CN102890819B - Image denoising method based on pixel spatial relativity judgment - Google Patents
Image denoising method based on pixel spatial relativity judgment Download PDFInfo
- Publication number
- CN102890819B CN102890819B CN201210329780.5A CN201210329780A CN102890819B CN 102890819 B CN102890819 B CN 102890819B CN 201210329780 A CN201210329780 A CN 201210329780A CN 102890819 B CN102890819 B CN 102890819B
- Authority
- CN
- China
- Prior art keywords
- data
- array
- noise
- close
- filtered
- 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)
Abstract
The invention relates to an image denoising method based on pixel spatial relativity judgment. The method mainly comprises the following steps of: (1) inputting a frame of original image to be filtered; (2) forming an array G1 by using unfiltered chromaticity spatial component data Pc in the original image and spatial adjacent component data existing in upper, lower, left and right directions; (3) calculating the number Nc of other data with similar numerical values in Pc and G1, and selecting adjacent data to form an array G2 according to the positions of Nc and Pc; (4) judging whether Pc is noisy data or not according to Nc, G2 and pixel spatial relativity, if so, filtering and replacing, otherwise, keeping the original value of Pc; (5) repeating the steps (2), (3) and (4) until all chromaticity spatial component data in the original image are processed; and (6) outputting a frame of filtered image. By the method, a good denoising effect can be achieved on the premise of keeping image details.
Description
Technical field
The invention belongs to digital image processing field, be specifically related to a kind of image de-noising method.
Background technology
At digital image arts, because a variety of causes makes image often can be subject to the pollution of noise.The noise corrupted original information of image, hinders people to the understanding of image, also brings difficulty to the relevant treatment of image simultaneously, therefore carry out denoising to image and have a wide range of applications space.
Median filter method is a kind of relatively more conventional filtering method, and ultimate principle is that the Mesophyticum of this point of the value of digital picture center position and neighborhood thereof is replaced.Although this wave filter has, computing is simple, speed is fast, can the advantage of filtering salt-pepper noise preferably, also there is the problem that the details of image is fogged.Medium filtering can be divided into square medium filtering, cross medium filtering, X-shape medium filtering etc. according to the shape of window.Square medium filtering can by the angle point of image mistakenly as noise spot filtering while filtering noise; The line of 45 ° and 135 ° and angle can scabble by cross medium filtering; Line on 0 ° and 90 ° of directions and angle then can scabble by X-shape medium filtering (see " Li Xiaohong, Jiang Jianguo, Wu Congzhong, Zhan Shu. image goes the research of salt-pepper noise wave filter. Journal of Engineering Graphics .2009,6. ").
In noise spot determination methods, application number is the joint probability distribution of the patent computed image of 201010610864.7, according to the threshold range of joint probability distribution determination gray scale, if the intermediate value of window drops within the scope of gray threshold, do not process, otherwise replace by intermediate value; Application number is that the patent of 200510002941.x adopts and obtains the gray average of image and variance yields, judges that the mode of image slices vegetarian refreshments whether in the upper and lower 3 times of variances of average is to judge noise.Above two kinds of methods can not effectively be protected image detail marked change parts such as the texture edge of image, single pixel texture, single pixel word lines, can lose more image information, image boundary process also exists the defect of excessive denoising.
Summary of the invention
While filtering noise, also lost the problem of image detail in order to overcome traditional median filter method, the invention provides a kind of image de-noising method based on pixel spatial relativity judgment, while maintenance image detail, effectively can remove picture noise.
In order to the technical scheme solving the problems of the technologies described above employing is:
Based on an image de-noising method for pixel spatial relativity judgment, said method comprising the steps of:
(1) original image that a frame is to be filtered is inputted;
(2) a unfiltered chrominance space component data P in original image is got
c, then by P
cand the space adjacent data composition array G that top, below, left and right exist
1;
(3) chrominance space component data P is first added up
cwith array G
1in other data difference absolute value diff be less than the number N of threshold value Th
c; Again according to N
cand P
carray G is chosen in residing locus expansion
2, specific as follows:
A () works as N
c=0, if P
cfor the chrominance space component data of summit or frontier point position, then do not expand and choose array G
2; Otherwise get P
cupper left, lower-left, upper right and lower right 4 adjacent datas composition array G
2;
B () works as N
c=1, by G
1in uniquely meet diff < Th component data be designated as P
newc, first get P
newcand the upper and lower, left and right of position and upper left, lower-left, upper right and the bottom right space adjacent data composition array G that exists of totally eight directions
temp; Get G again
tempmiddle removing G
1the part composition array G of middle data
2;
C () works as N
c>=2, then do not expand and choose array G
2;
(4) P is utilized
cpresent position, and the N obtained in step (3)
c, P
newcwith array G
2, and differentiate P according to pixel space relativity
cwhether be noise data, if noise data, then carry out corresponding filtering replacement, otherwise retain P
cinitial value is as filtered data;
(5) operation of step (2) to (4) is repeated, until process all chrominance space component datas in original image;
(6) the frame filtered image be made up of all filtered data is exported.
Further, in described step (4), utilize pixel space relativity as follows to carry out differentiation process:
A () works as P
cfor the chrominance space component data of vertex position, point following three kinds of situations process: 1) if N
c=0, i.e. G
1in other data and P
cnot close, then judge P
cg is used for noise data
1in other 2 data average replace; 2) if N
c=1 and G
2in do not exist and P
newcabsolute difference is less than the data of Th, namely with P
c1 data and P is only had in the data that space is adjacent
cclose, then judge P
cg is used for noise data
1in with P
cthe data that absolute difference is more than or equal to Th are replaced; 3) there are 3 data in other situation spatially mutually continuously and numerical value is close, judge P
cretain P for non-noise data
cinitial value, namely direct by P
cas filtered data;
B () works as P
cfor the chrominance space component data of the boundary position except 4 summits, point following three kinds of situations process: 1) if N
c=0, i.e. G
1in other data and P
cnot close, then judge P
cg is used for noise data
1in other 3 data intermediate value replace; 2) if N
c=1 and G
2in do not exist and P
newcabsolute difference is less than the data of Th, namely with P
c1 data and P is only had in the data that space is adjacent
cclose, then judge P
cg is used for noise data
1in with P
cthe average that absolute difference is more than or equal to 2 data of Th is replaced; 3) there are at least 3 data in other situation spatially mutually continuously and numerical value is close, judge P
cdirect by P for non-noise data
cinitial value is as filtered data;
C () works as P
cfor the chrominance space component data of rest position except border, point following three kinds of situations process: 1) if N
c=0 and G
2in data and P
cthe number that absolute difference is less than Th is less than 2, other data namely in G1 and P
cnot close and G
2in there is no data and the P of more than 2 or 2 yet
cclose, then judge P
cg is used for noise data
1in the intermediate value of all 5 data be used as filtered data; 2) if N
c=1 and G
2in do not exist and P
newcabsolute difference is less than the data of Th, namely with P
c1 data and P is only had in the data that space is adjacent
cclose, then judge P
cg is used for noise data
1in the intermediate value of all 5 data be used as filtered data; 3) there are at least 3 data in other situation spatially mutually continuously and numerical value is close, judge P
cdirect by P for non-noise data
cinitial value is as filtered data.
The technical conceive of this method is: in order to while removal noise, keep the detailed information of image as best one can, the removal process of noise can be divided into two steps: one is the judgement of noise spot; Two is choose the value that a suitable value carrys out alternative noise spot.Based on the filtering algorithm of this structure, overcome medium filtering in the past and all pixels done unified process, easily cause image blurring drawback, an equilibrium point can be have found in denoising and details protection.
If the proximity data that there are 2 or more than 2 in the spatial neighborhood of data to be filtered is close with value data to be filtered, namely data that data to be filtered have at least 3 numerical value close are comprised and spatially mutually continuous, then can judge that data to be filtered are as non-noise data, otherwise be judged to be noise data, and carry out corresponding filtering process according to noise data present position.
Compared with prior art; this method has following beneficial effect: can while filtering noise; protect image detail well; the particularly view data of the image detail marked change such as texture edge, single pixel texture, single pixel word lines of image, and also can carry out specific aim denoising to the border of image.
Accompanying drawing explanation
Fig. 1 is a kind of image de-noising method process flow diagram based on pixel spatial relativity judgment.
Fig. 2 is the array G of diverse location chrominance space component data in image
1choose schematic diagram.Wherein, (a) array G of choosing for left upper apex position data
1; B array G that () is chosen for right vertices position data
1; C array G that () is chosen for bottom left vertex position data
1; D array G that () is chosen for bottom right vertex position data
1; E array G that () is chosen for non-boundary member position data
1; F array G that () is chosen for left margin point position data
1; G array G that () is chosen for coboundary point position data
1; The array G that figure (h) chooses for right margin point position data
1; I array G that () is chosen for lower boundary point position data
1.
Fig. 3 is the array G of image vertex position, the removing boundary position on summit and the chrominance space component data of non-boundary position
2choose schematic diagram.Wherein, (a) is N
cwhen=0, the array G that non-frontier point position data is chosen
2; (b1) and (b2) be P
newcbe respectively P
dand P
rtime, the array G that left upper apex position data is chosen
2; (c1) arriving (c3) is P
newcbe respectively P
1, P
dand P
rtime, the array G that coboundary point position data is chosen
2; (c4) be P
newcfor P
ltime, the array G that the coboundary point position data adjacent with left upper apex is chosen
2; (c5) be P
newcfor P
rtime, the array G that the coboundary point position data adjacent with right vertices is chosen
2; (d1) arriving (d4) is P
newcbe respectively P
d, P
r, P
uand P
ltime, the array G that non-frontier point position data is chosen
2; (d5) be P
newcfor P
utime, the array G that the non-frontier point position data adjacent with coboundary is chosen
2; Array G
2in data identify out with letter e in the drawings.
Embodiment
Below in conjunction with accompanying drawing and example, the specific embodiment of the present invention is described.It should be pointed out that mode described below is one of optimal way, within the scope of the invention, those skilled in the art can also find out some other alternate embodiments.
With reference to Fig. 1, a kind of image de-noising method based on pixel spatial relativity judgment, comprises step as follows:
(1) original image that a frame is to be filtered is inputted.
(2) a unfiltered chrominance space component data P in original image is got
c, then get P
cand the space adjacent data P that top exists
u, below exist space adjacent data P
d, left exist space adjacent data P
1with the space adjacent data P that right exists
rcomposition array G
1.Fig. 2 (a) ~ 2 (d) is respectively the G of the upper left of original image, upper right, lower-left and lower right 4 vertex position data
1choose schematic diagram, Fig. 2 (f) ~ 2 (i) is respectively the G of the left margin of removing 4 vertex positions, coboundary, right margin and lower boundary position data
1choose schematic diagram, Fig. 2 (e) is the G of non-boundary position data
1choose schematic diagram.
(3) P is calculated
cwith G
1in the close number N of other value data
c, and according to N
cand P
cresiding locus, expands the array G choosing and be made up of space adjacent data
2.Namely first P is added up
cwith G
1in other data difference absolute value diff be less than the number N of threshold value Th
c, wherein Th rule of thumb gets 5 ~ 15 usually, gets 10 herein; Again according to N
cand P
carray G is chosen in residing locus expansion
2, specific as follows:
A () works as N
c=0, if P
cfor the chrominance space component data of summit or boundary position, then do not expand and choose array G
2; Otherwise get P
cupper left, lower-left, upper right and lower right 4 adjacent datas composition array G
2, as shown in Fig. 3 (a).
B () works as N
c=1, by G
1in uniquely meet the data that diff is less than Th and be designated as P
newc, get P
newcand P
newcupper and lower, left and right and upper left, lower-left, upper right and the bottom right space adjacent data composition array G that exists of totally eight directions
temp.Get G again
tempmiddle removing G
1the part composition array G of middle data
2.G
2mode of specifically choosing as follows:
(I) if P
cfor the chrominance space component data of vertex position, for the left upper apex of image, if P
newcfor P
d, then G is chosen by mode Fig. 3 (b1) Suo Shi
2if, P
newcfor P
r, then G is chosen by mode Fig. 3 (b2) Suo Shi
2.Other 3 vertex position data in image also do similar process.
(II) if P
cfor the chrominance space component data of the boundary position behind removing summit, for the coboundary of image, if P
newcfor P
1then choose G by mode Fig. 3 (c1) Suo Shi
2; If P
newcfor P
dthen choose G by mode Fig. 3 (c2) Suo Shi
2; If P
newcfor P
rthen choose G by mode Fig. 3 (c3) Suo Shi
2; If P
cfor the data of right, left upper apex position and the data of left, right vertices position and P
newcbe respectively P
land P
rtime, choose G by mode Fig. 3 (c4) and 3 (c5) Suo Shi
2.Other boundary position data in image also does similar process.
(III) if P
cfor the chrominance space component data of the rest position outside removing border, work as P
newcfor P
dshi Ze chooses G by mode Fig. 3 (d1) Suo Shi
2; Work as P
newcfor P
rshi Ze chooses G by mode Fig. 3 (d2) Suo Shi
2; Work as P
newcfor P
ushi Ze chooses G by mode Fig. 3 (d3) Suo Shi
2; Work as P
newcfor P
1shi Ze chooses G by mode Fig. 3 (d4) Suo Shi
2; Work as P
cfor the data of coboundary lower position and P
newcfor P
ushi Ze chooses G by mode Fig. 3 (d5) Suo Shi
2.In image, other chooses G with the data of adjacent position, border to corresponding border Directional Extension
2time also do similar process.
C () works as N
c>=2, then do not expand and choose array G
2.
(4) P is utilized
cpresent position, and the N obtained in step (3)
c, P
newcwith array G
2, and differentiate P according to pixel space relativity
cwhether be noise data, if noise data, then carry out corresponding filtering replacement, otherwise retain P
cinitial value is as filtered data.Concrete processing procedure is as follows:
A () works as P
cfor the chrominance space component data of vertex position, point following three kinds of situations process: 1) if N
c=0, i.e. G
1in other data and P
cnot close, then judge P
cg is used for noise data
1in other 2 data average replace; 2) if N
c=1 and G
2in do not exist and P
newcabsolute difference is less than the data of Th, namely with P
c1 data and P is only had in the data that space is adjacent
cclose, then judge this P
cvalue is noise data and uses G
1in with P
cthe data that absolute difference is more than or equal to Th are replaced; 3) there are 3 data in other situation spatially mutually continuously and numerical value is close, judge P
cretain P for non-noise data
cinitial value, namely direct by P
cas chrominance space component data corresponding after filtering.
B () works as P
cfor the chrominance space component data of the boundary position except 4 summits, point following three kinds of situations process: 1) if N
c=0, i.e. G
1in other data and P
cnot close, then judge P
cg is used for noise data
1in other 3 data intermediate value replace; 2) if N
c=1 and G
2in do not exist and P
newcabsolute difference is less than the data of Th, namely with P
c1 data and P is only had in the data that space is adjacent
cclose, then judge P
cg is used for noise data
1in with P
cthe average that absolute difference is more than or equal to 2 data of Th is replaced; 3) there are at least 3 data in other situation spatially mutually continuously and numerical value is close, judge P
cdirect by P for non-noise data
cinitial value is as filtered data.
C () works as P
cfor the chrominance space component data of rest position except border, point following three kinds of situations process: 1) if N
c=0 and G
2in data and P
cthe number that absolute difference is less than Th is less than 2, i.e. G
1in other data and P
cnot close and G
2in there is no data and the P of more than 2 or 2 yet
cclose, then judge this P
cvalue is noise data and uses G
1in the intermediate value of all 5 data be used as filtered data; 2) if N
c=1 and G
2in do not exist and P
newcabsolute difference is less than the data of Th, namely with P
c1 data and P is only had in the data that space is adjacent
cclose, then judge this P
cvalue is noise data and uses G
1in the intermediate value of all 5 data be used as filtered data; 3) there are at least 3 data in other situation spatially mutually continuously and numerical value is close, judge P
cdirect by P for non-noise data
cinitial value is as filtered data.
(5) operation of step (2) to (4) is repeated, until process all chrominance space component datas in original image.
(6) the frame filtered image be made up of all filtered data is exported.
Claims (1)
1. based on an image de-noising method for pixel spatial relativity judgment, it is characterized in that: said method comprising the steps of:
(1) original image that a frame is to be filtered is inputted;
(2) a unfiltered chrominance space component data P in original image is got
c, then by P
cand the space adjacent data composition array G that top, below, left and right exist
1;
(3) chrominance space component data P is first added up
cwith array G
1in other data difference absolute value diff be less than the number N of threshold value Th
c; Again according to N
cand P
carray G is chosen in residing locus expansion
2, specific as follows:
A () works as N
c=0, if P
cfor the chrominance space component data of summit or frontier point position, then do not expand and choose array G
2; Otherwise get P
cupper left, lower-left, upper right and lower right 4 adjacent datas composition array G
2;
B () works as N
c=1, by G
1in uniquely meet diff<Th component data be designated as P
newc, first get P
newcand the upper and lower, left and right of position and upper left, lower-left, upper right and the bottom right space adjacent data composition array G that exists of totally eight directions
temp; Get G again
tempthe part composition array G of data in middle removing G1
2;
C () works as N
c>=2, then do not expand and choose array G
2;
(4) P is utilized
cpresent position, and the N obtained in step (3)
c, P
newcwith array G
2, and differentiate P according to pixel space relativity
cwhether be noise data, if noise data, then carry out corresponding filtering replacement, otherwise retain P
cinitial value is as filtered data; Utilize pixel space relativity as follows to carry out differentiation process:
(4a) P is worked as
cfor the chrominance space component data of vertex position, point following three kinds of situations process: 1) if N
c=0, i.e. G
1in other data and P
cnot close, then judge P
cg is used for noise data
1in other 2 data average replace; 2) if N
c=1 and G
2in do not exist and P
newcabsolute difference is less than the data of Th, namely with P
c1 data and P is only had in the data that space is adjacent
cclose, then judge P
cg is used for noise data
1in with P
cthe data that absolute difference is more than or equal to Th are replaced; 3) there are 3 data in other situation spatially mutually continuously and numerical value is close, judge P
cretain P for non-noise data
cinitial value, namely direct by P
cas filtered data;
(4b) P is worked as
cfor the chrominance space component data of the boundary position except 4 summits, point following three kinds of situations process: 1) if N
c=0, i.e. G
1in other data and P
cnot close, then judge P
cg is used for noise data
1in other 3 data intermediate value replace; 2) if N
c=1 and G
2in do not exist and P
newcabsolute difference is less than the data of Th, namely with P
c1 data and P is only had in the data that space is adjacent
cclose, then judge P
cg is used for noise data
1in with P
cthe average that absolute difference is more than or equal to 2 data of Th is replaced; 3) there are at least 3 data in other situation spatially mutually continuously and numerical value is close, judge P
cdirect by P for non-noise data
cinitial value is as filtered data;
(4c) P is worked as
cfor the chrominance space component data of rest position except border, point following three kinds of situations process: 1) if N
c=0 and G
2in data and P
cthe number that absolute difference is less than Th is less than 2, i.e. G
1in other data and P
cnot close and G
2in there is no data and the P of more than 2 or 2 yet
cclose, then judge P
cg is used for noise data
1in the intermediate value of all 5 data be used as filtered data; 2) if N
c=1 and G
2in do not exist and P
newcabsolute difference is less than the data of Th, namely with P
c1 data and P is only had in the data that space is adjacent
cclose, then judge P
cg is used for noise data
1in the intermediate value of all 5 data be used as filtered data; 3) there are at least 3 data in other situation spatially mutually continuously and numerical value is close, judge P
cdirect by P for non-noise data
cinitial value is as filtered data;
(5) operation of step (2) to (4) is repeated, until process all chrominance space component datas in original image;
(6) the frame filtered image be made up of all filtered data is exported.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210329780.5A CN102890819B (en) | 2012-09-07 | 2012-09-07 | Image denoising method based on pixel spatial relativity judgment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210329780.5A CN102890819B (en) | 2012-09-07 | 2012-09-07 | Image denoising method based on pixel spatial relativity judgment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102890819A CN102890819A (en) | 2013-01-23 |
| CN102890819B true CN102890819B (en) | 2015-03-04 |
Family
ID=47534313
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210329780.5A Active CN102890819B (en) | 2012-09-07 | 2012-09-07 | Image denoising method based on pixel spatial relativity judgment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102890819B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1308889A2 (en) * | 2001-10-31 | 2003-05-07 | Eastman Kodak Company | Method of spatially filtering digital images for noise removal, noise estimation or digital image enhancement |
| CN1633159A (en) * | 2005-01-26 | 2005-06-29 | 北京中星微电子有限公司 | A method for removing image noise |
| CN102034227A (en) * | 2010-12-29 | 2011-04-27 | 四川九洲电器集团有限责任公司 | Method for de-noising image |
-
2012
- 2012-09-07 CN CN201210329780.5A patent/CN102890819B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1308889A2 (en) * | 2001-10-31 | 2003-05-07 | Eastman Kodak Company | Method of spatially filtering digital images for noise removal, noise estimation or digital image enhancement |
| CN1633159A (en) * | 2005-01-26 | 2005-06-29 | 北京中星微电子有限公司 | A method for removing image noise |
| CN102034227A (en) * | 2010-12-29 | 2011-04-27 | 四川九洲电器集团有限责任公司 | Method for de-noising image |
Non-Patent Citations (4)
| Title |
|---|
| An Adaptive Median Filter for Image Denoising;Chin-Chen Chang等;《Second International Symposium on Information Technology Application》;20081022;第346-350页 * |
| An Effective Adaptive Median Filter Algorithm for Removing Salt&Pepper Noise in Images;Jiafu Jiang等;《2010 Symposium on Photonics and Optoelectronic(SOPO)》;20101231;第1-4页 * |
| Selective switching median filter for the removal of salt & pepper impulse noise;Krishnan Nallaperumal等;《2006IFIP International Conference on Wireless and Optical Communication Newworks》;20061231;第1-5页 * |
| 高克芳等.一种基于噪声点检测的自适应中值滤波方法.《福建农林大学学报(自然科学版)》.2009,第38卷(第3期),第333-336页. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102890819A (en) | 2013-01-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9639918B2 (en) | Method for anti-aliasing of image with super-resolution | |
| CN102831591B (en) | Gaussian filter-based real-time defogging method for single image | |
| CN101930597B (en) | Mathematical morphology-based image edge detection method | |
| CN101226630B (en) | Interpolation method based on cubic spline function | |
| Horng et al. | Using sorted switching median filter to remove high-density impulse noises | |
| WO2006055693A3 (en) | System and method for a vector difference mean filter for noise suppression | |
| CN104835127B (en) | A kind of self-adaptive smooth filtering method | |
| CN111724315B (en) | Infrared image noise removal method based on self-adaptive weighted median filtering | |
| US9697434B2 (en) | Edge detection system and methods | |
| CN101908205B (en) | Magic square coding-based median filter method | |
| CN106910170A (en) | A kind of minimizing technology of image salt-pepper noise | |
| CN104299198A (en) | Fast image defogging method based on dark channels of pixels | |
| JP2009090950A (en) | Tire management system | |
| CN102890819B (en) | Image denoising method based on pixel spatial relativity judgment | |
| CN103268472A (en) | Dual-color-space-based lip detection method | |
| CN102509265B (en) | Digital image denoising method based on gray value difference and local energy | |
| JP4229320B2 (en) | Pneumatic tire | |
| CN105719257A (en) | Method for removing super-high-density salt-and-pepper noises of image | |
| CN107481253B (en) | Edge-based spot detection method | |
| Priestley et al. | A decision based switching median filter for restoration of images corrupted by high density impulse noise | |
| CN102842119A (en) | Quick document image super-resolution method based on image matting and edge enhancement | |
| JP2014038518A (en) | Handwritten document processing device, method, and program | |
| CN110853146B (en) | Relief modeling method, system and relief processing equipment | |
| CN106558029A (en) | A kind of image filtering method and device | |
| CN103136735A (en) | Single image defogging method based on dual-scale dark channel |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |