CN109584182A - A kind of image processing method and system - Google Patents
A kind of image processing method and system Download PDFInfo
- Publication number
- CN109584182A CN109584182A CN201811467759.5A CN201811467759A CN109584182A CN 109584182 A CN109584182 A CN 109584182A CN 201811467759 A CN201811467759 A CN 201811467759A CN 109584182 A CN109584182 A CN 109584182A
- Authority
- CN
- China
- Prior art keywords
- image
- pixel
- original image
- optimum
- dark areas
- 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- Withdrawn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
- 238000003672 processing method Methods 0.000 title claims abstract description 13
- 238000001914 filtration Methods 0.000 claims abstract description 31
- 230000002146 bilateral effect Effects 0.000 claims abstract description 27
- 238000000034 method Methods 0.000 claims abstract description 25
- 238000005286 illumination Methods 0.000 claims abstract description 23
- 238000012545 processing Methods 0.000 claims description 14
- 230000000007 visual effect Effects 0.000 claims description 9
- 230000004927 fusion Effects 0.000 claims description 8
- 230000011218 segmentation Effects 0.000 claims description 7
- 238000012795 verification Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 15
- 230000009466 transformation Effects 0.000 description 8
- 230000000694 effects Effects 0.000 description 5
- 230000002708 enhancing effect Effects 0.000 description 5
- 230000003014 reinforcing effect Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000009467 reduction Effects 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000007500 overflow downdraw method Methods 0.000 description 2
- 238000011426 transformation method Methods 0.000 description 2
- 230000000903 blocking effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000007499 fusion processing Methods 0.000 description 1
- 238000002156 mixing Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/80—Geometric correction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- 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
- G06T2207/20024—Filtering details
- G06T2207/20028—Bilateral filtering
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
Abstract
The embodiment of the invention discloses a kind of image processing method and systems, this method comprises: obtaining original image；Bilateral filtering algorithm is weighted using Gauss, the brightness of each pixel in the original image is adjusted, so that the luminance difference of each pixel in the original image is less than specified threshold；Output adjust after original image, by the above method, be conducive to improve intensity of illumination it is uneven caused by halation phenomenon, and then can obtain clearly and substantially without the image border of halation.
Description
Technical field
The present invention relates to field of computer technology, and in particular to a kind of image processing method and system.
Background technique
One of the main path of information is obtained as the mankind, digital picture is widely used in aeronautical and space technology, life
Object medicine, military public security, video and the various fields such as multimedia technology and e-commerce.However, being illuminated by the light, weather, acquisition
Often there is the problems such as uneven illumination, loss in detail in the restriction of the factors such as equipment, the image actually obtained, seriously affect image letter
The reception and registration of breath, therefore, the quality and visual effect that image is promoted using image enhancement technique have important application value.
In recent years, Retinex theory is increasingly becoming the research hotspot in the fields such as image enhancement, image defogging, however,
Retinex theory follows intensity of illumination consistency hypothesis, it is believed that the variation of the intensity of illumination of image is uniformly that this causes to enhance
Result images afterwards compare strong region in illumination such as edges and generate halation phenomenon, so that enhanced result images
Display effect is poor.
Summary of the invention
The embodiment of the present invention is designed to provide a kind of image processing method and system, ties in the prior art to solve
Fruit image compares strong region in illumination such as edges and leads to the problem of halation phenomenon.
To achieve the above object, the embodiment of the present invention provides a kind of image processing method, comprising:
Obtain original image；
Bilateral filtering algorithm is weighted using Gauss, the brightness of each pixel in the original image is adjusted, with
The luminance difference of each pixel in the original image is set to be less than specified threshold；
Original image after output adjusting.
Optionally, bilateral filtering algorithm is weighted using Gauss, the brightness of each pixel in the original image is carried out
Adjusting includes:
Using following formula, the brightness of each pixel in the original image is adjusted:
Wherein, p indicates filter window size, and S (i, j) is the original image, and L (i, j) is the light image estimated,
α is the distance between each pixel coefficient, and β is the coefficient of intensity of illumination,
(i,j)
For the coordinate of centre of neighbourhood pixel, (m, n) is the coordinate of other pixels in neighborhood, σdFor range difference scale parameter, σlIt is bright
Poor scale parameter is spent, γ=20, d (i, j) are central point pixel value in filter window, other pixels in d (m, n) filter window
Pixel value.
Optionally, before the original image after output is adjusted, the method also includes:
Using following formula, the processing that light image influences visual effect is removed on the original image after adjusting,
To obtain reflected image:
R'(i, j)=logS (i, j)-k × logL (i, j)；
The reflected image is quantized to the image between [0,255] by output；
Wherein, 0 < k < 1.
Optionally, after obtaining reflected image, the method also includes:
Using following formula, image stretch is carried out to the reflected image:
Reflected image after stretching is used into maximum variance between clusters Otsu Threshold Segmentation Algorithm, determines the reflection after stretching
The pixel threshold of image；
The pixel value and the pixel threshold of each pixel of reflected image after stretching are compared, to will be greater than or
It is determined as bright area equal to the region where the pixel of the pixel threshold, will be less than where the pixel of the pixel threshold
Region be determined as dark areas；
Using following formula, the comentropy of the bright area and the comentropy of the dark areas are calculated separately:
Using the maximum informational entropy in the bright area, the optimum image of the bright area is determined, and utilize described dark
Maximum informational entropy in region determines the optimum image of the dark areas；
Using following formula, the optimal figure of optimum image and the dark areas to the original image, the bright area
As being merged:
F (i, j)=wo×do(i,j)+wl×dl(i,j)+wd×dd(i,j)；
Wherein, 1 < λ < 20, R'maxFor the maximum value of the original image, R'minFor the minimum value of the original image, l
Indicate the specified pixel value in [0,255] section, p (l) indicates shared in the reflected image of the specified pixel value after the stretch
Ratio, do(i, j) is the pixel value of the sub-block in the original image, dl(i, j) is in the optimum image of the bright area
The pixel of sub-block, dd(i, j) is the pixel of the sub-block in the optimum image of the dark areas, woFor the weight of the original image,
wlFor the weight of the optimum image of the bright area, wdFor the weight of the optimum image of the dark areas,noFor the sub-block of the original image
Number, nlFor the number of the sub-block of the optimum image of the bright area, ndFor of the sub-block of the optimum image of the dark areas
Number.
Optionally, the method also includes:
Consistency desired result is carried out to fused image.
To achieve the above object, the embodiment of the present invention provides a kind of image processing system, comprising:
Acquiring unit, for obtaining original image；
Adjust unit, bilateral filtering algorithm weighted using Gauss, to the brightness of each pixel in the original image into
Row is adjusted, so that the luminance difference of each pixel in the original image is less than specified threshold；
Output unit, for exporting the original image after adjusting.
Optionally, it is used to weight bilateral filtering algorithm using Gauss in the adjusting unit, in the original image
When the brightness of each pixel is adjusted, it is used for:
Using following formula, the brightness of each pixel in the original image is adjusted:
Wherein, p indicates filter window size, and S (i, j) is the original image, and L (i, j) is the light image estimated,
α is the distance between each pixel coefficient, and β is the coefficient of intensity of illumination,
(i,
It j) is the coordinate of centre of neighbourhood pixel, (m, n) is coordinate of other pixels in neighborhood, σdFor range difference scale parameter, σlFor
Luminance difference scale parameter, γ=20, d (i, j) are central point pixel value in filter window, other pixels in d (m, n) filter window
The pixel value of point.
Optionally, the output unit is also used to before the original image after output is adjusted, right using following formula
Original image after adjusting is removed the processing that light image influences visual effect, to obtain reflected image:
R'(i, j)=logS (i, j)-k × logL (i, j)；
And the reflected image is quantized to the image between [0,255] by output；
Wherein, 0 < k < 1.
Optionally, the system also includes:
Draw unit, using following formula, carries out image drawing to the reflected image after obtaining reflected image
It stretches:
First determination unit is calculated for the reflected image after stretching using maximum variance between clusters Otsu Threshold segmentation
Method determines the pixel threshold of the reflected image after stretching；
Comparing unit compares for the pixel value of each pixel to the reflected image after stretching and the pixel threshold
Compared with being determined as bright area with the region where the pixel above or equal to the pixel threshold, the pixel threshold will be less than
Region where the pixel of value is determined as dark areas；
Computing unit calculates separately the comentropy of the bright area and the letter of the dark areas for utilizing following formula
Cease entropy:
Second determination unit, for determining the optimal figure of the bright area using the maximum informational entropy in the bright area
Picture, and using the maximum informational entropy in the dark areas, determine the optimum image of the dark areas；
Integrated unit, for utilizing following formula, optimum image to the original image, the bright area and described dark
The optimum image in region is merged:
F (i, j)=wo×do(i,j)+wl×dl(i,j)+wd×dd(i,j)；
Wherein, 1 < λ < 20, R'maxFor the maximum value of the original image, R'minFor the minimum value of the original image, l
Indicate the specified pixel value in [0,255] section, p (l) indicates shared in the reflected image of the specified pixel value after the stretch
Ratio, do(i, j) is the pixel value of the sub-block in the original image, dl(i, j) is in the optimum image of the bright area
The pixel of sub-block, dd(i, j) is the pixel of the sub-block in the optimum image of the dark areas, woFor the weight of the original image,
wlFor the weight of the optimum image of the bright area, wdFor the weight of the optimum image of the dark areas,noFor the sub-block of the original image
Number, nlFor the number of the sub-block of the optimum image of the bright area, ndFor of the sub-block of the optimum image of the dark areas
Number.
Optionally, the system also includes:
Verification unit, for carrying out consistency desired result to fused image.
The embodiment of the present invention has the advantages that
In embodiments of the present invention, original image, can be advantageous after weighting bilateral filtering algorithm process by Gauss
The luminance difference between each pixel in reduction original image, and the difference is less than specified threshold, that is, the original after adjusting
The luminance difference between each pixel in beginning image is smaller, thus be conducive to improve intensity of illumination it is uneven caused by
Halation phenomenon, and then can obtain clearly and substantially without the image border of halation.
Detailed description of the invention
Fig. 1 is a kind of flow diagram for image processing method that the embodiment of the present invention 1 provides；
Fig. 2 is a kind of flow diagram for image processing method that the embodiment of the present invention 2 provides；
Fig. 3 is a kind of flow diagram for image processing method that the embodiment of the present invention 3 provides；
Fig. 4 is a kind of image block schematic diagram that the embodiment of the present invention 3 provides；
Fig. 5 is a kind of structural schematic diagram for image processing system that the embodiment of the present invention 4 provides；
Fig. 6 is the structural schematic diagram for another image processing system that the embodiment of the present invention 4 provides；
Fig. 7 is the structural schematic diagram for another image processing system that the embodiment of the present invention 4 provides.
Specific embodiment
Embodiments of the present invention are illustrated by particular specific embodiment below, those skilled in the art can be by this explanation
Content disclosed by book is understood other advantages and efficacy of the present invention easily.
Embodiment 1
Fig. 1 is a kind of flow diagram for image processing method that the embodiment of the present invention 1 provides, as shown in Figure 1, this method
The following steps are included:
101, original image is obtained.
102, bilateral filtering algorithm is weighted using Gauss, the brightness of each pixel in the original image is adjusted
Section, so that the luminance difference of each pixel in the original image is less than specified threshold.
103, the original image after output is adjusted.
Specifically, in the prior art, traditional single scale Retinex algorithm is easy to produce halation phenomenon, and this is insufficient,
Bilateral filtering is weighted to estimate that light image, traditional single scale Retinex image enhancement are calculated using Gauss in the embodiment of the present invention
Based on method is assumed by illumination consistency, light image is estimated using Gauss model.However in practical applications, the illumination of image
Intensity is not often consistent.Therefore, enhanced image is easy to produce halation in the strong borderline region of illumination variation and shows
As.Bilateral filtering is a kind of innovatory algorithm proposed on the basis of gaussian filtering, it is increased on the basis of gaussian filtering
Parameter relevant to intensity of illumination.Bilateral filtering is increased on the basis of gaussian filtering between luminance difference degree pixel
Consider.Bilateral filtering is introduced into traditional single scale Retinex algorithm, can be effectively improved since intensity of illumination is uneven
Caused halation phenomenon.It is a kind of innovatory algorithm proposed on the basis of bilateral filtering that Gauss, which weights bilateral filtering, it is logical
Cross improve bilateral filtering in parameter relevant to intensity of illumination come achieve the purpose that adjust pixel between luminance difference degree.
In embodiments of the present invention, original image, can be advantageous after weighting bilateral filtering algorithm process by Gauss
The luminance difference between each pixel in reduction original image, and the difference is less than specified threshold, that is, the original after adjusting
The luminance difference between each pixel in beginning image is smaller, thus be conducive to improve intensity of illumination it is uneven caused by
Halation phenomenon, and then can obtain clearly and substantially without the image border of halation.
It, can be by following formula, to the original graph when executing step 102 in a feasible embodiment
The brightness of each pixel as in is adjusted:
Wherein, p indicates filter window size, and S (i, j) is the original image, and L (i, j) is the light image estimated,
α is the distance between each pixel coefficient, and β is the coefficient of intensity of illumination,(i, j) is neighborhood
The coordinate of central pixel point, (m, n) are the coordinate of other pixels in neighborhood, σdFor range difference scale parameter, σlFor luminance difference ruler
Parameter is spent, γ=20, d (i, j) are central point pixel value in filter window, the pixel of other pixels in d (m, n) filter window
Value.
Specifically, the value of parameter γ will affect the effect of image filtering, it, cannot effective reduced image when γ value is too small
Luminance difference between vegetarian refreshments, filter effect are equal to traditional bilateral filtering；When γ value is larger, will increase bright between pixel
Difference is spent, halation phenomenon is caused.Many experiments show as γ=20, weight bilateral filtering using Gauss and estimate light image
The inconsistent caused halation phenomenon of intensity of illumination can be effectively eliminated, to obtain the clear, image border without halation.
Embodiment 2
Fig. 2 is a kind of flow diagram for image processing method that the embodiment of the present invention 2 provides, as shown in Fig. 2, executing
When step 103, it can be realized by following steps:
201, using following formula, the place that light image influences visual effect is removed on the original image after adjusting
Reason, to obtain reflected image:
R'(i, j)=logS (i, j)-k × logL (i, j).
Wherein, 0 < k < 1.
202, the reflected image is quantized to the image between [0,255] by output.
Specifically, removing illumination pattern according to Retinex theory after estimating light image using Gauss weighting bilateral filtering
As the influence to visual effect, so that it may the reflected image of reflection image essential attribute is obtained, to make image more naturally, introducing
Parameter k (0 < k < 1), finally, obtaining result images in image quantization to [0,255] section.
Embodiment 3
Fig. 3 is a kind of flow diagram for image processing method that the embodiment of the present invention 3 provides, as shown in figure 3, obtaining
After reflected image, this method is further comprising the steps of:
301, using following formula, image stretch is carried out to the reflected image:
302, the reflected image after stretching is used into maximum variance between clusters Otsu Threshold Segmentation Algorithm, determined after stretching
The pixel threshold of reflected image.
303, the pixel value and the pixel threshold of each pixel of the reflected image after stretching are compared, it will be big
It is determined as bright area in or equal to region where the pixel of the pixel threshold, the pixel of the pixel threshold will be less than
The region at place is determined as dark areas.
304, using following formula, the comentropy of the bright area and the comentropy of the dark areas are calculated separately:
305, it using the maximum informational entropy in the bright area, determines the optimum image of the bright area, and utilizes institute
The maximum informational entropy in dark areas is stated, determines the optimum image of the dark areas.
306, using following formula, optimum image and the dark areas to the original image, the bright area are most
Excellent image is merged:
F (i, j)=wo×do(i,j)+wl×dl(i,j)+wd×dd(i,j)。
Wherein, 1 < λ < 20, R'maxFor the maximum value of the original image, R'minFor the minimum value of the original image, l
Indicate the specified pixel value in [0,255] section, p (l) indicates shared in the reflected image of the specified pixel value after the stretch
Ratio, do(i, j) is the pixel value of the sub-block in the original image, dl(i, j) is in the optimum image of the bright area
The pixel of sub-block, dd(i, j) is the pixel of the sub-block in the optimum image of the dark areas, woFor the weight of the original image,
wlFor the weight of the optimum image of the bright area, wdFor the weight of the optimum image of the dark areas, noFor the sub-block of the original image
Number, nlFor the number of the sub-block of the optimum image of the bright area, ndFor of the sub-block of the optimum image of the dark areas
Number.
Specifically, weighting bilateral filtering using Gauss to estimate that light image can be efficiently against halation phenomenon.However,
Since single scale Retinex algorithm only carries out enhancing processing to image in the overall situation, the part for having ignored image different zones is special
Sign, in enhanced image, dark areas details is been significantly enhanced, and bright area contrast declines, image graying, to make figure
The bright of picture, dark areas are all enhanced to the greatest extent, and the thought of image co-registration is proposed in the embodiment of the present invention, passes through fusion office
The method of portion's optimum image obtains global optimum's image of high quality.How to obtain the local optimum image of original image is this
Carve the matter of utmost importance to be solved.
For the deficiency of the image graying of single scale Retinex algorithm, proposition quantifies to reflect using gray-scale transformation method
Image, and then enhance the contrast of image.Gray-scale transformation method includes two kinds of linear transformation, nonlinear transformation.Wherein, linear to become
It is simple to change realization, but is often unable to fully meet the requirement of image enhancement.Nonlinear transformation is by neatly adjusting conversion curve
Slope, can the different zones of image be realized with different degrees of enhancing.Common image non-linear transform method includes pair
Transformation of variables, power rate transition etc..The embodiment of the present invention stretches reflected image using nonlinear transformation formula.
When λ value is smaller, for curve close to linear transformation, image stretch degree is relatively uniform, in conjunction with Retinex image
The significant feature of dark areas reinforcing effect, in result images after the stretch, the reinforcing effect of the dark areas of original image is significant, and
Bright area contrast is insufficient, image visual effect is partially grey.With the increase of λ value, R'(x, y) the lesser region stretching after image of value
Element value is close to 0, and R'(x, y) the biggish region curve slope of value is larger, and the reinforcing effect of bright area is significant in result images,
But the details missing of dark areas is more.Therefore, when λ value difference, image is bright, dark areas obtains different degrees of enhancing respectively
Effect.The embodiment of the present invention is optimal to obtain bright area optimum image and dark areas by adjusting the mode of nonlinear transformation parameter
Image, and then global optimum's image is obtained by Image Fusion.
During local optimum image obtains, main there are two affecting parameters, respectively R'(i, j)=logS (i, j)-k ×
Gain parameter k in logL (i, j) and
In nonlinear transformation parameter lambda, can reasonably choose the two parameters, for image enhancement final result have to close
Important role.
The embodiment of the present invention chooses bright area optimized parameter k using gradient ascent algorithmmaxl、λmaxlIt is optimal with dark areas
Parameter kmaxd、λmaxd.During seeking optimized parameter, using information entropy maximization as the basis for selecting of parameter.Comentropy is used
The abundant degree of information content included in image is measured, the comentropy the big, the information for showing that image is included is abundanter,
Image detail reinforcing effect is better.
The step of embodiment of the present invention obtains local optimum parameter using gradient ascent algorithm is as follows:
1) optimized parameter is initialized.Parametric variable: k=0.1, λ=1；Bright area optimized parameter: kbestl=0.1, λbestl
=1；Dark areas optimized parameter: kbestd=0.1, λbestd=1；Bright, dark areas comentropy variable: el=0, ed=0；It is bright, dark
Region maximum informational entropy: emaxl=0, emaxd=0；Parameter increase: Δ k=0.1, Δ λ=1.
2) result images under parameter current are acquired.Parametric variable k, λ are substituted into formula R'(i, j respectively)=logS (i,
J)-k × logL (i, j) andWith
Acquire the result images under parameter current.
3) bright, the dark areas of definitive result image.The threshold value of image is obtained using Otsu Threshold Segmentation Algorithm, and is sentenced according to this
Pixel value is less than point of the point of threshold value as dark picture areas domain, pixel value is greater than to the point of threshold value by bright, the dark areas of disconnected image
Point as image bright area.
4) bright, dark areas comentropy is calculated.Pass through formulaCalculate separately out image
The comentropy e of bright arealWith the comentropy e of dark areasd。
5) size of comparison information entropy.Compare elWith emaxl、edWith emaxdSize.If el> emaxl, then emaxl=el、
kbestl=k, λbestl=λ.Otherwise emaxl、kbestl、λbestlValue remain unchanged.E similarly can be obtainedmaxd、kbestd、λbestdValue.
6) increase the value of k and λ according to gradient ascent algorithm.λ=λ+Δ λ, k=k+ Δ k is set.It repeats step 2 and arrives step
5, until parameter k and λ reach the defined upper limit.Parameter k at this timebestl、λbestlObtain bright area comentropy most
The bright area optimized parameter being worth greatly, kbestd、λbestdDark areas comentropy is as set to obtain the dark areas optimized parameter of maximum value.
Many experiments show optimized parameter λbestl、λbestdValue range be [1,20] be advisable.
After estimating light image using Gauss weighting bilateral filtering, by the above-mentioned local optimum parameter k acquiredbestlIt substitutes into
Formula R'(i, j)=logS (i, j)-k × logL (i, j), unstretched reflected image can be obtained；By local optimum parameter
λbestlSubstitute into formulaIt can be obtained bright
The bright area optimum image of region acquirement maximum informational entropy.Similarly, by parameter kbestd、λbestdThe above-mentioned two public affairs of formula are substituted into respectively
Dark areas optimum image can be obtained in formula.
To avoid excessively enhancing phenomenon, the embodiment of the present invention is in fusion process while sufficiently enhancing image detail
The brightness that original image carrys out adjusted result image is added.Using segment fusion method to original image, bright area optimum image and
Dark areas optimum image carries out fusion treatment.
Fig. 4 is a kind of image block schematic diagram that the embodiment of the present invention 3 provides, and is carried out at identical piecemeal to three width images
Reason, divides the image into the identical square sub-block of size, image block schematic diagram is as shown in Figure 4.Assuming that image size is M × N, son
Block size is r × r.When image can be ignored by whole timesharing, region 2 and region 3, image is divided into (M/r) × (N/r)
Block；When image cannot be handled in region 2 and region 3 as two independent sub-blocks, image is divided by the whole timesharing of square sub-block
+ 2 pieces of [M/r] × [N/r] ([] indicates bracket function).Then, the comentropy for calculating the corresponding sub block of every piece image, takes phase
Answer selected digital image of the maximum image of sub-block information entropy as the sub-block region.Finally each selected sub-block fusion is got up, is obtained
The result images of fused high quality.In result images, each sub-block all has the most abundant detailed information.Using image point
Block fusion method carries out fusion treatment to the optimum image of original image, bright area and the optimum image of dark areas.It is using
In the result images that segment fusion algorithm obtains, each sub-block of image obtains preferable reinforcing effect, and picture contrast is higher.
However, block image blending algorithm is clearly disadvantageous there are one, i.e., in fused result images, there are serious piece of effects
It answers.
In a feasible embodiment, in optimum image to the original image, the bright area and described dark
After the optimum image in region is merged, consistency desired result is carried out to fused image.
Specifically, being handled using consistency desired result method fused for the blocking artifact for solving the problems, such as Image Fusion
Result images.
Embodiment 4
Fig. 5 is a kind of structural schematic diagram for image processing system that the embodiment of the present invention 4 provides, as shown in figure 5, the system
Include:
Acquiring unit 51, for obtaining original image；
Unit 52 is adjusted, bilateral filtering algorithm is weighted using Gauss, the brightness to each pixel in the original image
It is adjusted, so that the luminance difference of each pixel in the original image is less than specified threshold；
Output unit 53, for exporting the original image after adjusting.
In a feasible embodiment, it is used to weight bilateral filtering algorithm using Gauss in the adjusting unit 52,
When the brightness of each pixel in the original image is adjusted, it is used for:
Using following formula, the brightness of each pixel in the original image is adjusted:
Wherein, p indicates filter window size, and S (i, j) is the original image, and L (i, j) is the light image estimated,
α is the distance between each pixel coefficient, and β is the coefficient of intensity of illumination,
(i,
It j) is the coordinate of centre of neighbourhood pixel, (m, n) is coordinate of other pixels in neighborhood, σdFor range difference scale parameter, σlFor
Luminance difference scale parameter, γ=20, d (i, j) are central point pixel value in filter window, other pixels in d (m, n) filter window
The pixel value of point.
In a feasible embodiment, the output unit 51, be also used to output adjust after original image it
Before, using following formula, the processing that light image influences visual effect is removed on the original image after adjusting, to obtain
Reflected image:
R'(i, j)=logS (i, j)-k × logL (i, j)；
And the reflected image is quantized to the image between [0,255] by output；
Wherein, 0 < k < 1.
In a feasible embodiment, Fig. 6 is the knot for another image processing system that the embodiment of the present invention 4 provides
Structure schematic diagram, as shown in fig. 6, the system further include:
Draw unit 54, using following formula, carries out image drawing to the reflected image after obtaining reflected image
It stretches:
First determination unit 55 is calculated for the reflected image after stretching using maximum variance between clusters Otsu Threshold segmentation
Method determines the pixel threshold of the reflected image after stretching；
Comparing unit 56 is carried out for the pixel value of each pixel to the reflected image after stretching and the pixel threshold
Compare, bright area is determined as with the region where the pixel above or equal to the pixel threshold, the pixel will be less than
Region where the pixel of threshold value is determined as dark areas；
Computing unit 57 calculates separately the comentropy and the dark areas of the bright area for utilizing following formula
Comentropy:
Second determination unit 58, for determining the optimal of the bright area using the maximum informational entropy in the bright area
Image, and using the maximum informational entropy in the dark areas, determine the optimum image of the dark areas；
Integrated unit 59, for utilizing following formula, optimum image to the original image, the bright area and described
The optimum image of dark areas is merged:
F (i, j)=wo×do(i,j)+wl×dl(i,j)+wd×dd(i,j)；
Wherein, 1 < λ < 20, R'maxFor the maximum value of the original image, R'minFor the minimum value of the original image, l
Indicate the specified pixel value in [0,255] section, p (l) indicates shared in the reflected image of the specified pixel value after the stretch
Ratio, do(i, j) is the pixel value of the sub-block in the original image, dl(i, j) is in the optimum image of the bright area
The pixel of sub-block, dd(i, j) is the pixel of the sub-block in the optimum image of the dark areas, woFor the weight of the original image,
wlFor the weight of the optimum image of the bright area, wdFor the weight of the optimum image of the dark areas,noFor the sub-block of the original image
Number, nlFor the number of the sub-block of the optimum image of the bright area, ndFor of the sub-block of the optimum image of the dark areas
Number.
In a feasible embodiment, Fig. 7 is the knot for another image processing system that the embodiment of the present invention 4 provides
Structure schematic diagram, as shown in fig. 7, the system further include:
Verification unit 60, for carrying out consistency desired result to fused image.
It should be noted that can refer to embodiment one and embodiment three about being discussed in detail for page jump system, herein
No longer it is described in detail.
In embodiments of the present invention, original image, can be advantageous after weighting bilateral filtering algorithm process by Gauss
The luminance difference between each pixel in reduction original image, and the difference is less than specified threshold, that is, the original after adjusting
The luminance difference between each pixel in beginning image is smaller, thus be conducive to improve intensity of illumination it is uneven caused by
Halation phenomenon, and then can obtain clearly and substantially without the image border of halation.
Although above having used general explanation and specific embodiment, the present invention is described in detail, at this
On the basis of invention, it can be made some modifications or improvements, this will be apparent to those skilled in the art.Therefore,
These modifications or improvements without departing from theon the basis of the spirit of the present invention are fallen within the scope of the claimed invention.
Claims (10)
1. a kind of image processing method characterized by comprising
Obtain original image；
Bilateral filtering algorithm is weighted using Gauss, the brightness of each pixel in the original image is adjusted, so that institute
The luminance difference for stating each pixel in original image is less than specified threshold；
Original image after output adjusting.
2. the method as described in claim 1, which is characterized in that bilateral filtering algorithm is weighted using Gauss, to the original graph
The brightness of each pixel as in, which is adjusted, includes:
Using following formula, the brightness of each pixel in the original image is adjusted:
Wherein, p indicates filter window size, and S (i, j) is the original image, and L (i, j) is the light image estimated, and α is
The distance between each pixel coefficient, β are the coefficient of intensity of illumination,(i, j) is in neighborhood
The coordinate of imago vegetarian refreshments, (m, n) are the coordinate of other pixels in neighborhood, σdFor range difference scale parameter, σlFor luminance difference scale
Parameter, γ=20, d (i, j) are central point pixel value in filter window, the pixel of other pixels in d (m, n) filter window
Value.
3. the method as described in claim 1, which is characterized in that before the original image after output is adjusted, the method is also
Include:
Using following formula, the processing that light image influences visual effect is removed on the original image after adjusting, with
To reflected image:
R'(i, j)=logS (i, j)-k × logL (i, j)；
The reflected image is quantized to the image between [0,255] by output；
Wherein, 0 < k < 1.
4. method as claimed in claim 3, which is characterized in that after obtaining reflected image, the method also includes:
Using following formula, image stretch is carried out to the reflected image:
Reflected image after stretching is used into maximum variance between clusters Otsu Threshold Segmentation Algorithm, determines the reflected image after stretching
Pixel threshold；
The pixel value and the pixel threshold of each pixel of reflected image after stretching are compared, with above or equal to
Region where the pixel of the pixel threshold is determined as bright area, by the area where the pixel for being less than the pixel threshold
Domain is determined as dark areas；
Using following formula, the comentropy of the bright area and the comentropy of the dark areas are calculated separately:
Using the maximum informational entropy in the bright area, the optimum image of the bright area is determined, and utilize the dark areas
In maximum informational entropy, determine the optimum image of the dark areas；
Using following formula, the optimum image of optimum image and the dark areas to the original image, the bright area into
Row fusion:
F (i, j)=wo×do(i,j)+wl×dl(i,j)+wd×dd(i,j)；
Wherein, 1 < λ < 20, R'maxFor the maximum value of the original image, R'minFor the minimum value of the original image, l is indicated
Specified pixel value in [0,255] section, p (l) indicate ratio shared in the reflected image of the specified pixel value after the stretch
Example, do(i, j) is the pixel value of the sub-block in the original image, dl(i, j) is the sub-block in the optimum image of the bright area
Pixel, dd(i, j) is the pixel of the sub-block in the optimum image of the dark areas, woFor the weight of the original image, wlFor
The weight of the optimum image of the bright area, wdFor the weight of the optimum image of the dark areas,noFor the sub-block of the original image
Number, nlFor the number of the sub-block of the optimum image of the bright area, ndFor of the sub-block of the optimum image of the dark areas
Number.
5. method as claimed in claim 4, which is characterized in that the method also includes:
Consistency desired result is carried out to fused image.
6. a kind of image processing system characterized by comprising
Acquiring unit, for obtaining original image；
Unit is adjusted, bilateral filtering algorithm is weighted using Gauss, the brightness of each pixel in the original image is adjusted
Section, so that the luminance difference of each pixel in the original image is less than specified threshold；
Output unit, for exporting the original image after adjusting.
7. system as claimed in claim 6, which is characterized in that be used to weight bilateral filtering using Gauss in the adjusting unit
Algorithm is used for when the brightness of each pixel in the original image is adjusted:
Using following formula, the brightness of each pixel in the original image is adjusted:
Wherein, p indicates filter window size, and S (i, j) is the original image, and L (i, j) is the light image estimated, and α is
The distance between each pixel coefficient, β are the coefficient of intensity of illumination,(i, j) is neighborhood
The coordinate of central pixel point, (m, n) are the coordinate of other pixels in neighborhood, σdFor range difference scale parameter, σlFor luminance difference ruler
Parameter is spent, γ=20, d (i, j) are central point pixel value in filter window, the pixel of other pixels in d (m, n) filter window
Value.
8. system as claimed in claim 6, which is characterized in that the output unit is also used to original after output is adjusted
Before image, using following formula, the processing that light image influences visual effect is removed on the original image after adjusting,
To obtain reflected image:
R'(i, j)=logS (i, j)-k × logL (i, j)；
And the reflected image is quantized to the image between [0,255] by output；
Wherein, 0 < k < 1.
9. system as claimed in claim 8, which is characterized in that the system also includes:
Draw unit, using following formula, carries out image stretch to the reflected image after obtaining reflected image:
First determination unit uses maximum variance between clusters Otsu Threshold Segmentation Algorithm for the reflected image after stretching, really
The pixel threshold of reflected image after fixed stretching；
Comparing unit is compared for the pixel value of each pixel to the reflected image after stretching and the pixel threshold,
It is determined as bright area with the region where the pixel above or equal to the pixel threshold, the pixel threshold will be less than
Region where pixel is determined as dark areas；
Computing unit, for calculating separately the comentropy of the bright area and the comentropy of the dark areas using following formula:
Second determination unit, for determining the optimum image of the bright area using the maximum informational entropy in the bright area, with
And using the maximum informational entropy in the dark areas, determine the optimum image of the dark areas；
Integrated unit, for utilizing following formula, optimum image and the dark areas to the original image, the bright area
Optimum image merged:
F (i, j)=wo×do(i,j)+wl×dl(i,j)+wd×dd(i,j)；
Wherein, 1 < λ < 20, R'maxFor the maximum value of the original image, R'minFor the minimum value of the original image, l is indicated
Specified pixel value in [0,255] section, p (l) indicate ratio shared in the reflected image of the specified pixel value after the stretch
Example, do(i, j) is the pixel value of the sub-block in the original image, dl(i, j) is the sub-block in the optimum image of the bright area
Pixel, dd(i, j) is the pixel of the sub-block in the optimum image of the dark areas, woFor the weight of the original image, wlFor
The weight of the optimum image of the bright area, wdFor the weight of the optimum image of the dark areas,noFor the sub-block of the original image
Number, nlFor the number of the sub-block of the optimum image of the bright area, ndFor of the sub-block of the optimum image of the dark areas
Number.
10. system as claimed in claim 9, which is characterized in that the system also includes:
Verification unit, for carrying out consistency desired result to fused image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811467759.5A CN109584182A (en) | 2018-12-03 | 2018-12-03 | A kind of image processing method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811467759.5A CN109584182A (en) | 2018-12-03 | 2018-12-03 | A kind of image processing method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109584182A true CN109584182A (en) | 2019-04-05 |
Family
ID=65926206
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811467759.5A Withdrawn CN109584182A (en) | 2018-12-03 | 2018-12-03 | A kind of image processing method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109584182A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110188639A (en) * | 2019-05-20 | 2019-08-30 | 深圳供电局有限公司 | Face image processing process and its system, computer equipment, readable storage medium storing program for executing |
| CN110363720A (en) * | 2019-07-04 | 2019-10-22 | 北京奇艺世纪科技有限公司 | A kind of color enhancement method, apparatus, equipment and the storage medium of image |
| CN112006651A (en) * | 2020-09-10 | 2020-12-01 | 孙礼华 | Cataract surgery auxiliary diagnosis system and method thereof |
| WO2021212440A1 (en) * | 2020-04-23 | 2021-10-28 | 华为技术有限公司 | Image encoding/decoding method and apparatus |
| CN116993643A (en) * | 2023-09-27 | 2023-11-03 | 山东建筑大学 | Unmanned aerial vehicle photogrammetry image correction method based on artificial intelligence |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103617652A (en) * | 2013-11-02 | 2014-03-05 | 西北农林科技大学 | Visual saliency-based bas-relief generating method |
| US20150324961A1 (en) * | 2013-01-23 | 2015-11-12 | Tencent Technology (Shenzhen) Co., Ltd. | Method and apparatus for adjusting image brightness |
| CN105654433A (en) * | 2015-12-28 | 2016-06-08 | 桂林电子科技大学 | Color image enhancement method based on improved multi-scale Retinex |
-
2018
- 2018-12-03 CN CN201811467759.5A patent/CN109584182A/en not_active Withdrawn
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150324961A1 (en) * | 2013-01-23 | 2015-11-12 | Tencent Technology (Shenzhen) Co., Ltd. | Method and apparatus for adjusting image brightness |
| CN103617652A (en) * | 2013-11-02 | 2014-03-05 | 西北农林科技大学 | Visual saliency-based bas-relief generating method |
| CN105654433A (en) * | 2015-12-28 | 2016-06-08 | 桂林电子科技大学 | Color image enhancement method based on improved multi-scale Retinex |
Non-Patent Citations (1)
| Title |
|---|
| 常戬等: "基于图像融合技术的Retinex图像增强算法", 《计算机工程与科学》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110188639A (en) * | 2019-05-20 | 2019-08-30 | 深圳供电局有限公司 | Face image processing process and its system, computer equipment, readable storage medium storing program for executing |
| CN110363720A (en) * | 2019-07-04 | 2019-10-22 | 北京奇艺世纪科技有限公司 | A kind of color enhancement method, apparatus, equipment and the storage medium of image |
| WO2021212440A1 (en) * | 2020-04-23 | 2021-10-28 | 华为技术有限公司 | Image encoding/decoding method and apparatus |
| CN112006651A (en) * | 2020-09-10 | 2020-12-01 | 孙礼华 | Cataract surgery auxiliary diagnosis system and method thereof |
| CN116993643A (en) * | 2023-09-27 | 2023-11-03 | 山东建筑大学 | Unmanned aerial vehicle photogrammetry image correction method based on artificial intelligence |
| CN116993643B (en) * | 2023-09-27 | 2023-12-12 | 山东建筑大学 | Unmanned aerial vehicle photogrammetry image correction method based on artificial intelligence |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109584182A (en) | A kind of image processing method and system | |
| Ghani et al. | Enhancement of low quality underwater image through integrated global and local contrast correction | |
| Zhang et al. | Color correction and adaptive contrast enhancement for underwater image enhancement | |
| Wang et al. | Simple low-light image enhancement based on Weber–Fechner law in logarithmic space | |
| Huang et al. | Image contrast enhancement for preserving mean brightness without losing image features | |
| Wang et al. | Biologically inspired image enhancement based on Retinex | |
| Kaur et al. | Survey of contrast enhancement techniques based on histogram equalization | |
| CN107392866B (en) | A kind of facial image local grain Enhancement Method of illumination robust | |
| CN105931201B (en) | A kind of image subjective vision effect reinforcing method based on wavelet transformation | |
| CN104463804B (en) | Image enhancement method based on intuitional fuzzy set | |
| CN105654433A (en) | Color image enhancement method based on improved multi-scale Retinex | |
| CN108288258A (en) | A kind of low-quality images Enhancement Method under severe weather conditions | |
| CN111968041A (en) | Self-adaptive image enhancement method | |
| CN104268843A (en) | Image self-adaptation enhancing method based on histogram modification | |
| CN105303561A (en) | Image preprocessing grayscale space division method | |
| CN105225238A (en) | A kind of gray space division methods of the Image semantic classification based on mean filter | |
| CN111968065A (en) | Self-adaptive enhancement method for image with uneven brightness | |
| CN104166969B (en) | Digital image enhancement method and system | |
| Sun et al. | Low‐Light Image Enhancement Based on Guided Image Filtering in Gradient Domain | |
| CN109978789A (en) | A kind of image enchancing method based on Retinex algorithm and guiding filtering | |
| CN104616259B (en) | A kind of adaptive non-local mean image de-noising method of noise intensity | |
| CN108648160B (en) | Underwater sea cucumber image defogging enhancement method and system | |
| CN108550124B (en) | Illumination compensation and image enhancement method based on bionic spiral | |
| Bhandari et al. | Multi-exposure optimized contrast and brightness balance color image enhancement | |
| Zhu et al. | Image enhancement using fuzzy intensity measure and adaptive clipping histogram equalization |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20190424 Address after: 125000 Liaoning University of Engineering and Technology, 188 Longwan South Street, Xingcheng City, Huludao City, Liaoning Province Applicant after: LIAONING TECHNICAL University Address before: 112000 Longyuan Street, Wolong Xinyuan, Yinzhou District, Tieling City, Liaoning Province, 29 Applicant before: Dong Yuli |
|
| TA01 | Transfer of patent application right | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20190405 |
|
| WW01 | Invention patent application withdrawn after publication |