CN112465708A - Improved image defogging method based on dark channel - Google Patents
Improved image defogging method based on dark channel Download PDFInfo
- Publication number
- CN112465708A CN112465708A CN202011144335.2A CN202011144335A CN112465708A CN 112465708 A CN112465708 A CN 112465708A CN 202011144335 A CN202011144335 A CN 202011144335A CN 112465708 A CN112465708 A CN 112465708A
- Authority
- CN
- China
- Prior art keywords
- image
- dark channel
- value
- defogging
- foggy
- 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.)
- Withdrawn
Links
- 238000000034 method Methods 0.000 title claims abstract description 26
- 238000002834 transmittance Methods 0.000 claims abstract description 21
- 238000001914 filtration Methods 0.000 claims abstract description 10
- 230000015556 catabolic process Effects 0.000 claims abstract description 8
- 238000006731 degradation reaction Methods 0.000 claims abstract description 8
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000012937 correction Methods 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 230000001131 transforming effect Effects 0.000 claims description 3
- 241000255969 Pieris brassicae Species 0.000 abstract description 3
- 238000012545 processing Methods 0.000 description 6
- 230000007547 defect Effects 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 230000011218 segmentation Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000007670 refining Methods 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
Images
Classifications
-
- G06T5/73—
Abstract
The invention provides an improved image defogging method based on a dark channel, which comprises the steps of calculating an atmospheric light value of a foggy image and a dark channel image; performing Retinex enhancement on the foggy image to obtain an initial defogged image; calculating the atmosphere rough transmittance according to the dark channel image of the foggy image and the initial defogged image; guiding filtering is adopted to refine the coarse transmittance; and obtaining a restored image according to the atmospheric light value, the thinned atmospheric transmittance value and the image degradation model of the dark channel prior defogging algorithm. The invention effectively avoids errors generated by sky areas and large white areas.
Description
Technical Field
The invention belongs to the field of computer vision, and particularly relates to an improved image defogging method based on a dark channel.
Background
The fog day has great interference to the collection of images and videos, further influences subsequent detection, identification and the like, and has great influence on security protection, monitoring and the like. Therefore, the research on the image defogging has great value. In recent years, research in this area has advanced significantly, largely divided into two categories, based on physical models and non-physical models. The non-physical model mainly comprises a Retinex algorithm, such as an automobile license plate image defogging research based on Retinex proposed by the people of the Kishina, the Wen of yellow, and the like, and then a single-scale and multi-scale Retinex enhancement algorithm and the like are developed. This method performs defogging by changing the contrast of an image, has disadvantages of high complexity, and is liable to produce halation and the like, and thus is less used. The physical model is to establish a degradation model of the image and obtain the value of each parameter, thereby obtaining the defogged image. The representative algorithm is a dark channel prior defogging algorithm proposed by the hocamen, and the algorithm has achieved a more ideal result on the processing of the foggy image of the non-sky area, and has the defect that the processing is not in place on the sky area.
Aiming at the problem of sky area processing effect, an image defogging algorithm combining sky segmentation and a super-pixel level dark channel is proposed, but a few boundaries are lost by the method. Dark channel prior defogging algorithm for optimizing region segmentation is also provided, but has the defect that the sky region is easy to generate wrong segmentation.
Disclosure of Invention
The technical scheme for realizing the purpose of the invention is as follows: an image defogging method based on dark channel improvement comprises the following specific steps:
step 1, calculating an atmospheric light value of a foggy image;
step 2: calculating a dark channel image of the foggy image;
and step 3: performing Retinex enhancement on the foggy image to obtain an initial defogged image;
and 4, step 4: calculating the atmosphere rough transmittance according to the dark channel image of the foggy image and the initial defogged image;
and 5: guiding filtering is adopted to refine the coarse transmittance;
and 6, obtaining a restored image according to the atmospheric light value, the refined atmospheric transmittance value and the image degradation model of the dark channel prior defogging algorithm.
Preferably, the atmospheric light value is calculated by adopting a hierarchical search method based on the quadtree subspace division, and the method comprises the following specific steps:
step 1.1: dividing the foggy image into four rectangular areas;
step 1.2: evaluating the four rectangular areas, wherein the evaluation standard is that the standard deviation is subtracted from the average value of the pixels in the rectangular areas, and the rectangular area with the largest value is taken as a target area;
step 1.3: repeating the step 1.1 and the step 1.2 on the target area until the target area is smaller than a set threshold value; the pixel value of the point in the target region that minimizes | (ir (p), ig (p), ib (p)) - (255 ) | |, where ir (p), ig (p), ib (p) are R, G, B component values of the pixel points in the target region, respectively, is used as the atmospheric light value.
Preferably, the specific steps of calculating the dark channel image of the foggy image are as follows:
and (4) taking the minimum value from the three channels of the fog image rgb to form a gray image, and carrying out minimum value filtering on the gray image to obtain a dark channel image.
Preferably, the Retinex enhancement is performed on the foggy image, and the specific steps of obtaining the initial defogged image are as follows:
step 3.1: transforming the hazy image to a logarithmic region;
step 3.2: the central surround function F (x, y) is integrated into 1 and a Gaussian surround scale C, and a scale lambda is obtained:
∫∫F(x，y)dxdy＝1
in the formula, (x, y) is a pixel point coordinate;
step 3.3: the output image r (x, y) is found by the following equation:
wherein S (x, y) is an original foggy image, and F (x, y) is a center surrounding function;
step 3.4: converting the output image r (x, y) obtained in the step 3.4 into a real number domain to obtain an initial defogging image;
step 3.5: and mapping the gray value of the initial defogged image obtained in the step 3.4 to 0-255.
Preferably, the specific steps of calculating the atmospheric coarse transmittance from the dark channel image of the fogging image and the initial defogging image are as follows:
carrying out minimum value filtering on the image subjected to Retinex enhancement to obtain a dark channel image subjected to Retinex enhancement and defogging;
and (3) calculating the atmospheric crude transmittance, wherein the specific calculation formula is as follows:
in the formula Ic(y) Retinex enhanced defogged dark channel image, Jc(y) is a dark channel image of the foggy image, [ omega ] is a correction coefficient, AcIs the atmospheric light value, and t (x) is the atmospheric crude throw ratio.
Preferably, the image degradation model of the dark channel prior defogging algorithm is specifically:
I(x)＝J(x,t(x)+A[I-t(x)]
wherein J (x) is the original foggy image, t (x) is the refined atmospheric transmittance, A is the atmospheric light value, and I (x) is the final dehazed image.
Compared with the prior art, the invention has the following advantages: (1) the invention effectively solves the problem that the sky area is not processed in place; (2) when the atmospheric light value is calculated, errors generated by a sky area and a large white area are effectively avoided; (3) compared with other improved methods, the method has the advantages of small calculation amount and good real-time property.
Drawings
FIG. 1 is a flow chart of the present invention.
FIG. 2 is a target image acquired prior to defogging according to the present invention.
FIG. 3 is a target image acquired after defogging according to the present invention.
Detailed Description
As shown in fig. 1 to 3, aiming at the defects of the existing dark channel defogging method, the invention provides an image defogging method based on dark channel improvement, which is used for solving the problems that the processing of an empty area in a dark channel defogging sky is not available and the atmospheric light value is mistakenly obtained. Firstly, a hierarchical search method based on quadtree subspace division is adopted for the calculation of the atmospheric light value. Compared with a method for solving an atmospheric light value in a dark channel defogging algorithm, the method can effectively avoid errors generated in a sky area or a large white area. And then, obtaining a rough defogged image by using a non-physical model Retinex enhancement algorithm, wherein the algorithm has a good defogging effect on the sky area. And processing the defogged image to obtain a dark channel image, setting a reasonable threshold value for comparison, and enabling the image to be zero when the image is smaller than the threshold value, so that the dark channel value of the non-sky area meets the dark channel priori principle, and the dark channel value of the sky area is obtained. The method comprises the following steps of substituting an atmospheric scattering model formula to obtain a defogged picture by utilizing an atmospheric light value, a foggy picture dark channel value and a defogged picture dark channel value of a Retinex enhancement algorithm, wherein the method comprises the following steps:
step 1: calculating an atmospheric light value A of the foggy image;
in some embodiments, a hierarchical search method based on quadtree subspace partitioning is adopted to calculate an atmospheric light value, so as to solve the problem that the atmospheric light value calculation is interfered by a white area in a dark channel defogging algorithm, so that the white area is mistakenly taken as the atmospheric light value, and the specific steps are as follows:
step 1.1: dividing the foggy image into four rectangular areas;
step 1.2: evaluating the four rectangles, wherein the evaluation standard is that the standard deviation is subtracted from the average value of the pixels in the rectangular area, and the rectangular area with the maximum value is taken as a target area;
step 1.3: repeating the step 1.1 and the step 1.2 on the target area until the target area is smaller than a set threshold value; the pixel value of the point in the target region where | (ir (p), ig (p), ib (p)) - (255 ) | | is the smallest is taken as the atmospheric light value, and ir (p), ig (p), ib (p) are R, G, B component values of the pixel points in the target region, respectively.
Step 2: dark channel images of the hazy images are calculated.
In some embodiments, the minimum value in the three channels of the fogging image rgb is taken to form a gray map, and the gray map is subjected to minimum value filtering to obtain a dark channel image.
And step 3: and performing Retinex enhancement on the foggy image to obtain an initial defogged image.
In a further embodiment, the specific steps of performing Retinex enhancement on the foggy image to obtain an initial defogged image are as follows:
step 3.1: transforming the hazy image to a logarithmic region;
where R (x, y) is the output image, R (x, y) is the reflection image, L (x, y) is the luminance image, and S (x, y) is the original image.
Step 3.2: using the central surround function F (x, y) integrated to 1, and the gaussian surround scale C, the scale λ is found:
∫∫F(x，y)dxdy＝1
in the formula, (x, y) is the pixel point coordinate.
Step 3.3: the output image r (x, y) is found by the following equation:
where S (x, y) is the original foggy image and F (x, y) is the center-surround function.
Step 3.4: and (4) converting the output image r (x, y) obtained in the step (3.4) into a real number domain to obtain an initial defogged image.
Step 3.5: and mapping the gray value of the image obtained in the step 3.4 to 0-255.
And 4, step 4: the atmospheric coarse transmittance is calculated from the dark channel image of the hazy image and the initial defogged image.
And carrying out minimum value filtering on the image subjected to Retinex enhancement and defogging to obtain a dark channel image subjected to Retinex enhancement and defogging, and carrying out threshold processing on the dark channel image subjected to Retinex enhancement and defogging to ensure that a dark channel in a non-sky area is zero and meet dark channel priorality.
And (3) calculating the atmospheric crude transmittance, wherein the specific calculation formula is as follows:
in the formula Ic(y) Retinex enhanced defogged dark channel image, Jc(y) is a dark channel image of the foggy image, omega is a correction coefficient, and 0.95 is taken, AcIs the atmospheric light value, and t (x) is the atmospheric crude throw ratio.
And 5: and (3) refining the coarse transmittance by adopting guide filtering, wherein the specific formula is as follows:
qi＝∑jWij(I)pj
wherein I is the image after Retinex enhanced defogging, PjIs an atmospheric coarse transmittance image, qiFor images after thinning the transmission, WijIs a weight value.
And 6, substituting the atmospheric light value and the refined atmospheric transmittance value into an image degradation model of a dark channel prior defogging algorithm to obtain a restored image.
In a further embodiment, the image degradation model of the dark channel prior defogging algorithm is specifically as follows:
I(x)＝J(x)t(x)+A[I-t(x)]
wherein J (x) is the original foggy image, t (x) is the refined atmospheric transmittance, A is the atmospheric light value, and I (x) is the final dehazed image.
Claims (6)
1. An improved image defogging method based on a dark channel is characterized by comprising the following specific steps:
step 1, calculating an atmospheric light value of a foggy image;
step 2: calculating a dark channel image of the foggy image;
and step 3: performing Retinex enhancement on the foggy image to obtain an initial defogged image;
and 4, step 4: calculating the atmosphere rough transmittance according to the dark channel image of the foggy image and the initial defogged image;
and 5: guiding filtering is adopted to refine the coarse transmittance;
and 6, obtaining a restored image according to the atmospheric light value, the refined atmospheric transmittance value and the image degradation model of the dark channel prior defogging algorithm.
2. The image defogging method based on the dark channel improvement as claimed in claim 1, wherein the atmospheric light value is calculated by adopting a hierarchical search method based on the quadtree subspace division, and the specific steps are as follows:
step 1.1: dividing the foggy image into four rectangular areas;
step 1.2: evaluating the four rectangular areas, wherein the evaluation standard is that the standard deviation is subtracted from the average value of the pixels in the rectangular areas, and the rectangular area with the largest value is taken as a target area;
step 1.3: repeating the step 1.1 and the step 1.2 on the target area until the target area is smaller than a set threshold value; the pixel value of the point in the target region that minimizes | (ir (p), ig (p), ib (p)) - (255 ) | |, where ir (p), ig (p), ib (p) are R, G, B component values of the pixel points in the target region, respectively, is used as the atmospheric light value.
3. The method for improving image defogging based on the dark channel according to claim 1, wherein the step of calculating the dark channel image of the fogging image comprises the specific steps of:
and (4) taking the minimum value from the three channels of the fog image rgb to form a gray image, and carrying out minimum value filtering on the gray image to obtain a dark channel image.
4. The method of claim 1, wherein the Retinex enhancement is performed on the foggy image, and the specific steps of obtaining the initial foggy image are as follows:
step 3.1: transforming the hazy image to a logarithmic region;
step 3.2: the central surround function F (x, y) is integrated into 1 and a Gaussian surround scale C, and a scale lambda is obtained:
∫∫F(x，y)dxdy＝1
in the formula, (x, y) is a pixel point coordinate;
step 3.3: the output image r (x, y) is found by the following equation:
wherein S (x, y) is an original foggy image, and F (x, y) is a center surrounding function;
step 3.4: converting the output image r (x, y) obtained in the step 3.4 into a real number domain to obtain an initial defogging image;
step 3.5: and mapping the gray value of the initial defogged image obtained in the step 3.4 to 0-255.
5. The method for defogging images based on improvement of the dark channel according to claim 1, wherein the step of calculating the coarse transmittance of the atmosphere based on the image of the dark channel of the fogging image and the initial defogging image comprises the specific steps of:
carrying out minimum value filtering on the image subjected to Retinex enhancement to obtain a dark channel image subjected to Retinex enhancement and defogging;
and (3) calculating the atmospheric crude transmittance, wherein the specific calculation formula is as follows:
in the formula Ic(y) Retinex enhanced defogged dark channel image, Jc(y) is a dark channel image of the foggy image, [ omega ] is a correction coefficient, AcIs the atmospheric light value, and t (x) is the atmospheric crude throw ratio.
6. The dark channel-based improved image defogging method according to claim 1, wherein the image degradation model of the dark channel prior defogging algorithm is specifically:
I(x)＝J(x)t(x)+A[I-t(x)]
wherein J (x) is the original foggy image, t (x) is the refined atmospheric transmittance, A is the atmospheric light value, and I (x) is the final dehazed image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011144335.2A CN112465708A (en) | 2020-10-23 | 2020-10-23 | Improved image defogging method based on dark channel |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011144335.2A CN112465708A (en) | 2020-10-23 | 2020-10-23 | Improved image defogging method based on dark channel |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112465708A true CN112465708A (en) | 2021-03-09 |
Family
ID=74834433
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011144335.2A Withdrawn CN112465708A (en) | 2020-10-23 | 2020-10-23 | Improved image defogging method based on dark channel |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112465708A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113436095A (en) * | 2021-06-24 | 2021-09-24 | 哈尔滨理工大学 | Defogging method for sky area image |
| CN115861104A (en) * | 2022-11-30 | 2023-03-28 | 西安电子科技大学 | Remote sensing image defogging method based on transmissivity refinement |
| CN116188331A (en) * | 2023-04-28 | 2023-05-30 | 淄博市淄川区市政环卫服务中心 | Construction engineering construction state change monitoring method and system |
-
2020
- 2020-10-23 CN CN202011144335.2A patent/CN112465708A/en not_active Withdrawn
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113436095A (en) * | 2021-06-24 | 2021-09-24 | 哈尔滨理工大学 | Defogging method for sky area image |
| CN115861104A (en) * | 2022-11-30 | 2023-03-28 | 西安电子科技大学 | Remote sensing image defogging method based on transmissivity refinement |
| CN115861104B (en) * | 2022-11-30 | 2023-10-17 | 西安电子科技大学 | Remote sensing image defogging method based on transmissivity refinement |
| CN116188331A (en) * | 2023-04-28 | 2023-05-30 | 淄博市淄川区市政环卫服务中心 | Construction engineering construction state change monitoring method and system |
| CN116188331B (en) * | 2023-04-28 | 2023-07-18 | 淄博市淄川区市政环卫服务中心 | Construction engineering construction state change monitoring method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108596849B (en) | Single image defogging method based on sky region segmentation | |
| CN107680054B (en) | Multi-source image fusion method in haze environment | |
| CN107103591B (en) | Single image defogging method based on image haze concentration estimation | |
| CN106157267B (en) | Image defogging transmissivity optimization method based on dark channel prior | |
| CN112465708A (en) | Improved image defogging method based on dark channel | |
| CN108876743B (en) | Image rapid defogging method, system, terminal and storage medium | |
| CN107301624B (en) | Convolutional neural network defogging method based on region division and dense fog pretreatment | |
| KR101448164B1 (en) | Method for Image Haze Removal Using Parameter Optimization | |
| CN108389175B (en) | Image defogging method integrating variation function and color attenuation prior | |
| CN109087254B (en) | Unmanned aerial vehicle aerial image haze sky and white area self-adaptive processing method | |
| CN105913390B (en) | A kind of image defogging method and system | |
| CN108564597B (en) | Video foreground object extraction method fusing Gaussian mixture model and H-S optical flow method | |
| CN110009653A (en) | Increase limb recognition point sharp picture based on gray level threshold segmentation method and knows method for distinguishing | |
| CN114118144A (en) | Anti-interference accurate aerial remote sensing image shadow detection method | |
| CN110245600B (en) | Unmanned aerial vehicle road detection method for self-adaptive initial quick stroke width | |
| CN111861896A (en) | UUV-oriented underwater image color compensation and recovery method | |
| CN109003238B (en) | Image haze removal method based on model, histogram and gray level enhancement | |
| CN113313702A (en) | Aerial image defogging method based on boundary constraint and color correction | |
| CN108898132A (en) | A kind of terahertz image dangerous material recognition methods based on Shape context description | |
| Choi et al. | Fog detection for de-fogging of road driving images | |
| CN112750089B (en) | Optical remote sensing image defogging method based on local block maximum and minimum pixel prior | |
| CN112288780B (en) | Multi-feature dynamically weighted target tracking algorithm | |
| CN111091501A (en) | Parameter estimation method of atmosphere scattering defogging model | |
| Ding et al. | Restoration of single sand-dust image based on style transformation and unsupervised adversarial learning | |
| CN116883303A (en) | Infrared and visible light image fusion method based on characteristic difference compensation and fusion |
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 | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20210309 |