273936A (en) * | 2017-07-07 | 2017-10-20 | 广东工业大学 | A kind of GAN image processing methods and system |
| CN107563493A (en) * | 2017-07-17 | 2018-01-09 | 华南理工大学 | A kind of confrontation network algorithm of more maker convolution composographs |
| CN107563510A (en) * | 2017-08-14 | 2018-01-09 | 华南理工大学 | A kind of WGAN model methods based on depth convolutional neural networks |
| CN107590531A (en) * | 2017-08-14 | 2018-01-16 | 华南理工大学 | A kind of WGAN methods based on text generation |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109118467A (en) * | 2018-08-31 | 2019-01-01 | 武汉大学 | Based on the infrared and visible light image fusion method for generating confrontation network |
| CN109118467B (en) * | 2018-08-31 | 2021-11-16 | 武汉大学 | Infrared and visible light image fusion method based on generation countermeasure network |
| CN109636768A (en) * | 2018-12-12 | 2019-04-16 | 中国科学院深圳先进技术研究院 | Remote sensing image fusing method, device and electronic equipment |
| CN109636768B (en) * | 2018-12-12 | 2021-05-25 | 中国科学院深圳先进技术研究院 | Remote sensing image fusion method and device and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10755173B2 (en) | Video deblurring using neural networks | |
| CN104978715B (en) | A kind of non-local mean image de-noising method based on filter window and parameter adaptive | |
| CN107563493A (en) | A kind of confrontation network algorithm of more maker convolution composographs | |
| CN107590518A (en) | A kind of confrontation network training method of multiple features study | |
| CN107767413A (en) | A kind of image depth estimation method based on convolutional neural networks | |
| CN107871142A (en) | A kind of empty convolution method based on depth convolution confrontation network model | |
| CN108460720A (en) | A method of changing image style based on confrontation network model is generated | |
| CN104143186B (en) | A kind of SLIC super-pixel piecemeal optimization method | |
| CN107886162A (en) | A kind of deformable convolution kernel method based on WGAN models | |
| CN109829868B (en) | Lightweight deep learning model image defogging method, electronic equipment and medium | |
| WO2021189807A1 (en) | Image processing method, apparatus and system, and electronic device | |
| CN107689035A (en) | A kind of homography matrix based on convolutional neural networks determines method and device | |
| CN108805166B (en) | It is a kind of to establish image classification neural network model and image classification method, device | |
| CN107563509A (en) | A kind of dynamic adjustment algorithm for the condition DCGAN models that feature based returns | |
| CN108009568A (en) | A kind of pedestrian detection method based on WGAN models | |
| CN109858487A (en) | Weakly supervised semantic segmentation method based on watershed algorithm and image category label | |
| CN109344879A (en) | A kind of decomposition convolution method fighting network model based on text-image | |
| CN107274431A (en) | video content enhancement method and device | |
| CN112004078B (en) | Virtual reality video quality evaluation method and system based on generation countermeasure network | |
| CN107992944A (en) | It is a kind of based on be originally generated confrontation network model multiple dimensioned convolution method | |
| CN108470196A (en) | A method of handwritten numeral is generated based on depth convolution confrontation network model | |
| CN109360159A (en) | A kind of image completion method based on generation confrontation network model | |
| CN108399422A (en) | A kind of image channel fusion method based on WGAN models | |
| CN107564063A (en) | A kind of virtual object display methods and device based on convolutional neural networks | |
| CN107943750A (en) | A kind of decomposition convolution method based on WGAN models |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180814 |