CN105005966B - A kind of single image based on yellow haze physical characteristic goes haze method - Google Patents
A kind of single image based on yellow haze physical characteristic goes haze method Download PDFInfo
- Publication number
- CN105005966B CN105005966B CN201510023351.9A CN201510023351A CN105005966B CN 105005966 B CN105005966 B CN 105005966B CN 201510023351 A CN201510023351 A CN 201510023351A CN 105005966 B CN105005966 B CN 105005966B
- Authority
- CN
- China
- Prior art keywords
- image
- channel
- pixel
- haze
- numerical 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.)
- Active
Links
Landscapes
- Image Processing (AREA)
- Facsimile Image Signal Circuits (AREA)
Abstract
Haze algorithm is removed the invention discloses a kind of single image based on yellow haze physical characteristic, belongs to image processing field.The invention includes obtaining pending image, determine the sky areas in pending image, air luminous point is determined in region on high, determine the ratio of color channel in pending image, according to the ratio of color channel in pending passage, the transmission figure of pending image is determined, with reference to the saturation degree of the physical characteristic regulation transmission figure of yellow haze, the image after generation regulation.The present invention, can be by image adjustment to closer to real color, so as to recover the original brightness of image and saturation degree relative to the mode that mist and haze are uniformly processed in the prior art.
Description
Technical field
The present invention relates to image processing field, more particularly to a kind of single image based on yellow haze physical characteristic goes to haze side
Method.
Background technology
Nowadays air pollution is increasingly serious, especially haze weather, almost can all occur daily after the winter is entered, in outdoor bat
The picture taken the photograph has scattering process due to haze particle to light so that image detail is obscured, and overall image quality declines.
In the prior art, the image processing method of main flow is to obtain the air light value in picture, true according to air light value
The distribution map of propagation in atmosphere transmissivity in the fixed image, and then defogging expression formula is determined according to propagation in atmosphere transmissivity, so as to root
Haze expression formula is gone to handle image according to above-mentioned, so as to reach the influence for removing haze sky to image.
But, inventor has found there is problems with the prior art：
Brightness highest brightness value in image is determined by naked eyes to the selection of air light value in the prior art, not accounted for
The accurate meaning of brightness in atmospherical scattering model, and to mist and haze, both are not different during image procossing
Weather condition, which makes a distinction, to be treated, but carries out unified processing, because mist is different with the specific origin cause of formation of haze, so using unified
The result of processing mode is can not to reach the effect corrected aberration, recover the original brightness and contrast of image.
The content of the invention
In order to solve problem of the prior art, haze is removed the invention provides a kind of single image based on yellow haze physical characteristic
Method, the single image goes haze method to include：
Pending image is obtained, the sky areas in the pending image is determined；
Air luminous point is determined in the sky areas, the ratio of color channel in the pending image is determined；
According to the ratio of color channel in the pending passage, the transmission figure of the pending image is determined；
With reference to the physical characteristic of yellow haze, the saturation degree of the transmission figure, the image after generation regulation are adjusted.
Optionally, it is described to obtain pending image, the sky areas in the pending image is determined, including：
The pending image is divided into multiple regions, the first area in the pending image is extracted, to described
First area carries out Color edge detection, obtains edge image；
Binary conversion treatment, the near edge image after being handled are carried out to the edge image；
Detect in the near edge image, if meeting the first preparatory condition, judge the near edge figure correspondence
The first area be sky areas.
Optionally, it is described that air luminous point is determined in the sky areas, determine color channel in the pending image
Ratio, including：
Brightness highest point is chosen at least one described sky areas and is used as air luminous point；
The first number ratios of red channel, green channel, blue channel in the air luminous point are determined, by described first
Number ratios are used as color channel ratio in the pending image.
Optionally, the ratio according to color channel in the pending passage, determines the biography of the pending image
Defeated figure, including：
According to the ratio of color channel in the pending image, by the first adjustment formula to every in the pending image
Numerical value of the individual pixel in red channel, green channel and blue channel is adjusted, the numerical value after being adjusted, and described first
Adjusting formula is specially：
Wherein, R' is the numerical value of red channel after each pixel is adjusted, and G' is the number of green channel after each pixel is adjusted
Value, B' is the numerical value of blue channel after each pixel is adjusted, and R is the numerical value of red channel before each pixel is adjusted, and G is each picture
The numerical value of green channel before element adjustment, B is the numerical value of blue channel before each pixel is adjusted, RwFor air luminous point red channel
Numerical value, GwFor the numerical value of air luminous point green channel, BwFor the numerical value of atmosphere light point blue channel, T is in default passage value
Limit；
Based on the relation of red, green, blue scattered power and wavelength, with reference to human eye to described red, green, blueness quick
Sense degree, it is determined that when color space is converted to HSI (Hue-Saturation-Intensity, colourity-saturation degree-intensity)
Intensity-conversion formula, the intensity level of each pixel in the pending image is obtained according to the intensity-conversion formula, described strong
Spending conversion formula is specially：
I=0.4520*R'+0.5121*G'+0.0359*B',
Wherein, I is the intensity level in HSI color spaces；
Conversion to after the HSI color spaces, choose in the pending image each pixel the red channel,
The minimum value of numerical value in green channel, blue channel, gray-scale map is constituted according to the minimum value, the gray-scale map is carried out bilateral
Filtering, constitutes the transmission figure of the pending image.
Optionally, the physical characteristic of the yellow haze of the combination, adjusts the saturation degree of the transmission figure, the figure after generation regulation
Picture, including：
With reference to the second adjustment formula, the saturation degree to each pixel in the transmission figure is adjusted, after generation regulation
Image, described second, which adjusts formula, is specially：
Wherein, S is the initial saturation of each pixel in the pending image, and S' is the saturation of each pixel after processing
Degree.
The beneficial effect that the technical scheme that the present invention is provided is brought is：
By determining the sky areas in pending image, and then the most bright spot chosen in sky areas is used as white balance
Standard point, the color for then carrying out full image according to the color channel ratio of the standard point is adjusted, and carries out color space
Conversion, and the saturation degree after conversion is adjusted according to the characteristic of yellow haze, it is to avoid haze can not be directed in the prior art
Characteristic to image color implement regulation generation, reduce image aberration, as far as possible recover the original brightness and contrast of image.
Brief description of the drawings
In order to illustrate more clearly of technical scheme, the accompanying drawing used required in being described below to embodiment
It is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the present invention, general for this area
For logical technical staff, on the premise of not paying creative work, other accompanying drawings can also be obtained according to these accompanying drawings.
Fig. 1 is the schematic flow sheet that a kind of single image based on yellow haze physical characteristic that the present invention is provided removes haze algorithm；
Fig. 2 is the filtering mode that a kind of single image based on yellow haze physical characteristic that the present invention is provided is gone in haze algorithm
Contrast schematic diagram；
Fig. 3 is the position view of the original image of sample 1, near edge image and air luminous point；
Fig. 4 is the position view of the original image of sample 2, near edge image and air luminous point；
Fig. 5 is the position view of the original image of sample 3, near edge image and air luminous point；
Histogram is contrasted before and after original image, color balance simulation result and the emulation of Fig. 6 samples 4；
Histogram is contrasted before and after original image, color balance simulation result and the emulation of Fig. 7 samples 5；
The image that Fig. 8 is the original image of sample 6 and gone after haze emulation；
The image that Fig. 9 is the original image of sample 7 and gone after haze emulation.
Embodiment
To make the structure and advantage of the present invention clearer, the structure of the present invention is made further below in conjunction with accompanying drawing
Description.
Embodiment one
The present embodiment provides a kind of single image based on yellow haze physical characteristic and goes haze method, as shown in figure 1, the single width
Image goes haze method to include：
101st, pending image is obtained, the sky areas in the pending image is determined.
102nd, air luminous point is determined in the sky areas, the ratio of color channel in the pending image is determined.
103rd, according to the ratio of color channel in the pending passage, the transmission figure of the pending image is determined.
104th, with reference to the physical characteristic of yellow haze, the saturation degree of the transmission figure, the image after generation regulation are adjusted.
In force, sky areas is determined in pending image first, is determined on high in region after air luminous point, root
The ratio of color channel in pending image is determined according to fixed air luminous point, and then is treated according to the determination of the ratio of color channel
The transmission figure of image is handled, finally according to the physical characteristic of yellow haze, the saturation degree to the transmission figure is adjusted, generation regulation
Image afterwards.It is overall remove haze during, be used as the color in pending image due to choosing air luminous point in sky areas
Passage adjustment standard, can so recover the original color of image as far as possible, plus the physical characteristic for yellow haze to figure
The saturation degree of picture is adjusted, so can be with respect to the mode with mist and haze are uniformly processed in the prior art, can be by image
Regulation is to closer to real color, so as to recover the original brightness of image and saturation degree.
Optionally, it is described to obtain pending image, the sky areas in the pending image is determined, including：
Step one, the pending image is divided into multiple regions, extracts the first area in the pending image,
Color edge detection is carried out to the first area, edge image is obtained.
In force, analyzed according to investigation statisticses, most of picture under yellow haze weather is shot on daytime outside room,
And in the environment outside daytime room, the main source of illumination is direct sunlight and the diffusing reflection daylight, correspondence intensity of illumination
Highest region is exactly sky areas, other parts more or less blocking due to building or other objects, corresponding illumination
Intensity is all without higher than sky areas, in this case, even if there is the interference of yellow haze, the loss of the brightness value of sky areas
Also will not be too high, therefore determine in pending image sky areas, just into one it is critically important the problem of.
In above-mentioned steps one, region division is carried out to pending image in advance, multiple regions are obtained, an area is chosen every time
The corresponding image in domain carries out Color edge detection, and specific Color edge detection method, which can be used, calculates color gradient
ColorGrad algorithms, carry out rim detection for the coloured image in rgb color space, obtain edge image, the edge image
For gray level image.
Step 2, binary conversion treatment, the near edge image after being handled are carried out to the edge image.
In force, because edge image is gray level image, comprising multiple excessive GTG colors, at the later stage
Reason, introduces binaryzation principle here, carries out binary conversion treatment to the edge image obtained in step one, is exactly briefly to set
Vertical differentiation threshold value, will be higher than the pixel assignment 1 of the differentiation threshold value, i.e., in the corresponding gray value of each pixel in edge image
For white, the pixel assignment 0 of the differentiation threshold value will be equal to or less than in the corresponding gray value of each pixel, as black, so
What is obtained is the edge graph of only black and white, and the edge image a kind of relative to step is more succinct fine, is referred to as fine
Edge image.
Step 3, detects in the near edge image, if meeting the first preparatory condition, judges the near edge
The corresponding first area is schemed for sky areas.
In force, the near edge image generated is detected with reference to the first preparatory condition, here first
Preparatory condition is specially：
(1) the corresponding assignment of pixel is 0；
(2) the corresponding luminance component of pixel is more than predetermined threshold value It；
(3) the corresponding saturation degree component of pixel is less than predetermined threshold value St；
Wherein, predetermined threshold value It=0.65*Imax+0.35*Imin,ImaxAnd IminRespectively
The maximum and minimum value of region P luminance components, SmedFor the intermediate value of region P saturation degree components.
In order to make it easy to understand, being illustrated here in conjunction with specific data.For example, choose in pending image be located at (50,
380) pixel is handled, and numerical value of the pixel RGB namely respectively in red channel, green channel, blue channel is
(0.8902,0.8471,0.7607), the predetermined threshold value formula in the first preparatory condition, predetermined threshold value ImaxFor 0.7843,
IminFor 0.2157, SmedFor 0.2298, the I obtained with reference to above-mentioned calculation formulat=0.5853, St=0.2298, for (50,
380) I=0.7608 of pixel>It, S=0.0863<St, meets the requirements, therefore the pixel in first area passes through first
Preparatory condition is detected, if pixel all in the first area all presses inspection of the above-mentioned steps by the first preparatory condition
Survey, then the first area is chosen to be sky areas.
It is noted that because in normal take pictures, sky areas is essentially in the upper plate portion of pending image
In point, so a kind of first area of selection of step is preferably the top half of pending image in most cases, if
, can be with the first region by meeting the sky areas of the first preparatory condition at above-mentioned steps one to step 3 not selection
According to actual use demand, it is considered to whether need selection other parts to carry out the judgement as shown in step one to step 3 again
Journey, untill the sky areas for meeting the first preparatory condition is selected.By above-mentioned steps once can be from treating to step 3
The step of handling and sky areas is chosen in image, and be used for afterwards proceeds processing.
Optionally, it is described that air luminous point is determined in the sky areas, determine color channel in the pending image
Ratio, including：
Brightness highest point is chosen at least one described sky areas and is used as air luminous point；
The first number ratios of red channel, green channel, blue channel in the air luminous point are determined, by described first
Number ratios as color channel in the pending image ratio.
In force, the sky areas that may determine in step 3 has multiple, it is determined that at least one sky areas
In, brightness one point of highest is chosen as air luminous point, then obtains the air luminous point in pending image respectively red
Numerical value in chrominance channel, three passages of green channel and blue channel, and be defined by the numerical value of green channel, determine respectively red
The number ratios of passage, blue channel relative to green channel.
Here why with the standard of itself white balance of air luminous point as pending image adjustment white balance, be because
Air luminous point has the highest brightness value in pending image, relative to other pixels, and the effect influenceed by yellow haze is minimum, with
The pending image of white balance de-regulation of the air luminous point can have best regulating effect.
Optionally, the ratio according to color channel in the pending passage, determines the biography of the pending image
Defeated figure, including：
According to the ratio of color channel in the pending image, by the first adjustment formula to every in the pending image
Numerical value of the individual pixel in red channel, green channel and blue channel is adjusted, the numerical value after being adjusted, and described first
Adjusting formula is specially：
Wherein, R' is the numerical value of red channel after each pixel is adjusted, and G' is the number of green channel after each pixel is adjusted
Value, B' is the numerical value of blue channel after each pixel is adjusted, and R is the numerical value of red channel before each pixel is adjusted, and G is each picture
The numerical value of green channel before element adjustment, B is the numerical value of blue channel before each pixel is adjusted, RwFor air luminous point red channel
Numerical value, GwFor the numerical value of air luminous point green channel, BwFor the numerical value of atmosphere light point blue channel, T is in default passage value
Limit；
Based on the relation of red, green, blue scattered power and wavelength, with reference to human eye to described red, green, blueness quick
Sense degree, it is determined that when color space is converted to HSI (Hue-Saturation-Intensity, colourity-saturation degree-intensity)
Intensity-conversion formula, the intensity level of each pixel in the pending image is obtained according to the intensity-conversion formula, described strong
Spending conversion formula is specially：
I=0.4520*R'+0.5121*G'+0.0359*B',
Wherein, I is the intensity level in HSI color spaces；
Conversion to after the HSI color spaces, choose in the pending image each pixel the red channel,
The minimum value of numerical value in green channel, blue channel, gray-scale map is constituted according to the minimum value, the gray-scale map is carried out bilateral
Filtering, constitutes the transmission figure of the pending image.
In force, the numerical value by the air luminous point determined before respectively in red channel, green channel, blue channel
The R substituted into respectively in the first adjustment formulaw、Gw、BwIn, other pixels in pending image are logical in red channel, green
Numerical value in road, blue channel is substituted into R, G, B respectively, with reference to default passage value upper limit T, it is possible to public according to the first adjustment
Formula determines the numerical value in the red channel after each pixel adjustment, green channel, blue channel, that is, completes according to air
The step of luminous point white balance carries out blank level adjustment to the pending image of view picture.
Due to being mixed with the substantial amounts of molecule such as vapor, dust in an atmosphere so that " gas is molten for whole atmospheric environment composition
The colloid of glue ", so that the influence of scattering can be produced to the light by air, and for without the corresponding light of wavelength, dissipating
The degree penetrated also is not quite similar.Be actually based under haze weather due to caused by Rayleigh scattering each wave band coloured light to transfer rate tribute
The difference offered, it is considered to scattered power and the relation of wavelength under the conditions of Rayleigh scattering, with reference to the calculation formula of following transfer rate：
T (x)=e-βd(x),
The ratio that transfer rate can be obtained is about：0.6498:0.3679:0.1514.
With reference to difference in perception of the human eye to three primary colours light, for RGB three primary colours, human eye feels green glow brightness most
By force, feux rouges takes second place (the about half of green glow), and blue light is most weak (about 1/3rd of feux rouges), thus, it is supposed that the vision of green glow
Light intensity is 1, and the vision light intensity of feux rouges is only the 1/2 of its former intensity, and blue light is then only the 1/6 of its former intensity.
Further, process step before is all based on what rgb color space was handled entirely, under the color space
Easily there is oversaturated phenomenon in image, therefore in order to realize the accurate reparation of image color, needs exist for carrying out color space
Conversion, i.e., changed from RGB to HSI (Hue-Saturation-Intensity, colourity-saturation degree-intensity), so as to realize full
With the accurate control of degree.
Three reasons of summary, finally combine the rgb value determination after each pixel adjustment strong under HSI color spaces
Spending conversion formula is：
I=0.4520*R'+0.5121*G'+0.0359*B',
Wherein, R' is the numerical value of red channel after each pixel is adjusted, and G' is the number of green channel after each pixel is adjusted
Value, B' is the numerical value of blue channel after each pixel is adjusted, in HSI color spaces, is determined by above-mentioned intensity-conversion formula
Intensity (i.e. brightness) value of each pixel.
Further, after conversion to HSI color spaces, each pixel in pending image is chosen logical in the red
The minimum value of numerical value in road, green channel, blue channel, gray-scale map is constituted according to the minimum value, and the gray-scale map is carried out
Bilateral filtering, constitutes the transmission figure of the pending image.
Above-mentioned gray-scale map is exactly inherently a gray-scale map, the gray value of each pixel in transmission figure each pixel three
The minimum value planted in Color Channel is determined, if numerical value of that is, one pixel in red channel, green channel, blue channel
Respectively (0.8902,0.8471,0.7607), then wherein minimum value 0.7607 is taken as the corresponding gray scale of the pixel in gray-scale map
Value.
It is noted that bilateral filtering here is for the mini-value filtering that conventional method is commonly used, pass through
Above-mentioned gray-scale map is handled in spatial domain and codomain simultaneously, it is specific as shown in Fig. 2 for pixel A, minimum value is filtered
Ripple is obtained centered on pixel A, the pixel value of adjacent 8 pixels of surrounding, and above-mentioned 8 pixel values are asked for after minimum value,
Using minimum value as pixel A value；And after bilateral filtering is then adjacent 8 pixel values around by pixel A, with reference to each
The weight a-h of value is calculated, using obtained numerical value as pixel A value.Bilateral filtering, can relative to mini-value filtering
Value to pixel A in terms of spatial domain and codomain two is handled, it is to avoid mini-value filtering can not fully reflect surrounding picture
The defect of plain value.
Optionally, the physical characteristic of the yellow haze of the combination, adjusts the saturation degree of the transmission figure, the figure after generation regulation
Picture, including：
With reference to the second adjustment formula, the saturation degree to each pixel in the transmission figure is adjusted, after generation regulation
Image, described second, which adjusts formula, is specially：
Wherein, S is the initial saturation of each pixel in the pending image, and S' is the saturation of each pixel after processing
Degree.
In force, the aerosol system that mist is made up of the small water droplet or ice crystal that are largely suspended in surface air
System.Haze is due to that the particles such as dust, sulfuric acid, nitric acid, organic hydrocarbon compounds in air make the muddy phenomenon of air.Thus may be used
See, mist and haze are in essence and different, targetedly the image under yellow haze weather is handled in the present embodiment, because
This, with reference to yellow haze physical characteristic, it is necessary to the saturation degree of transmission figure generated to previous step is adjusted, adjusted with specific reference to second
Whole formula, will transmit the saturation degree S of each pixel in figure according to whether classified calculating is carried out more than 0.025, so as to be calculated
The intensity value of each pixel afterwards, the saturation degree of each pixel in transmission figure is replaced with after the intensity value after calculating, obtained
Image after to regulation.
It should be noted that span in practice is 0 to 255, it is logical in red channel, green in the present embodiment
The span of numerical value in road, blue channel be 0 to 1, be because by 0 to 255 numerical value made divided by 255 processing,
In order to avoid misreading, spy explains herein.
In order to show the processing advantage of this method, concrete implementation effect is as follows：
Emulation 1, the emulation to atmosphere light estimation algorithm.
Emulating 1 simulated conditions is carried out under MATLAB R2008a softwares.
Estimation atmosphere light is carried out to test sample 1-3 and carries out emulation experiment.
Fig. 3 is the position view of the original image of sample 1, near edge image and air luminous point；
Fig. 4 is the position view of the original image of sample 2, near edge image and air luminous point；
Fig. 5 is the position view of the original image of sample 3, near edge image and air luminous point.
In from Fig. 3 to Fig. 5 as can be seen that can be very good to avoid the interference of white object for sample 1, it is to avoid tradition
Algorithm caused by ignorance atmosphere light physical significance due to missing detecting leakage problem；For sample 2, due to having paid close attention to the color of image
Information, thus preferably avoid because of the interference that interpolation is produced at image depth mutation, correctly brightness abnormity point is excluded
Part is outer on high；For sample 3, due to being detected in the brightness of image and saturation degree component simultaneously, thus people is avoided
Make the interference of light source.
Emulation 2, the emulation to colour-balance algorithm.
Emulating 2 simulated conditions is carried out under MATLAB R2008a softwares.
Color balance emulation experiment is carried out to test sample 4-5.
Histogram is contrasted before and after original image, color balance simulation result and the emulation of Fig. 6 samples 4；
Histogram is contrasted before and after original image, color balance simulation result and the emulation of Fig. 7 samples 5.
From Fig. 6 and 7 as can be seen that the present invention proposes that colour-balance algorithm has good adaptivity.For in the absence of
The image of color distortion such as Fig. 6, from Fig. 6 in original image and emulating image histogram curve essentially coincide as can be seen that
This algorithm can't make an amendment, and can be very good to keep the color character of image script；For the more obvious example of color distortion
Such as Fig. 7, from Fig. 7 in original image and emulating image histogram, original image original image and emulating image
Refined image curve has many places not overlap as can be seen that this algorithm is on a balanced basis to brightness of image etc.
Information has certain change.
Emulation 3, the emulation of the final effect to removing haze.
Emulating 3 simulated conditions is carried out under MATLAB R2008a softwares.
Emulation experiment is carried out to test sample.The image that Fig. 8 is the original image of sample 6 and gone after haze emulation, Fig. 9 is sample
This 7 original image and the image gone after haze emulation.
From Fig. 8,9 as can be seen that the algorithms in the present invention achieve and good go haze effect.Can be with for sample this algorithm
The color distortion brought by dense haze is eliminated well, and is not in that satiety and/or halo (are artificially produced at depth of field mutation
Abnormal halation) phenomenon；It can retain local detailed information well in the deeper place of the depth of field for this algorithm of sample.
A kind of single image based on yellow haze physical characteristic proposed in the present embodiment removes haze algorithm, by pending figure
Air luminous point is determined as in, the ratio of color channel in pending image is determined according to fixed air luminous point, and then according to
The ratio of color channel determines the transmission figure of pending image, finally according to the physical characteristic of yellow haze, and the transmission figure is satisfied
It is adjusted with degree, the image after generation regulation.The original color of image can be recovered as far as possible, and it is special for the physics of yellow haze
Property the saturation degree of image is adjusted, so can the relative mode with mist and haze are uniformly processed in the prior art, can
By image adjustment to closer to real color, so as to recover the original brightness of image and saturation degree.
Embodiments of the invention are the foregoing is only, are not intended to limit the invention, it is all in the spirit and principles in the present invention
Within, any modification, equivalent substitution and improvements made etc. should be included in the scope of the protection.
Claims (4)
1. a kind of single image based on yellow haze physical characteristic goes haze method, it is characterised in that the single image goes haze method
Including：
Pending image is obtained, the sky areas in the pending image is determined；
Air luminous point is determined in the sky areas, the ratio of color channel in the pending image is determined；
According to the ratio of color channel in the pending image, the transmission figure of the pending image is determined, including：
According to the ratio of color channel in the pending image, by the first adjustment formula to each picture in the pending image
Numerical value of the element in red channel, green channel and blue channel is adjusted, the numerical value after being adjusted, first adjustment
Formula is specially：
Wherein, R' is the numerical value of red channel after each pixel is adjusted, and G' is the numerical value of green channel after each pixel is adjusted, B'
The numerical value of blue channel after being adjusted for each pixel, R is the numerical value of red channel before each pixel is adjusted, and G adjusts for each pixel
The numerical value of whole preceding green channel, B is the numerical value of blue channel before each pixel is adjusted, RwFor the number of air luminous point red channel
Value, GwFor the numerical value of air luminous point green channel, BwFor the numerical value of atmosphere light point blue channel, T is in default passage value
Limit；
Based on the relation of red, green, blue scattered power and wavelength, with reference to human eye to the red, green, the sensitive journey of blueness
Degree, it is determined that the intensity-conversion formula when color space is converted to HSI, obtains described pending according to the intensity-conversion formula
The intensity level of each pixel in image, the intensity-conversion formula is specially：
I=0.4520*R'+0.5121*G'+0.0359*B',
Wherein, I is the intensity level in HSI color spaces；
After conversion to the HSI color spaces, each pixel is in the red channel, green in the selection pending image
The minimum value of numerical value in passage, blue channel, gray-scale map is constituted according to the minimum value, and bilateral filter is carried out to the gray-scale map
Ripple, constitutes the transmission figure of the pending image；
With reference to the physical characteristic of yellow haze, the saturation degree of the transmission figure, the image after generation regulation are adjusted.
2. the single image according to claim 1 based on yellow haze physical characteristic goes haze method, it is characterised in that described to obtain
Pending image is taken, the sky areas in the pending image is determined, including：
The pending image is divided into multiple regions, the first area in the pending image is extracted, to described first
Region carries out Color edge detection, obtains edge image；
Binary conversion treatment, the near edge image after being handled are carried out to the edge image；
The near edge image is detected, if meeting the first preparatory condition, judges that the near edge figure is corresponding described
First area is sky areas.
3. the single image according to claim 1 based on yellow haze physical characteristic goes haze method, it is characterised in that it is described
Air luminous point is determined in the sky areas, the ratio of color channel in the pending image is determined, including：
Brightness highest point is chosen at least one described sky areas and is used as air luminous point；
The first number ratios of red channel, green channel, blue channel in the air luminous point are determined, by first numerical value
Ratio is used as color channel ratio in the pending image.
4. the single image according to claim 1 based on yellow haze physical characteristic goes haze method, it is characterised in that the knot
The physical characteristic of yellow haze is closed, the saturation degree of the transmission figure is adjusted, generates the image after regulation, including：
With reference to the second adjustment formula, the saturation degree to each pixel in the transmission figure is adjusted, the image after generation regulation,
Described second, which adjusts formula, is specially：
Wherein, S is the initial saturation of each pixel in the pending image, and S' is the saturation degree of each pixel after processing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510023351.9A CN105005966B (en) | 2015-01-16 | 2015-01-16 | A kind of single image based on yellow haze physical characteristic goes haze method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510023351.9A CN105005966B (en) | 2015-01-16 | 2015-01-16 | A kind of single image based on yellow haze physical characteristic goes haze method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105005966A CN105005966A (en) | 2015-10-28 |
| CN105005966B true CN105005966B (en) | 2017-07-28 |
Family
ID=54378625
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510023351.9A Active CN105005966B (en) | 2015-01-16 | 2015-01-16 | A kind of single image based on yellow haze physical characteristic goes haze method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105005966B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106954022B (en) * | 2017-03-08 | 2019-10-25 | Oppo广东移动通信有限公司 | Image processing method, device and terminal |
| CN106998445A (en) * | 2017-03-28 | 2017-08-01 | 深圳市科美集成电路有限公司 | Haze penetrates camera circuit and system |
| CN109361907A (en) * | 2018-11-16 | 2019-02-19 | 深圳市趣创科技有限公司 | Method, apparatus, terminal and the storage medium of camera shooting night scene sky color |
| CN109919859B (en) * | 2019-01-25 | 2021-09-07 | 暨南大学 | Outdoor scene image defogging enhancement method, computing device and storage medium thereof |
| CN112330559B (en) * | 2020-11-05 | 2022-03-04 | 山东交通学院 | Early warning method for image information recovery and lane keeping of severe foggy roads |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7409082B2 (en) * | 1999-08-18 | 2008-08-05 | Fujifilm Corporation | Method, apparatus, and recording medium for processing image data to obtain color-balance adjusted image data based on white-balance adjusted image data |
| CN101908210A (en) * | 2010-08-13 | 2010-12-08 | 北京工业大学 | Method and system for color image defogging treatment |
| CN103150708A (en) * | 2013-01-18 | 2013-06-12 | 上海交通大学 | Image quick defogging optimized method based on black channel |
| 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 |
-
2015
- 2015-01-16 CN CN201510023351.9A patent/CN105005966B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7409082B2 (en) * | 1999-08-18 | 2008-08-05 | Fujifilm Corporation | Method, apparatus, and recording medium for processing image data to obtain color-balance adjusted image data based on white-balance adjusted image data |
| CN101908210A (en) * | 2010-08-13 | 2010-12-08 | 北京工业大学 | Method and system for color image defogging treatment |
| CN103150708A (en) * | 2013-01-18 | 2013-06-12 | 上海交通大学 | Image quick defogging optimized method based on black channel |
| 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 |
Non-Patent Citations (4)
| Title |
|---|
| Automatic Correction of Saturated Regions in Photographs using Cross-channel Correlation;SyedZ.Masood et al.;《Pacific Graphics 2009》;20091231;论文第1-9页 * |
| Image Dehazing Based on Haziness Analysis;Fan Guo et al.;《International Journal of Automation and Computing》;20140228;第11卷(第1期);第78-86页 * |
| Physics-based Fast Single Image Fog Removal;Jing Yu et al.;《ICS 2010 Proceedings》;20101231;第1048-1052页 * |
| 雾天降质图像的增强复原算法研究;胡学友;《中国博士学位论文全文数据库 信息科技辑》;20120315;正文全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105005966A (en) | 2015-10-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105005966B (en) | A kind of single image based on yellow haze physical characteristic goes haze method | |
| CN106056559A (en) | Dark-channel-prior-method-based non-uniform-light-field underwater target detection image enhancement method | |
| CN104063853B (en) | Method for improving traffic video image definition based on dark channel technology | |
| CN103996382B (en) | Improve the method and system of RGBW image saturation | |
| CN105374013B (en) | Image visibility restorative procedure and image processing apparatus based on double dark channel priors | |
| CN102982514B (en) | The colour-image reinforcing method under water that a kind of dark primary and white balance combine | |
| CN106485681B (en) | Underwater color image restoration method based on color correction and red channel prior | |
| TW201120814A (en) | Method for determining if an input image is a foggy image, method for determining a foggy level of an input image and cleaning method for foggy images | |
| CN105872510B (en) | A kind of image white balancing treatment method and device | |
| CN107403418A (en) | Defogging and the underwater picture Enhancement Method of color correction are carried out based on passage transmissivity | |
| CN107451966A (en) | A kind of real-time video defogging method realized using gray-scale map guiding filtering | |
| CN107316278A (en) | A kind of underwater picture clearness processing method | |
| CN107240079A (en) | A kind of road surface crack detection method based on image procossing | |
| CN106408526B (en) | A kind of visibility detecting method based on multilayer polar plot | |
| CN107481202A (en) | A kind of method of dynamic range of images enhancing | |
| WO2014084199A1 (en) | Image pickup device | |
| CN102436643A (en) | Image defogging method facing to atmospheric scattering proximity effect | |
| CN106384117A (en) | Vehicle color recognition method and device | |
| CN105957023B (en) | A kind of enhancing of laser rays stripe pattern and denoising method based on chroma space | |
| CN107895357A (en) | A kind of real-time water surface thick fog scene image Enhancement Method based on FPGA | |
| CN106782385A (en) | A kind of light-dimming method and its device for wearing display device | |
| CN105959510A (en) | Video rapid defogging method | |
| CN113327675A (en) | Retina blue light radiation damage evaluation method and device, computer equipment and medium | |
| CN103268472B (en) | Lip detection method based on double-colored color space | |
| CN107977941A (en) | A kind of bright areas color fidelity and the image defogging method of contrast enhancing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |