CN107085833A - Remote sensing images filtering method based on the equal intermediate value fusion of gradient self-adaptive switch reciprocal - Google Patents
Remote sensing images filtering method based on the equal intermediate value fusion of gradient self-adaptive switch reciprocal Download PDFInfo
- Publication number
- CN107085833A CN107085833A CN201710241062.5A CN201710241062A CN107085833A CN 107085833 A CN107085833 A CN 107085833A CN 201710241062 A CN201710241062 A CN 201710241062A CN 107085833 A CN107085833 A CN 107085833A
- Authority
- CN
- China
- Prior art keywords
- noise
- pixel
- intermediate value
- remote sensing
- mrow
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000001914 filtration Methods 0.000 title claims abstract description 50
- 238000000034 method Methods 0.000 title claims abstract description 28
- 230000004927 fusion Effects 0.000 title claims abstract description 13
- 239000006002 Pepper Substances 0.000 claims abstract description 26
- 238000002156 mixing Methods 0.000 claims abstract description 10
- 230000009467 reduction Effects 0.000 claims abstract description 6
- 239000011159 matrix material Substances 0.000 claims description 15
- 238000012545 processing Methods 0.000 claims description 7
- 230000009977 dual effect Effects 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 4
- 238000005070 sampling Methods 0.000 claims description 2
- 238000009499 grossing Methods 0.000 claims 1
- 238000005192 partition Methods 0.000 abstract description 5
- 238000004321 preservation Methods 0.000 abstract 1
- 230000003044 adaptive effect Effects 0.000 description 12
- 230000000694 effects Effects 0.000 description 10
- 230000008901 benefit Effects 0.000 description 6
- 210000001367 artery Anatomy 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 150000003839 salts Chemical class 0.000 description 2
- 210000003462 vein Anatomy 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000001939 inductive effect Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011160 research Methods 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
- G06T5/70—Denoising; Smoothing
-
- 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
- G06T2207/10032—Satellite or aerial image; Remote sensing
-
- 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
- G06T2207/20032—Median filtering
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses the remote sensing images filtering method based on the equal intermediate value fusion of gradient self-adaptive switch reciprocal, step 1：Obtain unmanned aerial vehicle remote sensing image；Step 2：The template that size is n × n is chosen, the Grad in template center's point and template between remaining (n × n 1) pixel and preservation is obtained；Step 3：Grad and threshold value obtained by step 2 is compared, judges the pixel whether as caused by salt-pepper noise or random noise；Step 4：The result drawn according to step 3, if the pixel is caused by salt-pepper noise or random noise, is smoothed with Gradient Inverse Weight algorithm；If the pixel is caused by salt-pepper noise or random noise, denoising is filtered with the equal intermediate value blending algorithm of self-adaptive switch；Step 5：The unmanned aerial vehicle remote sensing image after filtering and noise reduction is obtained by step 4.Instant invention overcomes the limitation of existing remote sensing images filtering algorithm scope of application when in face of polytype partition noise.
Description
Technical field
The invention belongs to technical field of image processing, and in particular to one kind is melted based on the gradient equal intermediate value of self-adaptive switch reciprocal
The remote sensing images filtering method of conjunction.
Background technology
Along with developing rapidly for modern space flight and unmanned air vehicle technique, remote sensing technology achieves what is advanced by leaps and bounds in recent years
Development.Unmanned plane has build compact because of it, and the flexible advantage of scouting mode is used widely, wherein, to unmanned aerial vehicle remote sensing
The analysis and research of image have turned into people and have obtained one of main path of information.
However, remote sensing images are during acquisition and transmission, influenceed inevitable by factors such as sensor and air
Meeting introduce noise, for obtain clearly, high-quality remote sensing images, it is very must that remote sensing images are filtered with denoising
Want.
When being filtered denoising to unmanned aerial vehicle remote sensing image, traditional remote sensing images filtering is both for having determined
Types noise and by noise jamming degree carry out select suitable filtering and noise reduction algorithm, these algorithms are in remote sensing images denoising
Although respectively there is feature, and good filter effect can be obtained, its restricted application has certain limitation, no
It is well positioned to meet the denoising requirement of unmanned aerial vehicle remote sensing image.For unmanned aerial vehicle remote sensing image, due to in-flight easily by appearance
Signal such as is disturbed at the influence of reason in state interference, the intrinsic speciality of sensing equipment, optical aberration, transmitting procedure, therefore typically
Will not can often there is the noise of polytype distribution in image only by single noise pollution, polytype make an uproar is being distributed with
During sound, traditional remote sensing images filtering algorithm cannot obtain preferable denoising effect.
The content of the invention
It is an object of the invention to provide a kind of remote sensing images filter based on the equal intermediate value fusion of gradient self-adaptive switch reciprocal
Wave method, to overcome the defect that above-mentioned prior art is present, the present invention to a certain extent, overcomes existing remote sensing images filtering
The limitation of algorithm scope of application when in face of polytype partition noise.
To reach above-mentioned purpose, the present invention is adopted the following technical scheme that：
Based on the remote sensing images filtering method of the equal intermediate value fusion of gradient self-adaptive switch reciprocal, comprise the following steps：
Step 1：Obtain unmanned aerial vehicle remote sensing image；
Step 2：Choose the template that size is n × n, obtain in template center's point and template remaining (n × n-1) pixel it
Between gradient value matrix q and preserve；
Step 3：Grad and threshold value obtained by step 2 is compared, judges whether the pixel is made an uproar by the spiced salt
Caused by sound or random noise；
Step 4：The result drawn according to step 3, if the pixel is caused by salt-pepper noise or random noise, is used
Gradient Inverse Weight algorithm is smoothed；If the pixel is caused by salt-pepper noise or random noise, with adaptive
The equal intermediate value blending algorithm of inductive switch is filtered denoising；
Step 5：The unmanned aerial vehicle remote sensing image after filtering and noise reduction is obtained by step 4.
Further, template center's point is (x, y) in step 2, and the gray value of template center's point is f (x, y), is obtained respectively
Grad between f (x, y) and template remaining (n × n-1) individual pixel.
Further, given threshold T=f (x, y) × 10% in step 3, if at least there is pixel (x+i, a y+
J) so that | f (x+i, y+j)-f (x, y) |≤T, then it is not caused by salt-pepper noise or random noise to judge pixel (x, y)；
If all pixels point (x+i, y+j) so that | f (x+i, y+j)-f (x, y) | ＞ T, then judging pixel (x, y) is made an uproar by the spiced salt
Caused by sound or random noise.
Further, Gradient Inverse Weight is used when the pixel is caused by salt-pepper noise or random noise in step 4
Algorithm is smoothed, and is specially：
For n × n template, the matrix q that step 2 is obtained is：
Weight matrix w is obtained by matrix q：
If f (x+j, y+j)=f (x, y), then Grad is 0, regulation central element w (x, y)=0.5, remaining n*n-1
Weighted elements sum is 0.5, w each elements summation is equal to 1, then has：
Wherein, it is 0 when i, j=-1,0 or 1, but i, j are different；
Finally, template elements are multiplied with the weights correspondence corresponding to it at each pixel, then sum and be somebody's turn to do
Output g (x, y) of the pixel using gradient inverse after smooth.
Further, the pixel is caused by salt-pepper noise or random noise in step 4, then with self-adaptive switch it is equal in
Value blending algorithm is filtered denoising, is specially：
The equal intermediate value blending algorithm of self-adaptive switch, is switched over using dual threshold, is realized in switch mean filter and switch
The fusion treatment of value filtering, is expressed as follows with mathematical formulae：
Wherein, σ (x, y) is that abscissa is x in image, ordinate for y point grayscale shift value；μ (x, y) is in image
Abscissa is x, and ordinate is the gray scale difference value of y point；F (x, y) is that abscissa is x, and ordinate is the gray value of y point；
Median (x, y) is the intermediate value that abscissa is gray value in x, the neighborhood of a point that ordinate is y；Mean (x, y) is that abscissa is
X, ordinate is y neighborhood of a point average gray；G (x, y) is that the abscissa after the equal medium filtering denoising of switch is
X, ordinate is the gray value of y point；In dual threshold, θ is the threshold value of switch intermediate value；λ is the threshold value of switch average.
Further, in dual threshold sampling process, intermediate value of the threshold θ of intermediate value for the Grad in gradient matrix q is switched
M, i.e. M=medium (q), the threshold value λ of self-adaptive switch average are taken as 500 × M.
Compared with prior art, the present invention has following beneficial technique effect：
The inventive method to a certain extent, overcomes existing remote sensing images filtering algorithm and made an uproar in face of polytype distribution
The limitation of scope of application during sound.Gradient algorithm reciprocal and the equal intermediate value blending algorithm of self-adaptive switch are combined by the present invention, this
Sample just combines two kinds of respective advantages of algorithm, not only has the advantages that to retain image border and detailed information, and for not
Partition noise of same type, such as salt-pepper noise and impulsive noise etc. will be adaptive the adaptive intermediate value of selection or average filter
Ripple algorithm, the scope of application obtains a certain degree of lifting.In addition, traditional intermediate value and Mean Filtering Algorithm uses fixed threshold,
Thus can because of threshold value fixation, processing is filtered using same standard to each sub-block, some can be caused to be made an uproar
Sound pollution degree is higher and is realized that foot phenomenon did not occurred for denoising or denoising by noise pollution degree too low sub-block.And this
Inventive method uses adaptive threshold, the i.e. threshold value can be adaptive by the height progress of noise pollution degree with the sub-block
The change answered, avoiding problems because using the generation that denoising or denoising not foot phenomenon are crossed caused by fixed threshold.Meanwhile,
This method can take into account performance of both noise suppressed and details protection.
Brief description of the drawings
Fig. 1 is the schematic flow sheet of the inventive method；
Fig. 2 is that the inventive method is contrasted with other filtering algorithms to unmanned aerial vehicle remote sensing image filtering denoising effect, wherein,
(a) it is unmanned aerial vehicle remote sensing image, image after (b) is plus made an uproar, (c) switchs mean filter, and (d) switching median filter, (e) switch is equal
Intermediate value fused filtering, the equal intermediate value fused filtering of (f) self-adaptive switch, the equal intermediate value of (g) gradient inverse self-adaptive switch is diffusion-weighted
Filtering.
Embodiment
The present invention is described in further detail below in conjunction with the accompanying drawings：
Referring to Fig. 1 and Fig. 2, traditional remote sensing images filtering carries out selecting suitable both for the types noise having determined
Filtering and noise reduction algorithm, the scope of application obtains certain limitation.For above-mentioned deficiency, in order to realize unmanned aerial vehicle remote sensing image more
Effective denoising, the present invention proposes a kind of diffusion-weighted unmanned aerial vehicle remote sensing image of the equal intermediate value of gradient self-adaptive switch reciprocal
Filtering algorithm, the algorithm not only achieves good filter effect, and the more traditional filtering algorithm of the scope of application of the algorithm is obtained
Lifting to a certain extent.Whether concrete thought is, first by the comparison with threshold value, judge the point by salt-pepper noise or arteries and veins
Noise is rushed to be polluted.Then, if judging not to be contaminated, to not carrying out ladder by the point that salt-pepper noise or impulsive noise pollute
Degree smothing filtering reciprocal；If judging to be polluted by salt-pepper noise or impulsive noise, and then again by with it is adaptive in
The threshold value of value compares, selection suitable adaptive intermediate value or mean algorithm.During being somebody's turn to do, Filtering Template is in whole noisy image
It is mobile to realize traversal.Comprise the following steps that：
Step 1, acquisition unmanned aerial vehicle remote sensing image：Using unmanned plane image capture device, remote sensing images to be processed are obtained,
Switch to gray-scale map after the processing that carries out plus make an uproar, wait next step processing.
The template that step 2, selection size are n*n, is obtained in template center's point and template between remaining (n*n-1) pixel
Grad absolute value and be stored in g matrixes；
Step 3, by the comparison with threshold value, judge current pixel point whether by salt-pepper noise or impulsive noise dirt
Dye
By taking n*n templates as an example, the gray value for making central point (x, y) is f (x, y), obtain respectively f (x, y) and template remaining
Grad between (n × n-1) individual pixel, and given threshold T=f (x, y) × 10%.If at least there is a pixel (x+
I, y+j) so that | f (x+i, y+j)-f (x, y) |≤T, then it is not caused by salt-pepper noise or random noise to judge point (x, y)；
If all pixels point (x+i, y+j) so that | f (x+i, y+j)-f (x, y) | ＞ T, then judge point (x, y) be exactly salt-pepper noise or
Caused by random noise, according to surrounding pixel point and the Grad of template center's pixel come judge templet central point whether by noise
Cause；Have been detected by coming after we realize entire image with template traversal, namely all noise spots.
Step 4.1, the result drawn according to step 3, if it is determined that not being that salt-pepper noise or random noise cause, use gradient
Inverse weight algorithm is smoothed；Detailed process is as follows：
First, by taking n*n templates as an example, the matrix q drawn in step 2 is as follows：
Then, by q matrixes, we can obtain weight matrix w,
If f (x+j, y+j)=f (x, y), then gradient is 0, regulation central element w (x, y)=0.5, remaining eight weighting members
Plain sum is 0.5, w each elements summation is equal to 1.Then have：(i, j=-1,
0,1, but for 0) when i, j difference.
Finally, template elements are multiplied with the weights correspondence corresponding to it at each pixel, then it is exactly the picture to sum
Output g (x, y) of the vegetarian refreshments using gradient inverse after smooth, i.e., n*n templates pixel is multiplied with weight matrix w correspondences, then sums, and ties
Really as the output valve of template center's pixel.
Step 4.2：The result drawn according to step 3, if it is determined that being caused by salt-pepper noise or random noise, then with certainly
Adapt to the equal intermediate value blending algorithm of switch and be filtered denoising, output g (x, y)；Detailed process is as follows：
In self-adaptive switch-average fused filtering is expressed as follows with mathematical formulae：
Wherein, σ (x, y) is that abscissa is x in image, ordinate for y point grayscale shift value；μ (x, y) is in image
Abscissa is x, and ordinate is the gray scale difference value of y point；F (x, y) is that abscissa is x, and ordinate is the gray value of y point；
Median (x, y) is the intermediate value that abscissa is gray value in x, the neighborhood of a point that ordinate is y；Mean (x, y) is that abscissa is
X, ordinate is y neighborhood of a point average gray；G (x, y) is that the abscissa after the equal medium filtering denoising of switch is
X, ordinate is the gray value of y point；θ is the threshold value of switch intermediate value；λ is the threshold value of switch average；θ is chosen for gradient matrix q
In Grad intermediate value M, (wherein, M be matrix q intermediate value, i.e. M=medium (q)), the threshold value λ of self-adaptive switch average
It is chosen for 500*M.
In the first scenario, the pixel of this condition can be met, intermediate value is either switched and still switchs mean filter
It is not noise all to think it, and the gray scale to the pixel is not changed.
In the latter case, the pixel of this condition is met, switch mean filter thinks that it is not noise, but switch
Medium filtering thinks that it is noise, so in switch-average fused filtering thinks that it is noise, and with switching median filter pair
Its is handled.
In a third case, the pixel of this condition is met, switching median filter thinks that it is not noise, but switch
Average thinks that it is noise, so in switch-average fused filtering thinks that it is noise, and it is entered with mean filter is switched
Row processing.
In the fourth case, the pixel of this condition is met, either switching median filter still switchs average filter
Ripple all thinks that it is noise.So it is noise to be also considered as it in the switch-average fused filtering, and because meet this bar
The deviation affirmative of gray value is larger in the pixel of part, its neighborhood territory pixel point, such as can be larger with switch mean filter denoising error,
So carrying out denoising to pixel with switching median filter in this case.
In equal medium filtering is switched, if threshold value λ and threshold θ get maximum, g (x, y) be just equal to f (x,
Y), i.e., any processing is not carried out to image；If threshold value λ and threshold θ are all taken as zero, g (x, y) just with median (x, y)
It is essentially equal, i.e., medium filtering is carried out to original image；If threshold value λ gets maximum, if threshold θ is zero, g (x, y) just with
Median (x, y) is essentially equal to have carried out medium filtering to original image；If threshold θ gets maximum, if threshold value λ is zero, g
(x, y) is just essentially equal with mean (i, j), i.e., carried out mean filter to original image, so the selection of threshold value is algorithm performance
And the determinant of effect, different threshold value selections are possible to the result that can be differed greatly.And what invention algorithm was used
Adaptive threshold avoid or alleviate to a certain extent this case that appearance.
Step 5, obtain the unmanned aerial vehicle remote sensing image g (x, y) after filtering and noise reduction.
Gradient Inverse Weight algorithm and the equal intermediate value blending algorithm of self-adaptive switch are combined by the inventive method, it is proposed that one
The diffusion-weighted unmanned aerial vehicle remote sensing Image filter arithmetic of the equal intermediate value of gradient self-adaptive switch reciprocal is planted, the algorithm not only possesses ladder
Degree inverse simultaneously, retains the advantage of image border and detailed information in denoising, and possesses the equal intermediate value fused filtering calculation of switch
Method to salt-pepper noise, impulsive noise, etc. different type partition noise there is preferable denoising effect.And this is adaptively opened
Close in equal intermediate value blending algorithm and use adaptive threshold, the threshold value can be adaptive by the degree of noise pollution with the sub-block
The suitable threshold value of selection answered, can take into account performance of both noise suppressed and details protection.
(e) and figure (f) contrast are schemed in Fig. 2, adaptive filter effect acquired by median filtering algorithm is bright
The aobvious equal medium filtering of switch that is better than is calculated；As shown in Table 1, every evaluation index of the latter is significantly better than the former, especially average ladder
Degree, PSNR, MSE index become apparent.As shown in Table 1, every evaluation index of the inventive method is again equal significantly better than adaptive
Median filtering algorithm, this shows that the inventive method achieves more preferable filter effect, not only remains Gradient Inverse Weight algorithm
To retaining the advantage of image border and detailed information, and possesses the equal intermediate value fused filtering algorithm of switch to salt-pepper noise, arteries and veins
Rushing the different type such as noise partition noise has the advantages that preferable denoising effect.
The algorithms of different results contrast of table 1
Claims (6)
1. the remote sensing images filtering method based on the equal intermediate value fusion of gradient self-adaptive switch reciprocal, it is characterised in that including following
Step：
Step 1：Obtain unmanned aerial vehicle remote sensing image；
Step 2：The template that size is n × n is chosen, is obtained in template center's point and template between remaining (n × n-1) pixel
Gradient value matrix q is simultaneously preserved；
Step 3：Grad and threshold value obtained by step 2 is compared, judge the pixel whether by salt-pepper noise or
Caused by random noise；
Step 4：The result drawn according to step 3, if the pixel is caused by salt-pepper noise or random noise, uses gradient
Inverse weight algorithm is smoothed；If the pixel is caused by salt-pepper noise or random noise, with adaptively opening
Close equal intermediate value blending algorithm and be filtered denoising；
Step 5：The unmanned aerial vehicle remote sensing image after filtering and noise reduction is obtained by step 4.
2. the remote sensing images filtering method according to claim 1 based on the equal intermediate value fusion of gradient self-adaptive switch reciprocal,
Characterized in that, template center's point is (x, y) in step 2, the gray value of template center's point is f (x, y), and f (x, y) is obtained respectively
With the Grad between template remaining (n × n-1) individual pixel.
3. the remote sensing images filtering method according to claim 2 based on the equal intermediate value fusion of gradient self-adaptive switch reciprocal,
Characterized in that, given threshold T=f (x, y) × 10% in step 3, if at least there is a pixel (x+i, y+j) so that |
F (x+i, y+j)-f (x, y) |≤T, then it is not caused by salt-pepper noise or random noise to judge pixel (x, y)；If all pictures
Vegetarian refreshments (x+i, y+j) so that | f (x+i, y+j)-f (x, y) | ＞ T, then it is by salt-pepper noise or random to judge pixel (x, y)
Caused by noise.
4. the remote sensing images filtering method according to claim 1 based on the equal intermediate value fusion of gradient self-adaptive switch reciprocal,
Characterized in that, being entered in step 4 when the pixel is caused by salt-pepper noise or random noise with Gradient Inverse Weight algorithm
Row smoothing processing, be specially：
For n × n template, the matrix q that step 2 is obtained is：
Weight matrix w is obtained by matrix q：
If f (x+j, y+j)=f (x, y), then Grad is 0, regulation central element w (x, y)=0.5, remaining n*n-1 weighting
Element sum is 0.5, w each elements summation is equal to 1, then has：
<mrow>
<mi>w</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>+</mo>
<mi>i</mi>
<mo>,</mo>
<mi>y</mi>
<mo>+</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mn>2</mn>
</mfrac>
<mo>&times;</mo>
<mfrac>
<mrow>
<mi>q</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>;</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>0</mn>
</mrow>
</munder>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>0</mn>
</mrow>
</munder>
<mi>q</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>;</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
</mrow>
Wherein, it is 0 when i, j=-1,0 or 1, but i, j are different；
Finally, template elements are multiplied with the weights correspondence corresponding to it at each pixel, then sum and obtain the pixel
Output g (x, y) of the point using gradient inverse after smooth.
5. the remote sensing images filtering method according to claim 1 based on the equal intermediate value fusion of gradient self-adaptive switch reciprocal,
Characterized in that, the pixel of this in step 4 is caused by salt-pepper noise or random noise, then merged with the equal intermediate value of self-adaptive switch
Algorithm is filtered denoising, is specially：
The equal intermediate value blending algorithm of self-adaptive switch, is switched over using dual threshold, realizes switch mean filter and switch intermediate value filter
The fusion treatment of ripple, is expressed as follows with mathematical formulae：
Wherein, σ (x, y) is that abscissa is x in image, ordinate for y point grayscale shift value；μ (x, y) is horizontal seat in image
X is designated as, ordinate is the gray scale difference value of y point；F (x, y) is that abscissa is x, and ordinate is the gray value of y point；median
(x, y) is the intermediate value that abscissa is gray value in x, the neighborhood of a point that ordinate is y；Mean (x, y) is that abscissa is x, indulges and sits
It is designated as y neighborhood of a point average gray；G (x, y) is that the abscissa after the equal medium filtering denoising of switch is x, indulges and sits
It is designated as the gray value of y point；In dual threshold, θ is the threshold value of switch intermediate value；λ is the threshold value of switch average.
6. the remote sensing images filtering method according to claim 5 based on the equal intermediate value fusion of gradient self-adaptive switch reciprocal,
Characterized in that, in dual threshold sampling process, switching intermediate value M, i.e. M=of the threshold θ of intermediate value for the Grad in gradient matrix q
Medium (q), the threshold value λ of self-adaptive switch average is taken as 500 × M.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710241062.5A CN107085833B (en) | 2017-04-13 | 2017-04-13 | Remote sensing images filtering method based on the equal intermediate value fusion of gradient inverse self-adaptive switch |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710241062.5A CN107085833B (en) | 2017-04-13 | 2017-04-13 | Remote sensing images filtering method based on the equal intermediate value fusion of gradient inverse self-adaptive switch |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107085833A true CN107085833A (en) | 2017-08-22 |
| CN107085833B CN107085833B (en) | 2019-07-16 |
Family
ID=59611669
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710241062.5A Active CN107085833B (en) | 2017-04-13 | 2017-04-13 | Remote sensing images filtering method based on the equal intermediate value fusion of gradient inverse self-adaptive switch |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107085833B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109146816A (en) * | 2018-08-22 | 2019-01-04 | 浙江大华技术股份有限公司 | A kind of image filtering method, device, electronic equipment and storage medium |
| CN109859133A (en) * | 2019-01-30 | 2019-06-07 | 南京邮电大学 | A kind of median filtering image de-noising method |
| CN110111273A (en) * | 2019-04-25 | 2019-08-09 | 四川轻化工大学 | Image restoration method |
| CN110163868A (en) * | 2019-04-17 | 2019-08-23 | 安阳师范学院 | A kind of remote sensing image segmentation method |
| CN111179201A (en) * | 2019-12-31 | 2020-05-19 | 广州市百果园信息技术有限公司 | Video denoising method and electronic equipment |
| CN112200735A (en) * | 2020-09-18 | 2021-01-08 | 安徽理工大学 | Temperature identification method based on flame image and control method of low-concentration gas combustion system |
| CN112598610A (en) * | 2020-12-11 | 2021-04-02 | 杭州海康机器人技术有限公司 | Depth image obtaining method and device, electronic equipment and storage medium |
| CN113379629A (en) * | 2021-06-08 | 2021-09-10 | 深圳思谋信息科技有限公司 | Satellite image denoising method and device, computer equipment and storage medium |
| CN113537048A (en) * | 2019-03-25 | 2021-10-22 | 上海商汤智能科技有限公司 | Image processing method and device, electronic equipment and storage medium |
| CN117079143A (en) * | 2023-10-16 | 2023-11-17 | 南京佳格耕耘科技有限公司 | Farmland dynamic monitoring system based on remote sensing data |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103093443A (en) * | 2012-12-26 | 2013-05-08 | 南京信息工程大学 | Image impulse noise self-adaptive filtering method based on genetic algorithm-back propagation (GA-BP) neural network |
| CN104899842A (en) * | 2015-06-29 | 2015-09-09 | 济南大学 | Ranked-order adaptively extremum median filtering method for remote line structured light image |
-
2017
- 2017-04-13 CN CN201710241062.5A patent/CN107085833B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103093443A (en) * | 2012-12-26 | 2013-05-08 | 南京信息工程大学 | Image impulse noise self-adaptive filtering method based on genetic algorithm-back propagation (GA-BP) neural network |
| CN104899842A (en) * | 2015-06-29 | 2015-09-09 | 济南大学 | Ranked-order adaptively extremum median filtering method for remote line structured light image |
Non-Patent Citations (2)
| Title |
|---|
| 乔坤等: ""一种去除椒盐噪声的自适应开关中值滤波算法"", 《计算机应用与软件》 * |
| 魏丹等: ""梯度倒数加权平滑算法的改进与实现"", 《计算机应用研究》 * |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109146816B (en) * | 2018-08-22 | 2020-08-04 | 浙江大华技术股份有限公司 | Image filtering method and device, electronic equipment and storage medium |
| CN109146816A (en) * | 2018-08-22 | 2019-01-04 | 浙江大华技术股份有限公司 | A kind of image filtering method, device, electronic equipment and storage medium |
| CN109859133A (en) * | 2019-01-30 | 2019-06-07 | 南京邮电大学 | A kind of median filtering image de-noising method |
| CN109859133B (en) * | 2019-01-30 | 2022-08-02 | 南京邮电大学 | Median filtering image denoising method |
| CN113537048A (en) * | 2019-03-25 | 2021-10-22 | 上海商汤智能科技有限公司 | Image processing method and device, electronic equipment and storage medium |
| CN110163868A (en) * | 2019-04-17 | 2019-08-23 | 安阳师范学院 | A kind of remote sensing image segmentation method |
| CN110111273A (en) * | 2019-04-25 | 2019-08-09 | 四川轻化工大学 | Image restoration method |
| CN111179201A (en) * | 2019-12-31 | 2020-05-19 | 广州市百果园信息技术有限公司 | Video denoising method and electronic equipment |
| CN111179201B (en) * | 2019-12-31 | 2023-04-11 | 广州市百果园信息技术有限公司 | Video denoising method and electronic equipment |
| WO2021135702A1 (en) * | 2019-12-31 | 2021-07-08 | 百果园技术(新加坡)有限公司 | Video denoising method and electronic device |
| CN112200735A (en) * | 2020-09-18 | 2021-01-08 | 安徽理工大学 | Temperature identification method based on flame image and control method of low-concentration gas combustion system |
| CN112598610A (en) * | 2020-12-11 | 2021-04-02 | 杭州海康机器人技术有限公司 | Depth image obtaining method and device, electronic equipment and storage medium |
| CN113379629A (en) * | 2021-06-08 | 2021-09-10 | 深圳思谋信息科技有限公司 | Satellite image denoising method and device, computer equipment and storage medium |
| CN117079143A (en) * | 2023-10-16 | 2023-11-17 | 南京佳格耕耘科技有限公司 | Farmland dynamic monitoring system based on remote sensing data |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107085833B (en) | 2019-07-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107085833A (en) | Remote sensing images filtering method based on the equal intermediate value fusion of gradient self-adaptive switch reciprocal | |
| CN105046677B (en) | A kind of enhancing treating method and apparatus for traffic video image | |
| US8013917B2 (en) | Noise filter for bayer pattern image data | |
| CN108205804A (en) | Image processing method, device and electronic equipment | |
| CN110232670B (en) | Method for enhancing visual effect of image based on high-low frequency separation | |
| CN100563303C (en) | Image processing equipment, image capture device and image processing method | |
| CN106897981A (en) | A kind of enhancement method of low-illumination image based on guiding filtering | |
| CN104574337B (en) | Based on the image enchancing method that bilateral gamma correction and multi-scale image merge | |
| IES20080497A2 (en) | Image processing method and apparatus | |
| JPH08186714A (en) | Noise removal of picture data and its device | |
| CN1985274A (en) | Methods, system and program modules for restoration of color components in an image model | |
| KR20090101911A (en) | Detecting block artifacts in coded image and video | |
| CN103888764B (en) | Self-adaptation compensation system and method for video compression distortion | |
| CN106169181A (en) | A kind of image processing method and system | |
| JPH05225332A (en) | Method and apparatus for variable-space wave filtering | |
| CN105913404A (en) | Low-illumination imaging method based on frame accumulation | |
| CN105205792A (en) | Underwater image enhancement method based on brightness and chrominance separation | |
| CN104767913B (en) | A kind of adaptive video denoising system of contrast | |
| CN107292830A (en) | Low-light (level) image enhaucament and evaluation method | |
| CN104331863A (en) | Image filtering and denoising method | |
| CN106570833A (en) | Video image self-adaptive contrast enhancement method and device | |
| JPH04188283A (en) | Method and device for processing picture | |
| CN105913391A (en) | Defogging method based on shape variable morphological reconstruction | |
| CN109767407A (en) | A kind of quadratic estimate method of atmospheric transmissivity image during defogging | |
| CN109859138A (en) | A kind of infrared image enhancing method based on human-eye visual characteristic |
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 | ||
| CB03 | Change of inventor or designer information | ||
| CB03 | Change of inventor or designer information |
Inventor after: Huang He Inventor after: Du Jingjing Inventor after: Yuan Dongliang Inventor after: Huo Zixuan Inventor after: Du Yongzhe Inventor after: Song Jing Inventor after: Guo Lu Inventor after: Xu Zhe Inventor after: Wang Guiping Inventor after: Wang Ping Inventor after: Sheng Guangfeng Inventor after: Huang Ying Inventor after: Hui Xiaobin Inventor before: Huang He Inventor before: Du Jingjing Inventor before: Huo Zixuan Inventor before: Yuan Dongliang Inventor before: Du Yongzhe Inventor before: Wang Guiping Inventor before: Wang Ping Inventor before: Xu Zhe Inventor before: Guo Lu Inventor before: Sheng Guangfeng Inventor before: Song Jing Inventor before: Huang Ying Inventor before: Hui Xiaobin |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210202 Address after: 210012-015, 7th floor, building 11, yunmi City, No.19, ningshuang Road, Yuhuatai District, Nanjing City, Jiangsu Province Patentee after: Zhilian cloud big data technology Nanjing Co.,Ltd. Address before: 710064 Shaanxi province Xi'an Beilin District South Second Ring Road No. 33 in the middle Patentee before: CHANG'AN University |