CN106447617A - Improved Retinex image defogging method - Google Patents
Improved Retinex image defogging method Download PDFInfo
- Publication number
- CN106447617A CN106447617A CN201610173962.6A CN201610173962A CN106447617A CN 106447617 A CN106447617 A CN 106447617A CN 201610173962 A CN201610173962 A CN 201610173962A CN 106447617 A CN106447617 A CN 106447617A
- Authority
- CN
- China
- Prior art keywords
- result
- image
- retinex
- improved
- log
- 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
- 238000000034 method Methods 0.000 title claims abstract description 21
- 238000003384 imaging method Methods 0.000 claims abstract description 14
- 238000005286 illumination Methods 0.000 claims abstract description 10
- 238000001914 filtration Methods 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 9
- 238000010606 normalization Methods 0.000 claims description 5
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 235000013399 edible fruits Nutrition 0.000 claims description 2
- 230000000694 effects Effects 0.000 abstract description 7
- 238000004501 airglow Methods 0.000 abstract 1
- 230000008030 elimination Effects 0.000 description 13
- 238000003379 elimination reaction Methods 0.000 description 13
- 239000003595 mist Substances 0.000 description 13
- 230000008859 change Effects 0.000 description 4
- 230000008901 benefit Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000005855 radiation Effects 0.000 description 2
- 238000002310 reflectometry Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 238000010521 absorption reaction Methods 0.000 description 1
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 230000002238 attenuated effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000005352 clarification Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 230000000149 penetrating effect Effects 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
Classifications
-
- G06T5/73—
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- 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
- G06T2207/20172—Image enhancement details
-
- 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
- G06T2207/20212—Image combination
- G06T2207/20224—Image subtraction
Abstract
The invention discloses an improved Retinex image defogging method. The method comprises the following steps of: inputting a haze image, estimating n(x,y) firstly and removing n(x,y) according to a new expression formula, wherein n(x,y) is estimated by use of gaussian filter because n(x,y) is caused by direct imaging of airglow, transforms slowly and belongs to a low frequency part; subtracting the estimated n(x,y) from an original drawing, normalizing the obtained values, and acquiring an intermediate result; carrying out gaussian filter on the intermediate result according to a single-scale Retinex algorithm, estimating an illumination component, and removing the estimated illumination component L(x,y) from the intermediate result in a Log domain; and carrying out automatic color gradation processing on the obtained result, eliminating an overexposure phenomenon in the result, and outputting the processed result. According to the improved Retinex image defogging method, the defogging effect is greatly improved, the distant scenery becomes clear, detail parts become obvious, the running speed is rapid and the real-time requirement can be satisfied.
Description
Technical field
The present invention relates to a kind of image clarification method in foggy day, particularly to a kind of improved Retinex image mist elimination side
Method.
Background technology
In the case of the greasy weather, imaging device is by the scattering of air suspended particles and absorption, and atmosphere light participates in imaging
Impact, become picture contrast declines, and visibility reduces, and details is smudgy, and picture quality significantly declines, thus significantly
Have impact on the normal work of outdoor vision system, therefore the mist elimination of image is processed is the task of top priority.
At present, the single image mist elimination algorithm of main flow mainly has two big classifications, and a kind of is that mist elimination based on image enhaucament is calculated
Method, this algorithm does not consider the image-forming principle of image, only from the angle of image procossing, improves the contrast of image, prominent
Image detail, thus improve the visual effect of final image, and meet the demands of production and living.Another kind is multiple based on image
Former mist elimination algorithm, this algorithm, based on atmospherical scattering model, is asked restored by some auxiliary information or priori law
Fogless image.
Atmospherical scattering model refers to, the total radiation receiving during imaging device imaging, is the spoke after being decayed by incident light
The summation of the amount of radiation of the amount of penetrating and atmosphere light imaging system, i.e.：
I (x)=I∞ρe-βd+I∞(1-e-βd) (1)
Wherein I∞Representing ambient light, ρ is the reflectivity of scene, and β represents the scattering coefficient of air, and d represents scene depth.Formula
Middle I∞ρ represents the radiancy of scene, the i.e. fogless situation of picture jointly.e-βdIt is the transmissivity of scene, which characterizes incident light and exist
Reach imaging device through air and be not attenuated the ratio shared by part, it can be seen that as the change of scene depth d is big, e-βd's
It is worth less, say, that decay also more.I∞ρe-βdThen jointly represent incident light attenuation term, I∞(1-e-βd) then illustrate ring
Border light participates in the impact of imaging.
In numerous mist elimination algorithms, Retinex algorithm belongs to the mist elimination algorithm of image enhaucament class.And Retinex theory be by
One color how regulating the object perceiving with regard to human visual system of Land et al. proposition and the model of brightness.Its number
Learn expression formula as follows：
I (x, y)=L (x, y) R (x, y) (2)
Wherein (x, y) is the image that people is finally perceived to I, and (x, y) is ambient light illumination to L, and (x y) is reflecting component to R.Its
(x, y) determines the dynamic range that image can reach to middle ambient light illumination L, can change with the change of environment.And reflect and divide
(x y) is then the inwardness of object to amount R.And the basic thought of Retinex theory is through, and by ambient light illumination L, (x, y) from I
(x, y) middle removing, thus restore the picture rich in detail not affected by illumination.
Further we can obtain single scale Retinex algorithm, and its mathematic(al) representation is as follows：
R (x, y)=log (I (x, y))-log (F (x, y) * I (x, y)) (3)
WhereinBelonging to gaussian filtering, c is the yardstick of Gaussian function, decides fuzzy degree,
And k represents normalization factor.
By the Retinex color constancy that had of theory so that Retinex algorithm not enough to scene illumination or
Image enhaucament in the case of uneven illumination achieves prominent effect.But work as and traditional Retinex algorithm is directly applied
When in image mist elimination, although still maintain the characteristic of its color constancy, but mist elimination result is not but very good.
Theoretical according to atmospheric scattering and Retinex theoretical, we are it is estimated that knot after Retinex algorithm is processed
Really,
The luminance component estimated by Retinex algorithm belongs to the property taken advantage of component, therefore we can be approximated and estimate here
Calculating is transmissivity e-βd, therefore we just can obtain above formula.From above formula, it will be seen that through single scale Retinex
In result after calculation process, except I∞Also have outside ρImpact, so with the intensification of the depth of field, image is subject to
Also bigger to the impact of right formula Section 2, this is the reason that namely image depth higher depth cannot remove fog.
Content of the invention
The present invention is mainly on the basis of single scale Retinex algorithm, theoretical in conjunction with atmospheric scattering, proposes a kind of improved
Based on the single image mist elimination algorithm of Retinex, promote it and remove fog effect, thus reach the purpose that can use in the greasy weather.
Here, the Retinex in the case of the present invention proposes a new greasy weather is theoretical.In the case of the greasy weather, Wo Mensuo
The color experienced and brightness, be not only to be determined by the reflectivity of ambient light illumination and object itself, and it also receives atmosphere light
Directly participate in imaging impact.So new expression formula is：
I (x, y)=R (x, y) L (x, y)+n (x, y) (5)
Wherein (x is y) that atmosphere light directly participates in imaging moiety to n.
A kind of improved Retinex image mist elimination algorithm, the theoretical basis of the Retinex in the case of based on the new greasy weather
On, comprise the following steps that：
Step 1：Input haze image, according to formula I, ((x, y) (x, y) (x, y), first we should estimate+n L for x, y)=R
(x, y) and be removed, here due to n, (x, y) belongs to atmosphere light direct imaging and causes, therefore its conversion is mild, belongs to low to go out n
Frequently part, here we use gaussian filtering carry out estimation n (x, y).
Step 2：By estimate n (x, y) deducts from artwork, and will obtain value normalization, obtain intermediate object program.
Step 3：According to single scale Retinex algorithm R, ((((x, y) * I (x, y)), right for F for I (x, y))-log for x, y)=log
Intermediate object program carries out gaussian filtering, estimates luminance component, and in Log territory, by luminance component L of estimation, (x, y) from centre
Result is removed.
Step 4：The result obtaining is carried out Auto Laves process, eliminates the overexposure phenomenon in result, and by after process
Result output.
Further, in step 1, and the size of the Gaussian filter function in step 3, unified value c=80 can obtain
Preferably remove fog effect.
Compared with prior art, the invention have the advantages that and technique effect：
Due to the fact that it is to improve on the basis of Retinex algorithm, so still retaining the look of Retinex algorithm
Color shape constancy, has preferable color fidelity.And after improvement, the fog effect that goes of image is substantially improved, scenery at a distance
Being apparent from, detail section becomes obvious.Further, this invention only has this variable element of size of Gaussian function, arranges letter
Single, it is 80 by its default setting, i.e. can tackle most situation.And this algorithm speed of service is fast, disclosure satisfy that in real time
Requirement.
Brief description
Fig. 1 is the flow chart of improved Retinex image mist elimination algorithm in example.
Fig. 2 is the original haze image of the embodiment of the present invention.
Fig. 3 is tried to achieve n (x, y) part by embodiment of the present invention gaussian filtering.
Fig. 4 is embodiment of the present invention intermediate object program figure.
Fig. 5 is embodiment of the present invention final result figure (not carrying out Auto Laves process).
Fig. 6 is embodiment of the present invention final result figure (carrying out Auto Laves process).
Detailed description of the invention
Below with reference to brief description, the present invention is described further, but the enforcement of the present invention and protection are not limited to
This, if it is noted that in place of having special detailed description in detail below, being all that those skilled in the art can refer to prior art realization
Or understand.
As it is shown in figure 1, this improved Retinex image defogging method step is as follows：
Step 1：Input haze image, according to formula I, ((x, y) (x, y) (x, y), first we should estimate+n L for x, y)=R
(x, y) and be removed, here due to n, (x, y) belongs to atmosphere light direct imaging and causes, therefore its conversion is mild, belongs to low to go out n
Frequently part, here we use gaussian filtering carry out estimation n (x, y).
Step 2：By estimate n (x, y) deducts from artwork, and will obtain value normalization, obtain intermediate object program.
Step 3：According to single scale Retinex algorithm R, ((((x, y) * I (x, y)), right for F for I (x, y))-log for x, y)=log
Intermediate object program carries out gaussian filtering, estimates luminance component, and in Log territory, by luminance component L of estimation, (x, y) from centre
Result is removed.
Step 4：The result obtaining is carried out Auto Laves process, eliminates the overexposure phenomenon in result, and by after process
Result output.
As a example by Fig. 2, carry out the detailed description of experimental procedure, and provide mist elimination result.
1) reading in haze image to be processed, such as Fig. 2, its size is 525x600x3；
2) first, we are by artwork I, (x, y) carries out gaussian filtering, and we can obtain additive component n, and (x y), ties
Fruit is as it is shown on figure 3, for the ease of viewing, to n, (x y) has carried out normalized for we.
3) by the n that obtains, (x, y) deducts from artwork, and is normalized the result obtaining, and just can obtain
Intermediate object program, as shown in Figure 4.It will be seen that this image detail ratio is more visible, but image integral color is partially dark.
4) finally we are processed according to formula (3), can obtain the result of Fig. 5, it can be seen that the result obtaining
In have the phenomenon of slight overexposure.
5) carry out Auto Laves process to Fig. 5, eliminate the overexposure phenomenon in result, and obtain final result, such as Fig. 6 institute
Showing, image has preferable color fidelity, and after improvement, the fog effect that goes of image is substantially improved, and scenery becomes at a distance
Obtaining clearly, detail section becomes obvious.
Although the embodiment that disclosed herein is as above, but described content is only to facilitate understand the present invention and adopt
Embodiment, be not limited to the present invention.Technical staff in any the technical field of the invention, without departing from this
On the premise of inventing disclosed spirit and scope, any modification and change can be made in the formal and details implemented,
But broadly fall into protection scope of the present invention.
Claims (2)
1. an improved Retinex image defogging method, it is characterised in that comprise the steps：
Step 1：Input haze image, according to formula I (x, y)=R (and x, y) L (x, y)+n (x, y), first should estimate n (x, y)
And be removed, due to n, (x, y) belongs to atmosphere light direct imaging and causes, and conversion is mild, belongs to low frequency part, uses Gauss to filter
Ripple carry out estimation n (x, y)；Wherein (x, y) is the image that people is finally perceived to I, and (x, y) is ambient light illumination to L, and (x y) is R
Reflecting component, (x is y) that atmosphere light directly participates in imaging moiety to n；
Step 2：By estimate n (x, y) deducts from artwork, and will obtain value normalization, obtain intermediate object program；
Step 3：According to single scale Retinex algorithm R, ((((x, y) * I (x, y)), to centre for F for I (x, y))-log for x, y)=log
Result carries out gaussian filtering, estimates luminance component, and in Log territory, by luminance component L of estimation, (x, y) from intermediate object program
Middle removal；Belonging to gaussian filtering, c is the yardstick of Gaussian function, decides fuzzy degree, and k table
Show normalization factor；
Step 4：The result obtaining is carried out Auto Laves process, eliminates the overexposure phenomenon in result, and the knot after processing
Fruit output.
2. one according to claim 1 improved Retinex image defogging method, it is characterised in that step 1 and step 3
In the size of Gaussian filter function, unified value c=80.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610173962.6A CN106447617A (en) | 2016-03-24 | 2016-03-24 | Improved Retinex image defogging method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610173962.6A CN106447617A (en) | 2016-03-24 | 2016-03-24 | Improved Retinex image defogging method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106447617A true CN106447617A (en) | 2017-02-22 |
Family
ID=58183738
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610173962.6A Pending CN106447617A (en) | 2016-03-24 | 2016-03-24 | Improved Retinex image defogging method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106447617A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106846280A (en) * | 2017-03-08 | 2017-06-13 | 沈阳工业大学 | Image defogging method based on discrete warp wavelet |
| CN108364261A (en) * | 2017-12-13 | 2018-08-03 | 湖北工业大学 | A kind of TV-Retinex single-frame images defogging methods of gradient guiding |
| CN108428217A (en) * | 2018-01-17 | 2018-08-21 | 南京理工大学 | A kind of image defogging method based on frequency-domain visual perception estimation |
| CN112102182A (en) * | 2020-08-31 | 2020-12-18 | 华南理工大学 | Single image reflection removing method based on deep learning |
| CN112965083A (en) * | 2021-02-02 | 2021-06-15 | 英飞拓(杭州)信息系统技术有限公司 | Multi-sensing information fusion system of intelligent automobile |
| CN113012067A (en) * | 2021-03-16 | 2021-06-22 | 华南理工大学 | Retinex theory and end-to-end depth network-based underwater image restoration method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040091164A1 (en) * | 2002-11-11 | 2004-05-13 | Minolta Co., Ltd. | Image processing program product and device for executing retinex processing |
| JP2005026804A (en) * | 2003-06-30 | 2005-01-27 | Matsushita Electric Ind Co Ltd | Device and method of correcting image luminance |
| CN103606134A (en) * | 2013-11-26 | 2014-02-26 | 国网上海市电力公司 | Enhancing method of low-light video images |
| CN103914813A (en) * | 2014-04-10 | 2014-07-09 | 西安电子科技大学 | Colorful haze image defogging and illumination compensation restoration method |
| CN104200437A (en) * | 2014-09-04 | 2014-12-10 | 北京工业大学 | Image defogging method |
-
2016
- 2016-03-24 CN CN201610173962.6A patent/CN106447617A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040091164A1 (en) * | 2002-11-11 | 2004-05-13 | Minolta Co., Ltd. | Image processing program product and device for executing retinex processing |
| JP2005026804A (en) * | 2003-06-30 | 2005-01-27 | Matsushita Electric Ind Co Ltd | Device and method of correcting image luminance |
| CN103606134A (en) * | 2013-11-26 | 2014-02-26 | 国网上海市电力公司 | Enhancing method of low-light video images |
| CN103914813A (en) * | 2014-04-10 | 2014-07-09 | 西安电子科技大学 | Colorful haze image defogging and illumination compensation restoration method |
| CN104200437A (en) * | 2014-09-04 | 2014-12-10 | 北京工业大学 | Image defogging method |
Non-Patent Citations (2)
| Title |
|---|
| 张赛楠 等: "改进的单尺度Retinex雾天图像增强算法", 《激光与红外》 * |
| 黄黎红: "单幅图像的快速去雾算法", 《光电子.激光》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106846280A (en) * | 2017-03-08 | 2017-06-13 | 沈阳工业大学 | Image defogging method based on discrete warp wavelet |
| CN108364261A (en) * | 2017-12-13 | 2018-08-03 | 湖北工业大学 | A kind of TV-Retinex single-frame images defogging methods of gradient guiding |
| CN108428217A (en) * | 2018-01-17 | 2018-08-21 | 南京理工大学 | A kind of image defogging method based on frequency-domain visual perception estimation |
| CN108428217B (en) * | 2018-01-17 | 2023-06-16 | 南京理工大学 | Image defogging method based on frequency domain visual perception estimation |
| CN112102182A (en) * | 2020-08-31 | 2020-12-18 | 华南理工大学 | Single image reflection removing method based on deep learning |
| CN112965083A (en) * | 2021-02-02 | 2021-06-15 | 英飞拓(杭州)信息系统技术有限公司 | Multi-sensing information fusion system of intelligent automobile |
| CN113012067A (en) * | 2021-03-16 | 2021-06-22 | 华南理工大学 | Retinex theory and end-to-end depth network-based underwater image restoration method |
| CN113012067B (en) * | 2021-03-16 | 2022-11-18 | 华南理工大学 | Retinex theory and end-to-end depth network-based underwater image restoration method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106447617A (en) | Improved Retinex image defogging method | |
| CN106296612B (en) | A kind of stagewise monitor video sharpening system and method for image quality evaluation and weather conditions guidance | |
| CN107527332B (en) | Low-illumination image color retention enhancement method based on improved Retinex | |
| CN110148095B (en) | Underwater image enhancement method and enhancement device | |
| CN108734670B (en) | Method for restoring single night weak-illumination haze image | |
| CN102789635B (en) | Image enhancement method and image enhancement device | |
| CN106157267B (en) | Image defogging transmissivity optimization method based on dark channel prior | |
| CN111292258B (en) | Image defogging method based on dark channel prior and bright channel prior | |
| CN103020920B (en) | Method for enhancing low-illumination images | |
| CN103914813B (en) | The restored method of colored haze image defogging and illumination compensation | |
| CN103279931B (en) | Mist elimination image denoising method based on absorbance | |
| CN108537756B (en) | Single image defogging method based on image fusion | |
| CN108765342A (en) | A kind of underwater image restoration method based on improvement dark | |
| CN106846263A (en) | The image defogging method being immunized based on fusion passage and to sky | |
| CN107103591A (en) | A kind of single image to the fog method based on image haze concentration sealing | |
| CN102289791B (en) | Method for quickly demisting single image | |
| CN104240194A (en) | Low-light-level image enhancement algorithm based on parabolic function | |
| CN105631823A (en) | Dark channel sky area defogging method based on threshold segmentation optimization | |
| CN104318529A (en) | Method for processing low-illumination images shot in severe environment | |
| CN105046658A (en) | Low-illumination image processing method and device | |
| CN108288258A (en) | A kind of low-quality images Enhancement Method under severe weather conditions | |
| CN105513015B (en) | A kind of clearness processing method of Misty Image | |
| CN105931208A (en) | Physical model-based low-illuminance image enhancement algorithm | |
| CN107203977A (en) | A kind of underwater image restoration method based on dark primary priori and rarefaction representation | |
| CN104091307A (en) | Frog day image rapid restoration method based on feedback mean value filtering |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into 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: 20170222 |