CN110033419A - A kind of processing method being adapted to warship basic image defogging - Google Patents
A kind of processing method being adapted to warship basic image defogging Download PDFInfo
- Publication number
- CN110033419A CN110033419A CN201910308652.4A CN201910308652A CN110033419A CN 110033419 A CN110033419 A CN 110033419A CN 201910308652 A CN201910308652 A CN 201910308652A CN 110033419 A CN110033419 A CN 110033419A
- Authority
- CN
- China
- Prior art keywords
- image
- warship
- algorithm
- defogging
- neural networks
- 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
- 238000003672 processing method Methods 0.000 title claims abstract description 11
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 23
- 238000013135 deep learning Methods 0.000 claims abstract description 13
- 230000003044 adaptive effect Effects 0.000 claims abstract description 12
- 230000006870 function Effects 0.000 claims abstract description 11
- 230000009466 transformation Effects 0.000 claims abstract description 4
- 230000001131 transforming effect Effects 0.000 claims abstract description 4
- 238000013316 zoning Methods 0.000 claims abstract description 4
- 238000000034 method Methods 0.000 claims description 10
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 230000008030 elimination Effects 0.000 claims description 3
- 238000003379 elimination reaction Methods 0.000 claims description 3
- 239000003595 mist Substances 0.000 claims description 3
- 238000012876 topography Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 5
- 238000003062 neural network model Methods 0.000 abstract description 3
- 230000011514 reflex Effects 0.000 abstract description 3
- 238000001514 detection method Methods 0.000 description 2
- 230000007812 deficiency Effects 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/73—Deblurring; Sharpening
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; 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
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
A kind of processing method being adapted to warship basic image defogging, S1. obtain warship base RGB original image;S2. judge whether the original image obtained needs to handle by image defogging;S3. the image for needing to carry out image defogging processing is handled using adaptive histogram equalization algorithm;S4. using self-adapting histogram equilibrium algorithm and bilinear interpolation algorithm to each pixel zoning histogram and corresponding transforming function transformation function in image;S5. it adaptively brightens and denoises to realize using the feature of the different low-light (level) picture signals of convolutional neural networks training of deep learning.Aerial steam reflex is distributed in the contrast of image in order to reduce, the present invention improves the contrast of image using a kind of algorithm of adaptive histogram equalization combination deep neural network model, to realize the effect so that image clearly, reach the ability to warship basic image defogging.
Description
Technical field
The present invention relates to a kind of processing methods for being adapted to warship basic image defogging, belong to the skill of computer visual image processing
Art field.
Background technique
Image processing techniques is the important research field of current information technology.Target detection based on warship basic image can be real
When effectively find target, for warship safety navigation effective Information Assurance is provided.At sea due to naval vessel traveling, it gets
Warship basic image be easy to be influenced by foggy weather.Warship basic image with fog is brought to the object detection and recognition of image
Greatly challenge and difficulty, needs image carrying out defogging processing thus, obtains relatively clear image.
Original image with fog is more fuzzy, and the reflex that main cause is distributed across aerial steam makes figure
The contrast of picture reduces.In order to make image clearly just need to improve the contrast of image, used here as a kind of self-adapting histogram
The algorithm in conjunction with deep neural network model is equalized to improve the contrast of image, to realize the effect so that image clearly
Fruit reaches the ability of image defogging.
Summary of the invention
In view of the deficiencies of the prior art, the present invention discloses a kind of processing method for being adapted to warship basic image defogging.The present invention
So that the warship base original image got is apparent after image procossing, convenient for the use in later period.
Technical scheme is as follows:
A kind of processing method being adapted to warship basic image defogging characterized by comprising
S1. warship base RGB original image is obtained, wherein what the warship base RGB original image obtained was made of three kinds of Color Channels;
S2. judge whether the original image obtained needs to handle by image defogging: the RGB color of clear image block
In have a Color Channel very dark, numerical value is very low or even close to zero;The part for analyzing image, if some face of topography
Chrominance channel value is less than other color channel values, then it is assumed that the image is substantially fogless, does not need to carry out defogging;Conversely, then needing
Carry out mist elimination image processing;
S3. the image for needing to carry out image defogging processing is handled using adaptive histogram equalization algorithm;
S4. using self-adapting histogram equilibrium algorithm and bilinear interpolation algorithm to each pixel zoning in image
Histogram and corresponding transforming function transformation function;To improve calculating speed;
S5. the feature using the different low-light (level) picture signals of convolutional neural networks training of deep learning is adaptive to realize
It brightens and denoises;There is larger promotion more originally using the picture contrast after adaptive histogram equalization, but equally increased
The noise in image is added, therefore, the convolutional neural networks training of deep learning has been introduced in the present invention, to eliminate self-adaptive direct
Bring noise after side's figure equalization.
Preferred according to the present invention, the convolutional neural networks of the deep learning are CNN convolutional neural networks, convolution mind
It include coding and decoded symmetrical structure, including symmetrical 12 layers of convolutional layer through network, wherein first 6 layers are coding layer, latter 6 layers are
Decoding layer;In the coding layer of the convolutional neural networks network, just coding layer is connected to symmetrically with a wire jumper every 2 layers
Decoding layer, therefore can directly carry out forward and reverse propagation.
Preferred according to the present invention, the convolutional neural networks of the deep learning further include neural network of making an uproar, with output with
The l2 norm of noise is loss function to train network.The technological merit of the technical characteristic is: the number of plies is easy more deeply
Lead to gradient disperse in training process, it is not easy to optimal solution is converged to, in order to solve gradient disperse caused by the network number of plies is deepened
Effect is similar to residual error learning process using present networks.
The invention also discloses a kind of devices for realizing the above method, which is characterized in that including for acquiring RGB image
Equipment, video storage equipment and computer;
The equipment and video storage equipment of the RGB image are connected with computer respectively;
The computer is the computer equipped with image processing algorithm, and described image Processing Algorithm includes: adaptive
Histogram equalization algorithm, bilinear interpolation algorithm and the convolutional neural networks using deep learning.
The technical advantages of the present invention are that:
Aerial steam reflex is distributed in the contrast of image in order to reduce, and the present invention uses a kind of self-adaptive direct
Side figure algorithm of the equalization in conjunction with deep neural network model improve the contrast of image, to realize so that image clearly
Effect reaches the ability to warship basic image defogging.
Detailed description of the invention
Fig. 1 is the flow chart of the method for the invention.
Specific embodiment
The present invention is described in detail below with reference to embodiment and Figure of description, but not limited to this.
As shown in Fig. 1.
Embodiment 1,
A kind of processing method being adapted to warship basic image defogging, comprising:
S1. warship base RGB original image is obtained, wherein what the warship base RGB original image obtained was made of three kinds of Color Channels;
S2. judge whether the original image obtained needs to handle by image defogging: the RGB color of clear image block
In have a Color Channel very dark, numerical value is very low or even close to zero;The part for analyzing image, if some face of topography
Chrominance channel value is less than other color channel values, then it is assumed that the image is substantially fogless, does not need to carry out defogging;Conversely, then needing
Carry out mist elimination image processing;
S3. the image for needing to carry out image defogging processing is handled using adaptive histogram equalization algorithm;
S4. using self-adapting histogram equilibrium algorithm and bilinear interpolation algorithm to each pixel zoning in image
Histogram and corresponding transforming function transformation function;To improve calculating speed;
S5. the feature using the different low-light (level) picture signals of convolutional neural networks training of deep learning is adaptive to realize
It brightens and denoises;There is larger promotion more originally using the picture contrast after adaptive histogram equalization, but equally increased
The noise in image is added, therefore, the convolutional neural networks training of deep learning has been introduced in the present invention, to eliminate self-adaptive direct
Bring noise after side's figure equalization.
Embodiment 2,
A kind of processing method being adapted to warship basic image defogging as described in Example 1, the convolution mind of the deep learning
It is CNN convolutional neural networks through network, which includes encoding and decoded symmetrical structure, including symmetrical 12 layers
Convolutional layer, wherein first 6 layers are coding layer, latter 6 layers are decoding layer;In the coding layer of the convolutional neural networks network, every 2 layers
With regard to coding layer is connected to symmetrical decoding layer with a wire jumper, therefore it can directly carry out forward and reverse propagation.
The convolutional neural networks of the deep learning further include neural network of making an uproar, to export with the l2 norm of noise as loss
Function trains network.The technological merit of the technical characteristic is: the number of plies is easy to lead to gradient in the training process more deeply
Disperse, it is not easy to optimal solution is converged to, in order to solve gradient dispersion effect caused by the network number of plies is deepened, using present networks,
Similar to residual error learning process.
Embodiment 3,
A kind of device for realizing the above method, including equipment, video storage equipment and the calculating for acquiring RGB image
Machine;
The equipment and video storage equipment of the RGB image are connected with computer respectively;
The computer is the computer equipped with image processing algorithm, and described image Processing Algorithm includes: adaptive
Histogram equalization algorithm, bilinear interpolation algorithm and the convolutional neural networks using deep learning.
Claims (4)
1. a kind of processing method for being adapted to warship basic image defogging characterized by comprising
S1. warship base RGB original image is obtained;
S2. judge whether the original image obtained needs to handle by image defogging: analyzing the part of image, if topography
Some color channel values is less than other color channel values, then it is assumed that the image is substantially fogless, does not need to carry out defogging;Instead
It, then need to carry out mist elimination image processing;
S3. the image for needing to carry out image defogging processing is handled using adaptive histogram equalization algorithm;
S4. using self-adapting histogram equilibrium algorithm and bilinear interpolation algorithm to each pixel zoning histogram in image
Figure and corresponding transforming function transformation function;
S5. it is adaptively brightened using the feature of the different low-light (level) picture signals of convolutional neural networks training of deep learning with realizing
And denoising.
2. a kind of processing method for being adapted to warship basic image defogging according to claim 1, which is characterized in that the depth
The convolutional neural networks of study are CNN convolutional neural networks, which includes coding and decoded symmetrical structure, packet
Symmetrical 12 layers of convolutional layer is included, wherein first 6 layers are coding layer, latter 6 layers are decoding layer;In the coding of the convolutional neural networks network
In layer, coding layer is just connected to symmetrical decoding layer with a wire jumper every 2 layers.
3. a kind of processing method for being adapted to warship basic image defogging according to claim 1, which is characterized in that the depth
The convolutional neural networks of study further include neural network of making an uproar, and train network as loss function using output and the l2 norm of noise.
4. a kind of device for realizing method as described in claim 1, which is characterized in that including for acquiring RGB image equipment,
Video storage equipment and computer;
The equipment and video storage equipment of the RGB image are connected with computer respectively;
The computer is the computer equipped with image processing algorithm, and described image Processing Algorithm includes: adaptive histogram
Figure equalization algorithm, bilinear interpolation algorithm and the convolutional neural networks using deep learning.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910308652.4A CN110033419A (en) | 2019-04-17 | 2019-04-17 | A kind of processing method being adapted to warship basic image defogging |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910308652.4A CN110033419A (en) | 2019-04-17 | 2019-04-17 | A kind of processing method being adapted to warship basic image defogging |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110033419A true CN110033419A (en) | 2019-07-19 |
Family
ID=67238827
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910308652.4A Pending CN110033419A (en) | 2019-04-17 | 2019-04-17 | A kind of processing method being adapted to warship basic image defogging |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110033419A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113012168A (en) * | 2021-03-24 | 2021-06-22 | 哈尔滨理工大学 | Brain glioma MRI image segmentation method based on convolutional neural network |
CN113465497A (en) * | 2020-03-31 | 2021-10-01 | 宝山钢铁股份有限公司 | Hot-rolled strip steel movement position online detection method capable of overcoming water mist interference |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104794705A (en) * | 2015-03-27 | 2015-07-22 | 努比亚技术有限公司 | Image defogging method and method based on image partial content characteristics |
CN105225210A (en) * | 2015-10-14 | 2016-01-06 | 南京第五十五所技术开发有限公司 | A kind of self-adapting histogram based on dark strengthens defogging method capable |
CN105574830A (en) * | 2016-02-04 | 2016-05-11 | 沈阳工业大学 | Low-quality image enhancement method under extreme weather conditions |
CN106296612A (en) * | 2016-08-09 | 2017-01-04 | 南京工业大学 | Hierarchical monitoring video sharpening system and method for image quality evaluation and weather condition guidance |
CN107248144A (en) * | 2017-04-27 | 2017-10-13 | 东南大学 | A kind of image de-noising method based on compression-type convolutional neural networks |
CN107516330A (en) * | 2017-08-31 | 2017-12-26 | 上海联影医疗科技有限公司 | Model generating method, image processing method and medical imaging devices |
CN107895145A (en) * | 2017-10-31 | 2018-04-10 | 南京信息工程大学 | Method based on convolutional neural networks combination super-Gaussian denoising estimation finger stress |
CN108460743A (en) * | 2018-03-19 | 2018-08-28 | 西安因诺航空科技有限公司 | A kind of unmanned plane image defogging algorithm based on dark |
CN109242788A (en) * | 2018-08-21 | 2019-01-18 | 福州大学 | One kind being based on coding-decoding convolutional neural networks low-light (level) image optimization method |
-
2019
- 2019-04-17 CN CN201910308652.4A patent/CN110033419A/en active Pending
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104794705A (en) * | 2015-03-27 | 2015-07-22 | 努比亚技术有限公司 | Image defogging method and method based on image partial content characteristics |
CN105225210A (en) * | 2015-10-14 | 2016-01-06 | 南京第五十五所技术开发有限公司 | A kind of self-adapting histogram based on dark strengthens defogging method capable |
CN105574830A (en) * | 2016-02-04 | 2016-05-11 | 沈阳工业大学 | Low-quality image enhancement method under extreme weather conditions |
CN106296612A (en) * | 2016-08-09 | 2017-01-04 | 南京工业大学 | Hierarchical monitoring video sharpening system and method for image quality evaluation and weather condition guidance |
CN107248144A (en) * | 2017-04-27 | 2017-10-13 | 东南大学 | A kind of image de-noising method based on compression-type convolutional neural networks |
CN107516330A (en) * | 2017-08-31 | 2017-12-26 | 上海联影医疗科技有限公司 | Model generating method, image processing method and medical imaging devices |
CN107895145A (en) * | 2017-10-31 | 2018-04-10 | 南京信息工程大学 | Method based on convolutional neural networks combination super-Gaussian denoising estimation finger stress |
CN108460743A (en) * | 2018-03-19 | 2018-08-28 | 西安因诺航空科技有限公司 | A kind of unmanned plane image defogging algorithm based on dark |
CN109242788A (en) * | 2018-08-21 | 2019-01-18 | 福州大学 | One kind being based on coding-decoding convolutional neural networks low-light (level) image optimization method |
Non-Patent Citations (2)
Title |
---|
KAI ZHANG等: "Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising", 《ARXIV》 * |
XIAO-JIAO MAO等: "Image Restoration Using Very Deep Convolutional Encoder-Decoder Networks with Symmetric Skip Connections", 《ARXIV》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113465497A (en) * | 2020-03-31 | 2021-10-01 | 宝山钢铁股份有限公司 | Hot-rolled strip steel movement position online detection method capable of overcoming water mist interference |
CN113012168A (en) * | 2021-03-24 | 2021-06-22 | 哈尔滨理工大学 | Brain glioma MRI image segmentation method based on convolutional neural network |
CN113012168B (en) * | 2021-03-24 | 2022-11-04 | 哈尔滨理工大学 | Brain glioma MRI image segmentation method based on convolutional neural network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20200404340A1 (en) | Multi-frame quality enhancement method and device for lossy compressed video | |
CN105225210B (en) | A kind of self-adapting histogram enhancing defogging method based on dark | |
CN108650201B (en) | Neural network-based channel equalization method, decoding method and corresponding equipment | |
CN105205792A (en) | Underwater image enhancement method based on brightness and chrominance separation | |
CN106530257A (en) | Remote sensing image de-fogging method based on dark channel prior model | |
CN111145123B (en) | Image denoising method based on U-Net fusion retention details | |
CN107203977A (en) | A kind of underwater image restoration method based on dark primary priori and rarefaction representation | |
CN110033419A (en) | A kind of processing method being adapted to warship basic image defogging | |
CN106412383A (en) | Processing method and apparatus of video image | |
CN107085833A (en) | Remote sensing images filtering method based on the equal intermediate value fusion of gradient self-adaptive switch reciprocal | |
CN104751421A (en) | Method for achieving image defogging on FPGA | |
Sanila et al. | Underwater image enhancement using white balance, USM and CLHE | |
CN111476744B (en) | Underwater image enhancement method based on classification and atmospheric imaging model | |
Yu et al. | Image and video dehazing using view-based cluster segmentation | |
CN110136079A (en) | Image defogging method based on scene depth segmentation | |
Hegde et al. | Adaptive cubic spline interpolation in cielab color space for underwater image enhancement | |
CN111192213B (en) | Image defogging self-adaptive parameter calculation method, image defogging method and system | |
RU2008144840A (en) | METHOD AND DEVICE FOR CALCULATION AND FILTRATION OF THE CARD OF DISPARABILITY BASED ON STEREO IMAGES | |
CN115909064A (en) | Marine ship target detection method based on deep learning in sea fog environment | |
CN109003247B (en) | Method for removing color image mixed noise | |
CN112907470A (en) | Underwater image recovery method based on Lab color gamut transformation, classification and white balance | |
Liu et al. | Parameter-adaptive compensation (PAC) for processing underwater selective absorption | |
CN116503259A (en) | Mosaic interpolation method and system | |
CN114820395A (en) | Underwater image enhancement method based on multi-field information fusion | |
CN115034985A (en) | Underwater image enhancement method |
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: 20190719 |