CN109816605A - A kind of MSRCR image defogging method based on multichannel convolutive - Google Patents
A kind of MSRCR image defogging method based on multichannel convolutive Download PDFInfo
- Publication number
- CN109816605A CN109816605A CN201910041240.9A CN201910041240A CN109816605A CN 109816605 A CN109816605 A CN 109816605A CN 201910041240 A CN201910041240 A CN 201910041240A CN 109816605 A CN109816605 A CN 109816605A
- Authority
- CN
- China
- Prior art keywords
- image
- channel
- msrcr
- enhanced
- filtering
- 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
Landscapes
- Image Processing (AREA)
Abstract
The present invention provides a kind of MSRCR image defogging method based on multichannel convolutive.The method of the present invention, it include: that filtering processing is guided to source images, to treated, R, G, channel B carry out convolution with the Gaussian convolution core of 6 3*3 respectively, obtain the characteristic pattern of 6 with the same size of single input channel, corresponding 6 characteristic patterns in each channel are enhanced into laggard line Weighted Fusion by Retinex algorithm, detail pictures after the enhanced image of Retinex and secondary boot filtering processing are weighted fusion, reconstruct final mist elimination image.The present invention carries out convolution using multiple dimensioned Gaussian convolution core, extract finer feature assessment incident components, incident components are carried out with the Retinex enhancing of multiple dimensioned linear weighted function, the smoothness constraint of incident components and reflected image is considered in secondary boot filtering simultaneously, so that treated, image meets smoothness constraint, noise is reduced again, and two enhanced images are subjected to linear weighted function fusion, realize the defogging of image.
Description
Technical field
The present invention relates to technical field of image processing, specifically, more particularly to a kind of MSRCR based on multichannel convolutive
Image defogging method.
Background technique
In digital imaging arts, clear image is the critical prerequisite for understanding real scene.In outdoor environment, due to
The photo of the influence of the severe weather conditions such as illumination, mist and haze, shooting can seriously reduce visibility and contrast, this will lead to
Images dim and distortion.In order to effectively remove influence of the thick fog to picture quality and highlight the detailed information in thick fog, it is based on
The image enhancement of image procossing and image restoration based on physical model are common methods.
Defogging algorithm based on physical model is by establishing approximate atmospherical scattering model, inverting degenerative process to obtain nothing
The optimal estimation value of mist image.It is broadly divided into three classes: the method based on depth information, the method based on partial differential equation and being based on
The method of priori knowledge.What the defogging algorithm based on physical model needed Same Scene image deeply convinces breath, priori knowledge etc., and
It needs by some physical equipments, it is inconvenient in practical applications.Defogging algorithm based on image enhancement, which can be detached from, sets physics
Standby dependence, becomes current defogging algorithm main direction of studying, mainly there is histogram equalization, homomorphic filtering, bilateral filter at present
Wave, guiding filtering and Retinex algorithm.Since data of the histogram equalization to processing are indiscriminate, it may enhance back
The contrast of scape noise and the contrast for reducing useful signal；The computational complexity of homomorphic filtering and bilateral filtering is higher, algorithm
Efficiency and practicability it is not fully up to expectations；Guiding filtering as local linear image filter, have good edge keep and
Smothing filtering performance, when original image is more complex and larger noise, the phenomenon that enhancing image is likely to occur Noise enhancement；It is based on
The innovatory algorithms such as algorithm SSR, MSR and MSRCR of Retinex theory, the estimation and elimination of incident components are the key that defoggings, one
As using gaussian filtering estimate that incident components, SSR algorithm are mainly used for enhancing gray level image, but are difficult to balance the dynamic pressure of image
Contracting and color are constant；The SSR of multiple and different scales is carried out linear weighted function and carries out the enhancing of color image by MSR algorithm, but is brought
The problem of colored degeneration；MSRCR introduces color recovery factor on the basis of MSR so that enhanced image have compared with
Good color guarantee property, but the color of image can deviate original color, overexposure.
Summary of the invention
According to technical problem set forth above, a kind of MSRCR image defogging method based on multichannel convolutive is provided.This hair
It is bright mainly according to the statistical property of incident components and reflecting component in Retinex vision mode, merge guiding filtering, multiple dimensioned volume
Product, Retinex enhancing, quantization operation and details fusion can be to the effective defoggings of thick fog image.
The technological means that the present invention uses is as follows:
A kind of MSRCR (Multi-Scale Retinex with color restoration based on multichannel convolutive
Of Multi-Channel Convolutional, MC_MSRCR) image defogging method, which comprises the following steps:
Step S01: obtaining original RGB thick fog image, and pre-process to it, by original RGB thick fog picture breakdown at
R, G, channel B image；
Step S02: to the original RGB thick fog picture breakdown at R, G, channel B image guides at filtering respectively
Reason；
Step S03: above-mentioned steps S02 guiding filtering treated R, G, channel B image are used into 6 3*3 sizes not respectively
Etc. the Gaussian convolution core of parameters carry out multiple dimensioned convolution, obtain the identical characteristic pattern of corresponding 6 Zhang great little in each channel；
Step S04: by corresponding 6 characteristic patterns in each channel by Retinex enhance laggard line Weighted Fusion into
And improve enhanced effect, enhanced R, G, channel B image are obtained, then to enhanced R, G, channel B image introduction volume
Change operation, realizes color recovery, complete the enhancing process of MSRCR, obtain the enhanced R, G of MSRCR, channel B image；
Step S05: the enhanced R, G of MSRCR, channel B image are subjected to secondary boot filtering, secondary boot filters simultaneously
The smoothness constraint for considering incident components and reflected image, is not only met smoothness constraint condition while also reducing noise
R, G, channel B image；
Step S06: by R, G obtained in R, G obtained in step S04, channel B image and step S05, channel B image
Linear weighted function fusion is carried out, final defogging figure is reconstructed.
Further, guiding filtering processing method its Filtering Formula in the step S02 are as follows:
Q=guided_filter (p, I, r, ε)
Wherein, p is filtering input picture, and I is navigational figure, and r is filter window size, and ε is regularization coefficient, ε > 0, q
It is filtering output image；
Input picture p to be filtered is given, guiding filtering output image q is assumed to the linear transformation of navigational figure I:Wherein, wkIt is the local window centered on pixel k, coefficient akAnd bkBy minimizing cost
FunctionIt solves；
Wherein, ukWithIt is I respectively in window wkIn mean value and variance；r2It is window wkIn number of pixels；It is p
In window wkIn mean value；ε > 0 is to avoid the occurrence of biggish akThe regularization parameter of introducing.
Further, multiple dimensioned convolution method its formula in the step S03 is as follows:
Wherein, * indicates convolution algorithm, and λ is normalization constant, so that ∫ ∫ G (x, y) dxdy=1, σ ∈ { σ1,σ2,σ3,σ4,
σ5,σ6And 0≤σ1≤σ2< 50,50≤σ3≤σ4< 100,100≤σ5≤σ6。
Further, Retinex Enhancement Method its formula in the step S04 is as follows:
Wherein, i indicates some channel of image, and (x, y) indicates the pixel coordinate point of original image；Gfi(x, y) indicates certain
The guiding filtering function of a channel original image,N indicates Gaussian filter filter
The scale parameter of wave radius；Indicate i-th of channel corresponding incident light component under n-th of scale,ωnIt indicates the weight on n-th of scale, meets normalizing condition
It is described that quantization operation is introduced to enhanced R, G, channel B, it realizes the recovery of color, completes the enhancing of MSRCR
It is as follows to obtain enhanced MSRCR for journey:
Wherein, i indicates some channel of image；Indicate the mean value in i-th of channel；It indicates
The mean square deviation in i-th of channel；D indicates the control dynamic parameter D of image to realize the adjusting of no colour cast；MiniIndicate the channel i figure
The minimum value of picture；MaxiIndicate the maximum value of i channel image；I channel image after indicating quantization.
Further, the linear weighted function fusion process in the step S06 includes the following steps:
Step S061: R is setMSRCR、GMSRCR、BMSRCRR, G obtained in step S04, channel B image are respectively indicated,Respectively indicate R, G obtained in step S05, channel B image；
Step S062: to R, G obtained in R, G obtained in step S04, channel B image and step S05, channel B figure
Picture；Fusion is weighted using following formula:
RGB (i, j)=R (i, j)+G (i, j)+B (i, j)；
Wherein, λ is weighting coefficient, and 0≤λ≤1.
Further, the ω in the Retinex Enhancement Method in the step S04n=1/6, the present invention uses 6 kinds of differences
The Gaussian convolution verification multichannel of scale carries out convolution, extracts finer feature assessment incident components.
Further, the weighting coefficient λ can obtain preferable defog effect between 0.9~0.96.
Compared with the prior art, the invention has the following advantages that
1, bad in order to solve traditional Retinex defogging algorithm reinforcing effect, color fidelity is poor, fringe region halation
The problem of artifact.The present invention guides filtering processing to original image and remains edge letter from the point of view of incident components
Breath overcomes noise again, carries out convolution using multiple dimensioned Gaussian convolution verification multichannel, extracts finer feature assessment and enter
Component is penetrated, Retinex operation is carried out to incident components, enhances the detailed information and global contrast of image, in order to guarantee defogging
Image has preferable color fidelity to introduce quantization operation.
2, the image that the enhanced image of MSRCR algorithm and secondary boot of multichannel convolutive are filtered the present invention into
Row Weighted Fusion reconstructs clearly mist elimination image within the scope of certain proportion.
The present invention can be widely popularized in fields such as image procossings based on the above reasons.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to do simply to introduce, it should be apparent that, the accompanying drawings in the following description is this hair
Bright some embodiments for those of ordinary skill in the art without any creative labor, can be with
It obtains other drawings based on these drawings.
Fig. 1 is flow diagram of the invention.
Fig. 2 is the defog effect comparison diagram and each algorithm pair that the present invention is directed to short distance scene thick fog image with other algorithms
The grey level histogram answered.
Fig. 3 is the defog effect comparison diagram and each algorithm pair that the present invention is directed to remote scene thick fog image with other algorithms
The grey level histogram answered.
Fig. 4 is that the present invention is corresponding for the defog effect comparison diagram of underwater scene thick fog image and each algorithm with other algorithms
Grey level histogram.
Specific embodiment
In order to enable those skilled in the art to better understand the solution of the present invention, below in conjunction in the embodiment of the present invention
Attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is only
The embodiment of a part of the invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill people
The model that the present invention protects all should belong in member's every other embodiment obtained without making creative work
It encloses.
It should be noted that description and claims of this specification and term " first " in above-mentioned attached drawing, "
Two " etc. be to be used to distinguish similar objects, without being used to describe a particular order or precedence order.It should be understood that using in this way
Data be interchangeable under appropriate circumstances, so as to the embodiment of the present invention described herein can in addition to illustrating herein or
Sequence other than those of description is implemented.In addition, term " includes " and " having " and their any deformation, it is intended that cover
Cover it is non-exclusive include, for example, the process, method, system, product or equipment for containing a series of steps or units are not necessarily limited to
Step or unit those of is clearly listed, but may include be not clearly listed or for these process, methods, product
Or other step or units that equipment is intrinsic.
In order to verify the validity of defogging of the present invention, the thick fog image of different scenes is chosen as test set, while with
The experimental result of SSR, MSR, bilateral filtering MSRCR (MSRCR of bilateral filtering, B_MSRCR) and He algorithm
It is compared and analyzed in terms of qualitative and quantify two.Specific steps and principle are as follows:
As shown in Figure 1, the present invention provides a kind of MSRCR image defogging method based on multichannel convolutive, including it is following
Step:
Step S01: obtaining original RGB thick fog image, and pre-process to it, by original RGB thick fog picture breakdown at
R, G, channel B image；
Step S02: to the original RGB thick fog picture breakdown at R, G, channel B image guides at filtering respectively
Reason；Its Filtering Formula are as follows:
Q=guided_filter (p, I, r, ε)
Wherein, p is filtering input picture, and I is navigational figure, and r is filter window size, and ε is regularization coefficient, ε > 0, q
It is filtering output image；
Input picture p to be filtered is given, guiding filtering output image q is assumed to the linear transformation of navigational figure I:Wherein, wkIt is the local window centered on pixel k, coefficient akAnd bkBy minimizing cost
FunctionIt solves；
Wherein, ukWithIt is I respectively in window wkIn mean value and variance；r2It is window wkIn number of pixels；It is p
In window wkIn mean value；ε > 0 is to avoid the occurrence of biggish akThe regularization parameter of introducing.
Step S03: above-mentioned steps S02 guiding filtering treated R, G, channel B image are used into 6 3*3 sizes not respectively
Etc. the Gaussian convolution core of parameters carry out multiple dimensioned convolution, obtain the identical characteristic pattern of corresponding 6 Zhang great little in each channel；It is multiple dimensioned
Its formula of convolution method is as follows:
Wherein, * indicates convolution algorithm, and λ is normalization constant, so that ∫ ∫ G (x, y) dxdy=1, σ ∈ { σ1,σ2,σ3,σ4,
σ5,σ6And 0≤σ1≤σ2< 50,50≤σ3≤σ4< 100,100≤σ5≤σ6。
Step S04: by corresponding 6 characteristic patterns in each channel by Retinex enhance laggard line Weighted Fusion into
And improve enhanced effect, enhanced R, G, channel B image are obtained, its formula of Retinex Enhancement Method is as follows:
Wherein, i indicates some channel of image, and (x, y) indicates the pixel coordinate point of original image；Gfi(x, y) indicates certain
The guiding filtering function of a channel original image,N indicates Gaussian filter filtering
The scale parameter of radius；Indicate i-th of channel corresponding incident light component under n-th of scale,ωnIt indicates the weight on n-th of scale, meets normalizing condition
For the color distortion in MSR algorithm, from directly starting with from the mode of quantization, during adjusting color error ratio
The concept of mean value and mean square deviation is introduced, while introducing the control dynamic parameter D of image to realize the adjusting of no colour cast,
Not only color fidelity is improved, but also preferably adapts to various scene images, quantization behaviour is introduced to enhanced R, G, channel B image
Make, complete the enhancing process of MSRCR, it is as follows to obtain enhanced MSRCR:
Wherein, i indicates some channel of image；Indicate the mean value in i-th of channel；It indicates
The mean square deviation in i-th of channel；D indicates the control dynamic parameter D of image to realize the adjusting of no colour cast；MiniIndicate the channel i figure
The minimum value of picture；MaxiIndicate the maximum value of i channel image；I channel image after indicating quantization.
Step S05: the enhanced R, G of MSRCR, channel B image are subjected to secondary boot filtering, secondary boot filters simultaneously
The smoothness constraint for considering incident components and reflected image, is not only met smoothness constraint condition while also reducing noise
R, G, channel B image；The secondary boot in each channel filters i.e.:
Step S06: by R, G obtained in R, G obtained in step S04, channel B image and step S05, channel B image
Linear weighted function fusion is carried out, final defogging figure is reconstructed.
Step S061: R is setMSRCR、GMSRCR、BMSRCRR, G obtained in step S04, channel B image are respectively indicated,Respectively indicate R, G obtained in step S05, channel B image；
Step S062: to R, G obtained in R, G obtained in step S04, channel B image and step S05, channel B figure
Picture；Fusion is weighted using following formula:
RGB (i, j)=R (i, j)+G (i, j)+B (i, j)；
Wherein, λ is weighting coefficient, and 0≤λ≤1.
As the present embodiment preferred embodiment, the weighting coefficient λ can be obtained preferable between 0.9~0.96
Defog effect.
The ω in Retinex Enhancement Method as the present embodiment preferred embodiment, in the step S04n=1/
6, the present invention carries out convolution using the Gaussian convolution verification multichannel of 6 kinds of different scales, and it is incident to extract finer feature assessment
Component.
Embodiment
As shown in Fig. 2, the present invention provides the defog effect comparisons with other algorithms for short distance scene thick fog image
Scheme grey level histogram corresponding with each algorithm, can be seen that 5 kinds of defogging algorithms all improve to a certain extent from experiment effect figure
There is the phenomenon that color distortion and halo artifact in image overall contrast, SSR and MSR；He, B_MSRCR and the present invention exist
Effectively inhibit artifact while enhancing contrast, and there is dark primary and crosses enhancing in He.In terms of local detail, the present invention is effective
Noise reduction, enhance local detail.In terms of grey level histogram, mist elimination image grey value profile that the present invention is handled uniformly and
Height, reflection on the image, that is, enhance dark areas, and improve global contrast and detailed information.Therefore the present invention have compared with
Good defog effect improves global contrast, realizes details enhancing and color fidelity.
As shown in figure 3, the present invention provides the defog effect comparisons with other algorithms for remote scene thick fog image
Scheme grey level histogram corresponding with each algorithm, can be seen that 5 kinds of defogging algorithms all improve to a certain extent from experiment effect figure
There is the phenomenon that halo artifact in image overall contrast, SSR and MSR；He occurs that dark crosses enhancing, halo artifact shows
As；B_MSRCR and the present invention effectively inhibit artifact while enhancing contrast.In terms of local detail, this paper algorithm is effective
Noise reduction, improve local detail contrast, realize details enhancing.In terms of grey level histogram, this paper algorithm process is gone
Mist gray value of image is evenly distributed and height, and reflection on the image, that is, enhances dark areas, and improve global contrast and details
Information.Therefore this paper algorithm has preferable defog effect, improves global contrast, realizes details enhancing and color fidelity
Property.
As shown in figure 4, the present invention also provides the defog effect comparisons with other algorithms for underwater scene thick fog image
Scheme grey level histogram corresponding with each algorithm, can be seen that 5 kinds of defogging algorithms all improve to a certain extent from experiment effect figure
There is the phenomenon that halo artifact in image overall contrast, SSR and MSR；He occurs that dark crosses enhancing, halo artifact shows
As；B_MSRCR and the present invention effectively inhibit artifact while enhancing contrast.In terms of local detail, the present invention is effective
Noise reduction improves local detail contrast, realizes details enhancing.In terms of grey level histogram, defogging figure that the present invention is handled
Picture grey value profile is uniformly and high, and reflection on the image, that is, enhances dark areas, and improves global contrast and details letter
Breath.Therefore the present invention has preferable defog effect, improves global contrast, realizes details enhancing and color fidelity.
The present embodiment carries out the experimental result of algorithms of different from three kinds of average gradient, comentropy and clarity objective indicators
Comparison；From the data of table 1, table 2 and table 3 it is found that the average gradient of He, B_MSRCR and MC_MSRCR algorithm, comentropy and clear
Degree is all larger than original image；Although the average gradient and clarity of SSR and MSR algorithm are all larger than original image, part figure
As comentropy is slightly below original image, although this shows to improve image definition, image information has lost；Due to
SSR and MSR algorithm has ignored the reparation of color while improving clarity, and comentropy is determined by the abundant degree of color
It is fixed.In order to improve the color fidelity of enhancing image, MSRCR introduces color recovery on the basis of MSR, herein using amount
Change operation and realize the recovery of color, and increases spilling judgement.Bilateral filtering has preferable edge retention, but the time is complicated
It spends higher and insensitive to detailed information.Guiding filtering realizes details enhancing while guaranteeing marginal information.Therefore originally
Invention has biggish promotion for the average gradient of original image, comentropy and clarity, and is better than other enhancing algorithms.
1 inventive algorithm of table and the average gradient of other algorithm process results compare
The comentropy of 2 inventive algorithm of table and other algorithm process results compares
The clarity of 3 inventive algorithm of table and other algorithm process results compares
Finally, it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although
Present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: it still may be used
To modify to technical solution documented by previous embodiment, or some or all of the technical features are equal
Replacement；And these are modified or replaceed, the model for technical solution of the embodiment of the present invention that it does not separate the essence of the corresponding technical solution
It encloses.
Claims (7)
1. a kind of MSRCR image defogging method based on multichannel convolutive, which comprises the following steps:
Step S01: obtaining original RGB thick fog image, and pre-process to it, by original RGB thick fog picture breakdown at R, G, B
Channel image；
Step S02: to the original RGB thick fog picture breakdown at R, G, channel B image guide filtering processing respectively；
Step S03: 6 3*3 etc. are used to differ in size respectively above-mentioned steps S02 guiding filtering treated R, G, channel B image
The Gaussian convolution core of parameter carries out convolution, obtains the identical characteristic pattern of corresponding 6 Zhang great little in each channel；
Step S04: corresponding 6 characteristic patterns in each channel are enhanced into laggard line Weighted Fusion by Retinex and are changed
It is apt to enhanced effect, obtains enhanced R, G, channel B image, then quantization behaviour is introduced to enhanced R, G, channel B image
Make, realizes the recovery of color, complete the enhancing process of MSRCR, obtain the enhanced R, G of MSRCR, channel B image；
Step S05: the enhanced R, G of MSRCR, channel B image are subjected to secondary boot filtering, secondary boot filters while considering
The smoothness constraint of incident components and reflected image, not only met smoothness constraint condition and meanwhile also reduce noise R, G,
Channel B image；
Step S06: R, G obtained in R, G obtained in step S04, channel B image and step S05, channel B image are carried out
Linear weighted function fusion, reconstructs final defogging figure.
2. the MSRCR image defogging method according to claim 1 based on multichannel convolutive, which is characterized in that the step
Guiding filtering processing method its Filtering Formula in rapid S02 are as follows:
Q=guided_filter (p, I, r, ε)
Wherein, p is filtering input picture, and I is navigational figure, and r is filter window size, and ε is regularization coefficient, and ε > 0, q are filters
Wave exports image；
Input picture p to be filtered is given, guiding filtering output image q is assumed to the linear transformation of navigational figure I: qi=akIi+bk,Wherein, wkIt is the local window centered on pixel k, coefficient akAnd bkBy minimizing cost functionIt solves；
Wherein, ukWithIt is I respectively in window wkIn mean value and variance；r2It is window wkIn number of pixels；It is p in window
Mouth wkIn mean value；ε > 0 is to avoid the occurrence of biggish akThe regularization parameter of introducing.
3. the MSRCR image defogging method according to claim 1 based on multichannel convolutive, which is characterized in that the step
Multiple dimensioned convolution method its formula in rapid S03 is as follows:
Wherein, * indicates convolution algorithm, and λ is normalization constant, so that ∫ ∫ G (x, y) dxdy=1, σ ∈ { σ1,σ2,σ3,σ4,σ5,σ6}
And 0≤σ1≤σ2< 50,50≤σ3≤σ4< 100,100≤σ5≤σ6。
4. the MSRCR image defogging method according to claim 1 based on multichannel convolutive, which is characterized in that the step
Retinex Enhancement Method its formula in rapid S04 is as follows:
Wherein, i indicates some channel of image, and (x, y) indicates the pixel coordinate point of original image；Gfi(x, y) indicates that some is logical
The guiding filtering function of road original image, Gfi(x, y)=akIi(i,j)+bk,N indicates Gaussian filter filtering half
The scale parameter of diameter；Indicate i-th of channel corresponding incident light component under n-th of scale,ωnIt indicates the weight on n-th of scale, meets normalizing condition
It is described that quantization operation is introduced to enhanced R, G, channel B image, it realizes the recovery of color, completes the enhancing of MSRCR
It is as follows to obtain enhanced MSRCR for journey:
Wherein, i indicates some channel of image；Indicate the mean value in i-th of channel；Indicate i-th
The mean square deviation in a channel；D indicates the control dynamic parameter D of image to realize the adjusting of no colour cast；MiniIndicate i channel image
Minimum value；MaxiIndicate the maximum value of i channel image；I channel image after indicating quantization.
5. the MSRCR image defogging method according to claim 1 based on multichannel convolutive, which is characterized in that the step
Linear weighted function fusion process in rapid S06 includes the following steps:
Step S061: R is setMSRCR、GMSRCR、BMSRCRR, G obtained in step S04, channel B image are respectively indicated,Respectively indicate R, G obtained in step S05, channel B image；
Step S062: to R, G obtained in R, G obtained in step S04, channel B image and step S05, channel B image；It adopts
Fusion is weighted with following formula:
RGB (i, j)=R (i, j)+G (i, j)+B (i, j)；
Wherein, λ is weighting coefficient, and 0≤λ≤1.
6. the MSRCR image defogging method according to claim 1 based on multichannel convolutive, which is characterized in that the step
The ω in Retinex Enhancement Method in rapid S04n=1/6.
7. the MSRCR image defogging method according to claim 1 based on multichannel convolutive, which is characterized in that described to add
Weight coefficient λ is between 0.9~0.96.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910041240.9A CN109816605B (en) | 2019-01-16 | 2019-01-16 | MSRCR image defogging method based on multi-channel convolution |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910041240.9A CN109816605B (en) | 2019-01-16 | 2019-01-16 | MSRCR image defogging method based on multi-channel convolution |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109816605A true CN109816605A (en) | 2019-05-28 |
| CN109816605B CN109816605B (en) | 2022-10-04 |
Family
ID=66604464
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910041240.9A Active CN109816605B (en) | 2019-01-16 | 2019-01-16 | MSRCR image defogging method based on multi-channel convolution |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109816605B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110175964A (en) * | 2019-05-30 | 2019-08-27 | 大连海事大学 | A kind of Retinex image enchancing method based on laplacian pyramid |
| CN110503617A (en) * | 2019-08-29 | 2019-11-26 | 大连海事大学 | A kind of underwater picture Enhancement Method based on the fusion of high and low frequency information |
| CN111260589A (en) * | 2020-02-20 | 2020-06-09 | 国网陕西省电力公司电力科学研究院 | Retinex-based power transmission line monitoring image defogging method |
| CN111351449A (en) * | 2020-02-14 | 2020-06-30 | 广东工业大学 | Stereo matching method based on cost aggregation |
| CN112541869A (en) * | 2020-12-07 | 2021-03-23 | 南京工程学院 | Retinex image defogging method based on matlab |
| CN112862721A (en) * | 2021-02-24 | 2021-05-28 | 中国矿业大学(北京) | Underground pipeline image defogging method based on dark channel and Retinex |
| CN112950488A (en) * | 2020-12-31 | 2021-06-11 | 电子科技大学成都学院 | MSRCR image defogging algorithm based on multi-scale detail optimization |
| CN113284059A (en) * | 2021-04-29 | 2021-08-20 | Oppo广东移动通信有限公司 | Model training method, image enhancement method, device, electronic device and medium |
| CN113781416A (en) * | 2021-08-30 | 2021-12-10 | 武汉理工大学 | Conveyer belt tearing detection method and device and electronic equipment |
| CN117173070A (en) * | 2023-11-02 | 2023-12-05 | 南昌理工学院 | Image processing fusion method and system based on FPGA |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070083114A1 (en) * | 2005-08-26 | 2007-04-12 | The University Of Connecticut | Systems and methods for image resolution enhancement |
| US20080101719A1 (en) * | 2006-10-30 | 2008-05-01 | Samsung Electronics Co., Ltd. | Image enhancement method and system |
| CN105701783A (en) * | 2016-01-14 | 2016-06-22 | 浙江大学 | Single image defogging method based on ambient light model and apparatus thereof |
| CN106846274A (en) * | 2017-01-23 | 2017-06-13 | 重庆师范大学 | Haze method is gone with reference to the MSRCR traffic images of HE and guiding filtering |
-
2019
- 2019-01-16 CN CN201910041240.9A patent/CN109816605B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070083114A1 (en) * | 2005-08-26 | 2007-04-12 | The University Of Connecticut | Systems and methods for image resolution enhancement |
| US20080101719A1 (en) * | 2006-10-30 | 2008-05-01 | Samsung Electronics Co., Ltd. | Image enhancement method and system |
| CN105701783A (en) * | 2016-01-14 | 2016-06-22 | 浙江大学 | Single image defogging method based on ambient light model and apparatus thereof |
| CN106846274A (en) * | 2017-01-23 | 2017-06-13 | 重庆师范大学 | Haze method is gone with reference to the MSRCR traffic images of HE and guiding filtering |
Non-Patent Citations (1)
| Title |
|---|
| 黄允浒等: "多尺度Retinex算法与其它图像增强方法比较", 《新疆师范大学学报(自然科学版)》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110175964A (en) * | 2019-05-30 | 2019-08-27 | 大连海事大学 | A kind of Retinex image enchancing method based on laplacian pyramid |
| CN110503617A (en) * | 2019-08-29 | 2019-11-26 | 大连海事大学 | A kind of underwater picture Enhancement Method based on the fusion of high and low frequency information |
| CN110503617B (en) * | 2019-08-29 | 2022-09-30 | 大连海事大学 | Underwater image enhancement method based on high-frequency and low-frequency information fusion |
| CN111351449A (en) * | 2020-02-14 | 2020-06-30 | 广东工业大学 | Stereo matching method based on cost aggregation |
| CN111260589A (en) * | 2020-02-20 | 2020-06-09 | 国网陕西省电力公司电力科学研究院 | Retinex-based power transmission line monitoring image defogging method |
| CN112541869A (en) * | 2020-12-07 | 2021-03-23 | 南京工程学院 | Retinex image defogging method based on matlab |
| CN112950488A (en) * | 2020-12-31 | 2021-06-11 | 电子科技大学成都学院 | MSRCR image defogging algorithm based on multi-scale detail optimization |
| CN112862721A (en) * | 2021-02-24 | 2021-05-28 | 中国矿业大学(北京) | Underground pipeline image defogging method based on dark channel and Retinex |
| CN112862721B (en) * | 2021-02-24 | 2022-01-07 | 中国矿业大学(北京) | Underground pipeline image defogging method based on dark channel and Retinex |
| CN113284059A (en) * | 2021-04-29 | 2021-08-20 | Oppo广东移动通信有限公司 | Model training method, image enhancement method, device, electronic device and medium |
| CN113781416A (en) * | 2021-08-30 | 2021-12-10 | 武汉理工大学 | Conveyer belt tearing detection method and device and electronic equipment |
| CN117173070A (en) * | 2023-11-02 | 2023-12-05 | 南昌理工学院 | Image processing fusion method and system based on FPGA |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109816605B (en) | 2022-10-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109816605A (en) | A kind of MSRCR image defogging method based on multichannel convolutive | |
| CN106846263B (en) | Based on the image defogging method for merging channel and sky being immunized | |
| CN110689587B (en) | Underwater image enhancement method based on color correction and detail enhancement | |
| Gao et al. | Sand-dust image restoration based on reversing the blue channel prior | |
| CN110363727B (en) | Image defogging method based on multi-scale dark channel prior cascade deep neural network | |
| CN110782407B (en) | Single image defogging method based on sky region probability segmentation | |
| Pei et al. | Effective image haze removal using dark channel prior and post-processing | |
| CN105447825B (en) | Image defogging method and its system | |
| CN110675340A (en) | Single image defogging method and medium based on improved non-local prior | |
| CN110689504B (en) | Underwater image restoration method based on secondary guide transmission diagram | |
| Soni et al. | An improved image dehazing technique using CLAHE and guided filter | |
| CN112488948A (en) | Underwater image restoration method based on black pixel point estimation backscattering | |
| CN114219732A (en) | Image defogging method and system based on sky region segmentation and transmissivity refinement | |
| CN112419163B (en) | Single image weak supervision defogging method based on priori knowledge and deep learning | |
| Wu et al. | Image haze removal: Status, challenges and prospects | |
| CN111311502A (en) | Method for processing foggy day image by using bidirectional weighted fusion | |
| Wen et al. | A survey of image dehazing algorithm based on retinex theory | |
| Mathur et al. | Dual domain approach for colour enhancement of underwater images | |
| CN113298730B (en) | Defogging restoration method based on image decomposition | |
| Xie et al. | DHD-Net: A novel deep-learning-based dehazing network | |
| Jayanthi et al. | Underwater haze removal using contrast boosted grayscale image | |
| Li et al. | Haze density estimation and dark channel prior based image defogging | |
| Guo et al. | Single image haze removal using dark channel prior and adaptive transmission rate | |
| CN109961413A (en) | The image defogging iterative algorithm of atmosphere light direction optimal estimating | |
| CN112419166B (en) | Image defogging method based on combination of local region segmentation and SCN |
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 |