CN109978776A - Defogging system and method based on Morphological Reconstruction and adaptive-filtering - Google Patents
Defogging system and method based on Morphological Reconstruction and adaptive-filtering Download PDFInfo
- Publication number
- CN109978776A CN109978776A CN201910009877.XA CN201910009877A CN109978776A CN 109978776 A CN109978776 A CN 109978776A CN 201910009877 A CN201910009877 A CN 201910009877A CN 109978776 A CN109978776 A CN 109978776A
- Authority
- CN
- China
- Prior art keywords
- adaptive
- image
- pixel
- module
- value
- 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
- 238000001914 filtration Methods 0.000 title claims abstract description 30
- 238000000034 method Methods 0.000 title claims abstract description 18
- 230000000877 morphologic effect Effects 0.000 title claims abstract description 15
- 230000003044 adaptive effect Effects 0.000 claims abstract description 69
- 238000004364 calculation method Methods 0.000 claims description 41
- 239000000872 buffer Substances 0.000 claims description 24
- 238000010586 diagram Methods 0.000 claims description 20
- 238000011084 recovery Methods 0.000 claims description 19
- 238000002834 transmittance Methods 0.000 claims description 17
- 238000013316 zoning Methods 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 4
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000009738 saturating Methods 0.000 claims 1
- 230000000694 effects Effects 0.000 abstract description 4
- 230000005764 inhibitory process Effects 0.000 abstract 1
- 238000004904 shortening Methods 0.000 abstract 1
- 241000224489 Amoeba Species 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000003595 mist Substances 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000002203 pretreatment Methods 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
Classifications
-
- G06T5/70—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/155—Segmentation; Edge detection involving morphological operators
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of defogging system and method based on Morphological Reconstruction and adaptive-filtering, optimizes adaptive filter algorithm, thus more efficiently inhibition halation;It is proposed the hardware system circuit of parallel pipelining process line architecture, time needed for greatly shortening defogging.By the above-mentioned means, the present invention can optimize defog effect, system performance is improved.
Description
Technical field
The invention belongs to technical field of image recovery, in particular to a kind of image defogging system and prevention.
Background technique
In recent years, in outdoor monitoring system, advanced DAS (Driver Assistant System) and unmanned plane control loop, to embedded
The demand of intelligence system is increasing.Due to atmospheric scattering, picture quality is significantly reduced because of haze.The visibility of blurred picture
Deficiency will lead to intelligence system failure.Therefore, in such real-time system, a kind of sufficiently fast image preprocessing is needed to calculate
Method eliminates haze.
Traditionally, defogging is to be realized by software, and operate on CPU, DSP and GPU.The realization face of these methods
Face massive store, the long challenge for calculating time and delay, this is far beyond actual time demand.Since low calculating is imitated
Rate and high energy consumption, above-mentioned processor are not suitable for needing the mobile device of real-time defogging.Therefore, in real-time embedded system,
Hardware realization is vital.
Summary of the invention
The defogging system and method based on Morphological Reconstruction and adaptive-filtering that the purpose of the present invention is to provide a kind of, with
Solve above-mentioned technical problem.The present invention realizes image dehazing function by hardware system: it is adaptive to be able to use hardware circuit realization
It should filter, can be realized the parallel pipeline architecture design of defogging, to realize defogging pre-treatment in real time embedded system
Application.
To achieve the goals above, the present invention adopts the following technical scheme:
Defogging system based on Morphological Reconstruction and adaptive-filtering, comprising:
Three 2 row buffers, two 4 row buffers, RGB image turn gray level image module, transmittance calculation module, adaptive
Dark estimation module, air light value estimation module, fogless scene recovery module, maximum value calculation circuit, minimum value is answered to calculate
Circuit and two register array groups;
The first of the input terminal of the input terminal connection max calculation circuit of three 2 row buffers and fogless scene recovery module
Input terminal;
The first input end of the output end connection transmittance calculation module of maximum value calculation circuit, transmittance calculation module
Output end connects the second input terminal of fogless scene recovery module;
The output end that RGB image turns gray level image module passes sequentially through the one 4 row buffer, the connection of the first register group certainly
Adapt to the first input end of dark estimation module;
It is adaptive dark that the output end of minimum value counting circuit passes sequentially through the 2nd 4 row buffer, the connection of the second register group
Second input terminal of channel estimation module;
Second input terminal of the output end connection transmittance calculation module of adaptive dark estimation module;
Second register group connects the input terminal of air light value estimation module, the output end connection of air light value estimation module
The third input terminal of fogless scene recovery module.
Further, the adaptive dark estimation module is supported the calculating of self-adapting changeable structural elements and is exported adaptive
Answer dark channel diagram;
Transmittance calculation module is for calculating transmissivity t;
Air light value estimation module is for calculating air light value A;
Fogless scene recovery module is for restoring foggy image.
Defogging method based on Morphological Reconstruction and adaptive-filtering, comprising the following steps:
Step 1: gray level image module being turned by RGB image, gradation conversion is carried out to original foggy image, by gray level image
The adaptive structure member growth district Ω that radius is 2 is obtained by the one 4 row bufferk, calculate self-adapting changeable structural elements;
Step 2: original foggy image being carried out respectively by maximum value calculation circuit and minimum value counting circuit maximum, most
Small value filtering obtains bright channel figure and dark channel image;Dark channel diagram is obtained and adaptive structure by the 2nd 4 row buffer
The identical Filtering Template of first size;
Step 3: adaptive dark estimation module passes through the self-adapting changeable structural elements that step 1 obtains, and obtains to step 2
Dark channel diagram carry out self-adapting changeable structural elements mini-value filtering, obtain adaptive dark channel diagram;
Step 4: transmittance calculation module calculates separately out dark by adaptive dark channel image and bright channel image
Transmissivity tdcWith bright channel transmissivity tbc, then compare the larger value of the two as final optimization transmissivity t;
Step 5: air light value estimation module calculates air light value A by the dark channel diagram that step 2 obtains;
Step 6: fogless scene recovery module calculates the image after obtaining defogging by formula (1):
Further, the first register group and the second register group are used to realize dark channel diagram and adaptive structure element
Delay matching;Three 2 row buffers match for realizing the delay of dark transmissivity and bright channel transmissivity.
Further, the input signal for calculating adaptive shape varistructure element is guiding figure p, gradient weights λ and knot
Constitutive element limits range Ωk;Wherein, guiding figure p is the gray level image of input picture, and gradient weights λ takes 0.18, structural element limit
Range Ω processedkRadius be 2;Output signal is the corresponding adaptive shape varistructure element of pixel each on image, packet
Include following steps:
Step 1): centered on current pixel point, taking side length is the sub-rectangular areas Ω of 2rk;
Step 2): zoning ΩkPixel value distance L (σ) of the interior each pixel along different paths to central point;
Step 3): pixel value distance d of the minimum pixel value distance as the point is chosenλ;
Step 4): zoning ΩkInterior structural elements length dλMean value, as structural elements length limit rlimit;
Step 5): comparison domain ΩkInterior each pixel structural elements length dλWith upper limit rlimitRelationship:Less
In rlimit, otherwise it is 0 that adaptive shape varistructure member corresponding position, which is 1, generate the variable knot of the point self-adapted shape of current pixel
Constitutive element;
When
When
Step 6): mobile current pixel point to next adjacent pixel repeats step 1)~step 5), until traversing complete width
Image.
Further, in guiding figure p, actionradius r=2, the rectangle frame centered on pixel k; For pixel x withBetween path;The calculating formula of length of path σ are as follows:
Wherein: n is the number of pixels in the σ of path, dpixel(Image(xi), Image (xi+1))=| p (xi)-p(xi+1)|
Self-adapting changeable row structural elements length is defined as:
Minimum value is on all paths of x and y;The mean value that adaptive shape varistructure member is all d apart from the upper limit;
M is the sum of all pixels in rectangular box;Adaptive shape varistructure member is defined as:
In defogging, the gray level image for inputting foggy image is used as calculating the guiding figure of adaptive structure member;Make
Mini-value filtering is executed on minimum Color Channel with self-adapting changeable structural elements;Filtered image is adaptive dark
Figure.
Further, calculate self-adapting changeable structural elements specifically includes the following steps:
Step (1): go out rectangle template Ω in first clock calculationkInterior 24 effective adjacent pixel distances;
Step (2): in the pixel value distance of second clock calculation, 24 pixel distance template central points;
Step (3): in the path length of 24 pixels of third clock calculation;
Step (4): in the 4th clock, r is calculated using equation (4)limit;
Step (5): adaptive shape varistructure member is obtained by accounting equation (5).
Compared with the existing technology, the invention has the following advantages:
(1) the adaptive minimum filtering device based on deformable structure member, has filtered out fine texture texture, keeps transmissivity flat
Sliding, the grain details of defogging result are apparent.
(2) adaptive filter algorithm is realized using hardware circuit, and adaptive filter algorithm is optimized, obtained
Higher-quality defog effect;The meter of mean square error (MSE) and structural similarity (SSIM) index is carried out for mist elimination image
It calculates, square mean error amount (SSIM) is maintained between 0.1 ± 0.02, and structural similarity value (SSIM) is maintained between 0.9 ± 0.05.
(3) architecture design for using parallel pipeline, improves performance, realizes defogging processing in real time embedded system
Using.
The invention proposes a kind of Morphological Reconstruction adaptive-filtering defogging methods;Adaptive-filtering dark channel image and dark
Template image and tag image of the channel image respectively as Morphological Reconstruction.It is made a return journey using adaptive-filtering and Morphological Reconstruction
Except halation.The experimental results showed that the algorithm is suitable for universal foggy image.It has good effect to removal halation.The calculation
Method is easy to parallel computation.There are two features for present system: 1) realizing adaptive filter algorithm, and to adaptive filter algorithm
It is optimized, to obtain higher-quality defog effect;2) architecture design for using parallel pipeline, improves performance, realizes
The function of real-time defogging.For the image of 640 × 480 sizes, when system work clock is 170MHz, processing speed 1.8ms.
Realize the result shows that, the defogging system performance with higher and quality.
Detailed description of the invention
The accompanying drawings constituting a part of this application is used to provide further understanding of the present invention, and of the invention shows
Examples and descriptions thereof are used to explain the present invention for meaning property, does not constitute improper limitations of the present invention.In the accompanying drawings:
Fig. 1 is a kind of architecture diagram of the defogging system based on Morphological Reconstruction and adaptive-filtering of the present invention.
Fig. 2 is adaptive structure member path, and wherein Fig. 2 (a) is space length, and Fig. 2 (b) is pixel value distance.
Fig. 3 is the hardware architecture diagram of adaptive structure member estimation.
Specific embodiment
The present invention will be described in detail below with reference to the accompanying drawings and embodiments.It should be noted that in the feelings not conflicted
Under condition, the features in the embodiments and the embodiments of the present application be can be combined with each other.
Following detailed description is exemplary explanation, it is intended to provide further be described in detail to the present invention.Unless another
It indicates, all technical terms of the present invention contain with the normally understood of the application one of ordinary skill in the art
Justice is identical.Term used in the present invention is merely to describe specific embodiment, and be not intended to limit according to the present invention
Illustrative embodiments.
Refering to Figure 1, a kind of defogging system based on Morphological Reconstruction and adaptive-filtering of the present invention, by three 2
Row buffer, two 4 row buffers, RGB image turn gray level image module, transmittance calculation module, the estimation of adaptive dark
Module, air light value estimation module, fogless scene recovery module, maximum value calculation circuit, minimum value counting circuit and two
Register array composition.
The first of the input terminal of the input terminal connection max calculation circuit of three 2 row buffers and fogless scene recovery module
Input terminal;
The first input end of the output end connection transmittance calculation module of maximum value calculation circuit, transmittance calculation module
Output end connects the second input terminal of fogless scene recovery module;
The output end that RGB image turns gray level image module passes sequentially through the one 4 row buffer, the connection of the first register group certainly
Adapt to the first input end of dark estimation module;
It is adaptive dark that the output end of minimum value counting circuit passes sequentially through the 2nd 4 row buffer, the connection of the second register group
Second input terminal of channel estimation module;
Second input terminal of the output end connection transmittance calculation module of adaptive dark estimation module;
Second register group connects the input terminal of air light value estimation module, the output end connection of air light value estimation module
The third input terminal of fogless scene recovery module.
Adaptive dark estimation module supports calculating and the output adaptive dark channel diagram of self-adapting changeable structural elements.Thoroughly
Penetrate rate computing module calculation optimization transmissivity t.Air light value estimation module calculates air light value A.Fogless scene recovery module is extensive
Multiple foggy image.
A kind of defogging method based on Morphological Reconstruction and adaptive-filtering of the present invention, steps are as follows for concrete implementation:
Step 1: gray level image module being turned by RGB image, gradation conversion is carried out to original foggy image, by gray level image
The adaptive structure member growth district Ω that radius is 2 is obtained by the one 4 row bufferk, calculate self-adapting changeable structural elements;
Step 2: original foggy image being carried out respectively by maximum value calculation circuit and minimum value counting circuit maximum, most
Small value filtering obtains bright channel figure and dark channel image.Dark channel diagram is obtained and adaptive structure by the 2nd 4 row buffer
The identical Filtering Template of first size.
Step 3: adaptive dark estimation module passes through the self-adapting changeable structural elements that step 1 obtains, and obtains to step 2
Dark channel diagram carry out self-adapting changeable structural elements mini-value filtering, obtain adaptive dark channel diagram.
Step 4: transmittance calculation module calculates separately out dark by adaptive dark channel image and bright channel image
Transmissivity tdcWith bright channel transmissivity tbc, then compare the larger value of the two as final optimization transmissivity t.
Dark transmissivityBright channel transmissivityWherein DC is adaptive dark channel diagram
Picture, BC are bright channel image;Optimize transmissivity t=max (tdc, tbc)。
Step 5: air light value estimation module is compared by the dark channel diagram that step 2 obtains with air light value threshold value,
Compare the larger value in the two as air light value A, air light value threshold value is 180;
Step 6: fogless scene recovery module calculates the image after obtaining defogging by formula (1):
Wherein, two register arrays are used to realize that dark channel diagram and the delay of adaptive structure element match.Three 2 rows
Buffer area matches for realizing the delay of dark transmissivity and bright channel transmissivity.
The present invention improves the processing of the mini-value filtering in defogging algorithm using " amoeba " method, to realize adaptive
Varistructure member mini-value filtering.
The input signal for calculating adaptive shape varistructure element is that guiding figure p, gradient weights λ and structural element limit
Range Ωk(rectangle that radius is r).Wherein, guiding figure p is the gray level image of input picture, and gradient weights λ takes 0.18, structure
Element limits range ΩkRadius be 2.Output signal is the corresponding adaptive shape varistructure member of pixel each on image
Element, calculating process are as follows:
Step 1): centered on current pixel point, taking side length is the sub-rectangular areas Ω of 2rk。
Step 2): zoning ΩkPixel value distance L (σ) of the interior each pixel along different paths to central point.
Step 3): pixel value distance d of the minimum pixel value distance as the point is chosenλ。
Step 4): zoning ΩkInterior structural elements length dλMean value, as structural elements length limit Flimit。
Step 5): comparison domain ΩkInterior each pixel structural elements length dλWith upper limit rlimitRelationship:Less
In rlimit, otherwise it is 0 that adaptive shape varistructure member corresponding position, which is 1, generate the variable knot of the point self-adapted shape of current pixel
Constitutive element.
When
When
Step 6): mobile current pixel point to next adjacent pixel repeats step 1)~step 5), until traversing complete width
Image;
In guiding figure p, actionradius r=2, the rectangle frame centered on pixel k.For pixel x withBetween path.The calculating formula of length of path σ are as follows:
Wherein: n is the number of pixels in the σ of path, dpixel(Image(xi), Image (xi+1))=| p (xi)-p(xi+1)|.
Self-adapting changeable row structural elements length is defined as:
Minimum value is on all paths of x and y.The mean value that adaptive shape varistructure member is all d apart from the upper limit.
M is the sum of all pixels in rectangular box.Adaptive shape varistructure member is defined as:
In defogging, the gray level image for inputting foggy image is used as calculating the guiding figure of adaptive structure member.Make
Mini-value filtering is executed on minimum Color Channel with self-adapting changeable structural elements.Filtered image is adaptive dark
Figure.
Adaptive dark estimation module is the main modular in present system structure.
The calculating of self-adapting changeable structural elements is the major part of adaptive dark estimation module, by equation group (2)-(5)
It realizes.In equation (3), fixed route is selected to calculate pixel value distance, omittedCalculating.Choosing
It selects 2 and limits range Ω for structural elementkRadius.Then the length of the fixed route between x-axis and y-axis is calculated.Hardware realization
Shi Fangcheng (2) is reduced to:
Wherein, dpIt is the pixel value distance of fixed route between two o'clock, D8For the space length of pixel to central point.Such as:
The path of pixel 1 and pixel 13 is (p1, p7, p13).The calculating of pixel value distance are as follows: | Image (x1)-Image (x7) |+|
Image(x7)-Image(x13)|。
For self-adapting changeable structural elements calculating section, framework is as shown in Figure 3.It is accomplished by
Step (1): go out rectangle template Ω in first clock calculationkInterior 24 effective adjacent pixel distances, such as the left side Fig. 3
Inferior horn, xiWith xi+1Shown in pixel value distance.
Step (2): in the pixel value distance of second clock calculation, 24 pixel distance template central points.This 24 pictures
Shown in the fixed route such as Fig. 2 (b) of vegetarian refreshments to central point.
Step (3): 24 pixels of third clock calculation path length (enter pixel value distance+space away from
From).Shown in space length such as Fig. 2 (a), calculation formula is equation (6).
Step (4): in the 4th clock, r is calculated using equation (4)limit.Wherein, M (including the central point d that is 251313),
Equation (4) is reduced to when hardware realization
Step (5): adaptive shape varistructure member is obtained by accounting equation (5).
The above description is only an embodiment of the present invention, is not intended to limit the scope of the invention, all to utilize this hair
The equivalent structure or equivalent flow shift that bright specification and accompanying drawing content are done is applied directly or indirectly in other relevant skills
Art field, is included within the scope of the present invention.
Claims (7)
1. the defogging system based on Morphological Reconstruction and adaptive-filtering characterized by comprising
Three 2 row buffers, two 4 row buffers, RGB image turn gray level image module, transmittance calculation module, adaptive dark
Channel estimation module, air light value estimation module, fogless scene recovery module, maximum value calculation circuit, minimum value counting circuit
And two register array groups;
The input terminal of the input terminal connection max calculation circuit of three 2 row buffers and the first input of fogless scene recovery module
End;
The first input end of the output end connection transmittance calculation module of maximum value calculation circuit, the output of transmittance calculation module
End connects the second input terminal of fogless scene recovery module;
The output end that RGB image turns gray level image module passes sequentially through the one 4 row buffer, the connection of the first register group adaptively
The first input end of dark estimation module;
The output end of minimum value counting circuit passes sequentially through the 2nd 4 row buffer, the second register group connects adaptive dark
Second input terminal of estimation module;
Second input terminal of the output end connection transmittance calculation module of adaptive dark estimation module;
Second register group connects the input terminal of air light value estimation module, and the output end connection of air light value estimation module is fogless
The third input terminal of scene recovery module.
2. the defogging system according to claim 1 based on Morphological Reconstruction and adaptive-filtering, which is characterized in that
The adaptive dark estimation module supports calculating and the output adaptive dark channel diagram of self-adapting changeable structural elements;
Transmittance calculation module is for calculating transmissivity t;
Air light value estimation module is for calculating air light value A;
Fogless scene recovery module is for restoring foggy image.
3. the defogging method based on Morphological Reconstruction and adaptive-filtering, which comprises the following steps:
Step 1: gray level image module being turned by RGB image, gradation conversion is carried out to original foggy image, gray level image is passed through
One 4 row buffer obtains the adaptive structure member growth district Ω that radius is 2k, calculate self-adapting changeable structural elements;
Step 2: maximum, minimum value is carried out to original foggy image by maximum value calculation circuit and minimum value counting circuit respectively
Filtering, obtains bright channel figure and dark channel image;Dark channel diagram is obtained with adaptive structure member greatly by the 2nd 4 row buffer
Small identical Filtering Template;
Step 3: adaptive dark estimation module passes through the self-adapting changeable structural elements that step 1 obtains, and obtains to step 2 dark
Channel figure carries out self-adapting changeable structural elements mini-value filtering, obtains adaptive dark channel diagram;
Step 4: transmittance calculation module calculates separately out dark by adaptive dark channel image and bright channel image and transmits
Rate tdcWith bright channel transmissivity tbc, then compare the larger value of the two as final optimization transmissivity t;
Step 5: air light value estimation module calculates air light value A by the dark channel diagram that step 2 obtains;
Step 6: fogless scene recovery module calculates the image after obtaining defogging by formula (1):
4. according to the method described in claim 3, it is characterized in that, the first register group and the second register group are used to realize secretly
Channel figure is matched with the delay of adaptive structure element;Three 2 row buffers are saturating for realizing dark transmissivity and bright channel
Penetrate the delay matching of rate.
5. according to the method described in claim 3, it is characterized in that, calculating the input signal of adaptive shape varistructure element
Range Ω is limited for guiding figure p, gradient weights λ and structural elementk;Wherein, guiding figure p is the gray level image of input picture, ladder
Degree weight λ takes 0.18, and structural element limits range ΩkRadius be 2;Output signal is that each pixel is corresponding certainly on image
Adapt to shape varistructure element, comprising the following steps:
Step 1): centered on current pixel point, taking side length is the sub-rectangular areas Ω of 2rk;
Step 2): zoning ΩkPixel value distance L (σ) of the interior each pixel along different paths to central point;
Step 3): pixel value distance d of the minimum pixel value distance as the point is chosenλ;
Step 4): zoning ΩkInterior structural elements length dλMean value, as structural elements length limit rlimit;
Step 5): comparison domain ΩkInterior each pixel structural elements length dλWith upper limit rlimitRelationship:It is not more than
rlimit, otherwise it is 0 that adaptive shape varistructure member corresponding position, which is 1, generate the point self-adapted shape varistructure of current pixel
Element;
When
When
Step 6): mobile current pixel point to next adjacent pixel repeats step 1)~step 5), until traversing complete width figure
Picture.
6. according to the method described in claim 5, it is characterized in that, actionradius r=2 is in guiding figure p with pixel k
The rectangle frame at center;For pixel x withBetween path;The calculating formula of length of path σ
Are as follows:
Wherein: n is the number of pixels in the σ of path, dpixel(Image(xi), Image (xi+1))=| p (xi)-p(xi+1)|
Self-adapting changeable row structural elements length is defined as:
Minimum value is on all paths of x and y;The mean value that adaptive shape varistructure member is all d apart from the upper limit;
M is the sum of all pixels in rectangular box;Adaptive shape varistructure member is defined as:
In defogging, the gray level image for inputting foggy image is used as calculating the guiding figure of adaptive structure member;Using certainly
It adapts to varistructure member and executes mini-value filtering on minimum Color Channel;Filtered image is adaptive dark channel diagram.
7. according to the method described in claim 6, it is characterized in that, calculating self-adapting changeable structural elements specifically includes following step
It is rapid:
Step (1): go out rectangle template Ω in first clock calculationkInterior 24 effective adjacent pixel distances;
Step (2): in the pixel value distance of second clock calculation, 24 pixel distance template central points;
Step (3): in the path length of 24 pixels of third clock calculation;
Step (4): in the 4th clock, r is calculated using equation (4)limit;
Step (5): adaptive shape varistructure member is obtained by accounting equation (5).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910009877.XA CN109978776A (en) | 2019-01-05 | 2019-01-05 | Defogging system and method based on Morphological Reconstruction and adaptive-filtering |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910009877.XA CN109978776A (en) | 2019-01-05 | 2019-01-05 | Defogging system and method based on Morphological Reconstruction and adaptive-filtering |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109978776A true CN109978776A (en) | 2019-07-05 |
Family
ID=67076561
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910009877.XA Pending CN109978776A (en) | 2019-01-05 | 2019-01-05 | Defogging system and method based on Morphological Reconstruction and adaptive-filtering |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109978776A (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2009013696A2 (en) * | 2007-07-24 | 2009-01-29 | Koninklijke Philips Electronics N. V. | Framework system and method for low-frequency preservation in multiresolution nonlinear adaptive filtering |
CN105913391A (en) * | 2016-04-07 | 2016-08-31 | 西安交通大学 | Defogging method based on shape variable morphological reconstruction |
CN106023091A (en) * | 2016-04-22 | 2016-10-12 | 西安电子科技大学 | Image real-time defogging method based on graphics processor |
CN106204488A (en) * | 2016-07-12 | 2016-12-07 | 湖南翰博薇微电子科技有限公司 | The video defogging method that a kind of OpenCL accelerates |
-
2019
- 2019-01-05 CN CN201910009877.XA patent/CN109978776A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2009013696A2 (en) * | 2007-07-24 | 2009-01-29 | Koninklijke Philips Electronics N. V. | Framework system and method for low-frequency preservation in multiresolution nonlinear adaptive filtering |
CN105913391A (en) * | 2016-04-07 | 2016-08-31 | 西安交通大学 | Defogging method based on shape variable morphological reconstruction |
CN106023091A (en) * | 2016-04-22 | 2016-10-12 | 西安电子科技大学 | Image real-time defogging method based on graphics processor |
CN106204488A (en) * | 2016-07-12 | 2016-12-07 | 湖南翰博薇微电子科技有限公司 | The video defogging method that a kind of OpenCL accelerates |
Non-Patent Citations (1)
Title |
---|
董辉,张斌: ""基于显著图的可变模板形态学去雾方法"", 《自动化学报》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108665496B (en) | End-to-end semantic instant positioning and mapping method based on deep learning | |
CN108416327B (en) | Target detection method and device, computer equipment and readable storage medium | |
CN105681628B (en) | A kind of convolutional network arithmetic element and restructural convolutional neural networks processor and the method for realizing image denoising processing | |
US10043241B2 (en) | Lens distortion correction using a neurosynaptic circuit | |
CN109472818A (en) | A kind of image defogging method based on deep neural network | |
CN102968772B (en) | A kind of image defogging method capable based on dark channel information | |
CN109493300B (en) | Aerial image real-time defogging method based on FPGA (field programmable Gate array) convolutional neural network and unmanned aerial vehicle | |
CN108764336A (en) | For the deep learning method and device of image recognition, client, server | |
CN108376392A (en) | A kind of image motion ambiguity removal method based on convolutional neural networks | |
CN109961404A (en) | A kind of high clear video image Enhancement Method based on GPU parallel computation | |
CN109614941B (en) | Embedded crowd density estimation method based on convolutional neural network model | |
CN110059815B (en) | Artificial intelligence reasoning computing equipment | |
WO2022027917A1 (en) | Image processing method, apparatus and system, and electronic device and readable storage medium | |
CN111986108A (en) | Complex sea-air scene image defogging method based on generation countermeasure network | |
CN110969089A (en) | Lightweight face recognition system and recognition method under noise environment | |
CN104751421A (en) | Method for achieving image defogging on FPGA | |
CN108062559A (en) | A kind of image classification method based on multiple receptive field, system and device | |
CN112861727A (en) | Real-time semantic segmentation method based on mixed depth separable convolution | |
CN108629750A (en) | A kind of night defogging method, terminal device and storage medium | |
CN112651459A (en) | Defense method, device, equipment and storage medium for confrontation sample of deep learning image | |
Shu et al. | Adversarial differentiable data augmentation for autonomous systems | |
CN108846420B (en) | Network structure and client | |
CN103516959B (en) | Image processing method and equipment | |
CN110276739A (en) | A kind of video jitter removing method based on deep learning | |
CN109978776A (en) | Defogging system and method based on Morphological Reconstruction and adaptive-filtering |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190705 |