CN106447614B - A kind of super-resolution method based on Uniform Domains - Google Patents
A kind of super-resolution method based on Uniform Domains Download PDFInfo
- Publication number
- CN106447614B CN106447614B CN201611055298.1A CN201611055298A CN106447614B CN 106447614 B CN106447614 B CN 106447614B CN 201611055298 A CN201611055298 A CN 201611055298A CN 106447614 B CN106447614 B CN 106447614B
- Authority
- CN
- China
- Prior art keywords
- pixel
- resolution
- straight line
- value
- edge
- 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
- 238000000034 method Methods 0.000 title claims abstract description 58
- 238000001914 filtration Methods 0.000 claims abstract description 11
- 238000001514 detection method Methods 0.000 claims abstract description 6
- 230000008859 change Effects 0.000 claims description 3
- 238000007796 conventional method Methods 0.000 abstract description 3
- 230000000007 visual effect Effects 0.000 abstract description 2
- 238000012545 processing Methods 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- HUTDUHSNJYTCAR-UHFFFAOYSA-N ancymidol Chemical compound C1=CC(OC)=CC=C1C(O)(C=1C=NC=NC=1)C1CC1 HUTDUHSNJYTCAR-UHFFFAOYSA-N 0.000 description 2
- 230000002146 bilateral effect Effects 0.000 description 2
- 230000003321 amplification Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4053—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4023—Scaling of whole images or parts thereof, e.g. expanding or contracting based on decimating pixels or lines of pixels; based on inserting pixels or lines of pixels
-
- 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
- G06T2207/10016—Video; Image sequence
-
- 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
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
- Editing Of Facsimile Originals (AREA)
Abstract
The invention discloses a kind of super-resolution method based on Uniform Domains, the super-resolution method based on Uniform Domains is the following steps are included: first with Sobel operator come detection of edge pixels point；Secondly the set for participating in the pixel of estimation and filtering of high-resolution pixel value is determined；It will finally be filtered that non-edge pixels is selected and the pixel value of full-resolution picture vegetarian refreshments in this area to determine in Uniform Domains.The present invention carries out interpolation in the Uniform Domains of image as far as possible, to avoid the interpolation for crossing over two inconsistent regions；It, can be to avoid the shortcomings that edge blurry and obtaining better visual effect in conventional method due to only being filtered in Uniform Domains.
Description
Technical field
The invention belongs to technical field of image processing more particularly to a kind of super-resolution methods based on Uniform Domains.
Background technique
The image for the low resolution propagated on network due to being limited by system bandwidth is needed in high-resolution display
Be clearly displayed under print system, thus the super-resolution technique of single image is suggested to solve this recently to ask
Topic improves the resolution ratio of image more visiblely.This technology can also be applied in video monitoring neighborhood, to improve low point
The resolution ratio of the video containing suspect of resolution, more clearly shows the image containing suspect.Meanwhile it being led in pattern-recognition
Domain, the resolution ratio that this technology can be improved image carry out enlarged drawing, to improve discrimination.In existing document, there are three classes
The method of the super-resolution of single image.In the method for the first kind, high resolution graphics is estimated or filled using the method for interpolation
As upper unknown pixel value.The shortcomings that such methods, is to make the edge of image excessively fuzzy.In the method for the second class, using instruction
Set composed by the low resolution block extracted from existing image perfected and high-resolution block correspondingly is to predict
Current unknown high-resolution block.The shortcomings that such methods, is that its computation complexity is very big, it is difficult to it is real-time to be applied to needs
The occasion of application, as the Real-time High Resolution rate of video image is handled.In such method, existing simultaneously makes image quality decrease
Blocking artifact.In third class method, unknown high-resolution block is obtained using the good sparse dictionary of precondition.However,
Input picture and the image trained differ greatly in many cases, and third class method can generate very big distortion at this time.
From above discussion, it can be seen that the super-resolution processing of single image is still a hotter research topic, and attracts
A large amount of researcher conducts a research in this respect.The super-resolution method of the basic single image based on interpolation, bilinearity
In this approach, the curved surface that institute's interpolation is formed will such as above two methods be smooth for interpolation method.Assuming that a unit length,
Its pixel value is I (x, y), and the partial differential I on four angles on wide squarex(x,y),Iy(x, y), and Ixy(x,y)
It is all known.So, the surface equation of interpolation isHere coefficient ai,jPixel can be passed through
The first derivative and second dervative of value function are continuously assumed to obtain.From this equation, it can be seen that one share 14 here
A pixel value takes part in obtain the calculating of high-resolution pixel value.Currently, according to a large amount of experimental result, it can be deduced that general
In the case of, the performance of bilinear interpolation method is better than the conclusion of arest neighbors interpolation method and bilinear interpolation method.But this
In the high-definition picture that method still remains the problem of the obscure portions of image border, how to reduce edge blurry and be still
One challenge of research field.
In conclusion there are image border obscure portions in obtained high-definition picture in existing bi-cubic interpolation method
The problem of.
Summary of the invention
The purpose of the present invention is to provide a kind of super-resolution methods based on Uniform Domains, it is intended to solve existing double
Image border obscure portions in the high-definition picture that cube interpolation method there are problems that.
The invention is realized in this way the super-resolution side based on Uniform Domains in a kind of digital image processing field
Method, the super-resolution method based on Uniform Domains the following steps are included:
Carry out detection of edge pixels point first with Sobel operator；
Secondly the set for participating in the pixel of estimation and filtering of high-resolution pixel value is determined；
The pixel value to determine full-resolution picture vegetarian refreshments in this area will be finally filtered in Uniform Domains.
Further, the determination method of the Sobel operator and edge anchor point pixel includes:
It is expressed as in the template of the Sobel operator in x-axis and y-axis direction:
Meanwhile the modulus value and phase angle of gradient indicate are as follows:
The convolution algorithm of operator representation second order；Using a threshold T1, the pixel that gradient modulus value is greater than thresholding will be true
The edge pixel being set in image；To each high-resolution pixel, the pixel of four low resolution nearest with it is found, is used
(i, j) indicates the coordinate on low-resolution image.
Further, for the high-resolution pixel intermediate at (i, j) and (i, j+1), low-resolution pixel (i, j), (i, j+
1), (i-1, j), (i-1, j+1) will be determined four neighbor pixels of high-resolution pixel thus；
For the high-resolution pixel between (i, j) and (i+1, j), low-resolution pixel (i, j), (i+1, j), (i,
), j-1 (i+1, j-1) will be determined four neighbor pixels of high-resolution pixel thus；
For four neighbor pixels, if pixel (i, j) is edge pixel, it will be put by this and do a straight line, the angle of straight line
It is vertical with the angle φ (i, j) of the gradient of this pixel；For the straight line done near the pixel of Current high resolution, this picture is calculated
Element arrives the distance between each straight line；The straight line nearest from this pixel is the anchor straight line of current high-resolution pixel, generates this straight line
Low-resolution pixel is referred to as anchor point.
Further, the Uniform Domains around Current high resolution pixel are determined；Current edge anchor point pixel is p1=(ai,
aj), then increase by two anchor point p of selection2=(ai-1,aj- 1) and p3=(ai+1,aj+1)；For two anchor points, with two anchor points
Centered on do two straight lines, the direction of two straight lines is parallel with existing anchor rectilinear direction；On every straight line, select close
5 pixels of the anchor point that straight line and this straight line pass through, that is, minimize following function:
α*||pi-xi||+β*l(xi)；
||pi-xi| | it is pixel xiTo pixel piThe distance between, l (xi) it is pixel xiTo by anchor point piIt generates
The distance between straight line, α and β are two parameters in method；3 Uniform Domains set omegas comprising pixel will be generated1,
Ω2, and Ω3, and each set includes 5 pixels；The average value of the pixel value of pixel in each set is calculated, is gathered
Ω1,Ω2, and Ω3In average pixel value use m respectively1,m2, and m3To indicate；|m2-m1| > | m3-m1|, then use set omega3It replaces
For Ω2；Otherwise, Ω2Do not change.
Further, the determination method of high-resolution pixel value includes:
Its pixel value is determined using formula:
(u, v) is the position of current high-resolution pixel, and Ω is and the pixel of its nearest low resolution, f (x) and g (x) are
Two functions, take f (x)=x3, g (x)=x2, m (u, v) is the average value of the pixel value of the pixel nearest with current high-resolution,
W is normalized parameter, is had simultaneously:
For edge pixel, sets of pixel values Ω is used first1Substitute the formula of Ω use or moreTo obtain I1(u,v)；Then, it uses
Sets of pixel values Ω2Substitute the formula of Ω use or moreTo obtain I2(u,v)；To two
Value is combined to obtain high-resolution pixel value I (u, v), it may be assumed that
I (u, v)=w1*I1(u,v)+w2*I2(u,v)；
Parameter wiWith pixel collection ΩiThe variance of middle pixel value is inversely proportional.
Super-resolution method provided by the invention based on Uniform Domains, using Sobel operator come detection of edge pixels
Point；Determine the set for participating in the pixel of estimation and filtering of high-resolution pixel value；It will be filtered in Uniform Domains
To determine, non-edge pixels is selected and the pixel value of full-resolution picture vegetarian refreshments in this area；As far as possible in the Uniform Domains of image
Interpolation is carried out, to avoid the interpolation for crossing over two inconsistent regions；Due to only being filtered in Uniform Domains, can keep away
The shortcomings that exempting from edge blurry in conventional method simultaneously obtains better visual effect.Experimental result show that institute moving party
The superiority of method.The invention proposes a kind of new based on the method for interpolation to improve existing methods performance, such to reduce
The fuzzy problem of generally existing image border part in method.To obtain clearer high-resolution image.
Detailed description of the invention
Fig. 1 is the super-resolution method flow chart provided in an embodiment of the present invention based on Uniform Domains.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to embodiments, to the present invention
It is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not used to
Limit the present invention.
Application principle of the invention is explained in detail with reference to the accompanying drawing.
As shown in Figure 1, the super-resolution method provided in an embodiment of the present invention based on Uniform Domains the following steps are included:
S101: carry out detection of edge pixels point first with Sobel operator；
S102: the set for participating in the pixel of estimation and filtering of high-resolution pixel value is determined；
S103: it will be filtered in Uniform Domains to determine non-edge pixels point and full-resolution picture in this area
The pixel value of vegetarian refreshments.
Application principle of the invention is further described combined with specific embodiments below.
1 the embodiment of the present invention will carry out detection of edge pixels using Sobel gradient operator first.Then, determination is being counted
The pixel for needing to use during calculating high-resolution pixel value and in filtering.Then, high-resolution is determined using bilateral filtering
The pixel value of the edge pixel part of rate.It, can be to avoid figure since bilateral filtering is the trend along edge and carries out here
As the fuzzy of marginal portion and obtain better result.
The determination of 1.1Sobel operator and edge anchor point pixel
It can be indicated in the template of the Sobel operator in x-axis and y-axis direction are as follows:
Meanwhile the modulus value and phase angle of gradient can indicate are as follows:
HereThe convolution algorithm of operator representation second order.These operations will be implemented the pixel of low resolution.Then, it will adopt
With a threshold T1, gradient modulus value is greater than the edge pixel that the pixel of thresholding will be confirmed as in image.
After completing this operation, to each high-resolution pixel, the pixel of four low resolution nearest with it is found.
Here coordinate on low-resolution image is indicated with (i, j), for the high-resolution pixel in the centre (i, j) and (i, j+1),
Low-resolution pixel (i, j), (i, j+1), (i-1, j), (i-1, j+1) will be determined four neighbour's pictures of high-resolution pixel thus
Element.For the high-resolution pixel between (i, j) and (i+1, j), low-resolution pixel (i, j), (i+1, j), (i, j-1),
(i+1, j-1) will be determined four neighbor pixels of high-resolution pixel thus.For other situations, four neighbor pixels can be by very
It is easy and uniquely determines.
For four neighbor pixels, if pixel (i, j) is edge pixel, it will be put by this and do a straight line, the angle of this straight line
The angle φ (i, j) (being calculated by above formula 6) of the gradient of degree and this pixel is vertical.Near the pixel of Current high resolution
The straight line done calculates this pixel the distance between to each straight line.The straight line nearest from this pixel will be referred to as in the present invention
It is the anchor straight line of current high-resolution pixel, the low-resolution pixel for generating this straight line is referred to as anchor point.
1.2 pairs of each high-resolution pixels determine set composed by the pixel of the participation filtering corresponding to it
When four neighborhood pixels of Current high resolution pixel determined above are not edge pixels, Current high resolution
Pixel is confirmed as smooth pixel.9 low-resolution pixels of smooth pixel and its arest neighbors will be confirmed as and this picture
Pixel in the corresponding filtered pixel set of element.Here, when determining neighbor pixel, if there is more than one pixel and current picture
Element distance having the same, then the pixel of preferred upper left angular direction, these pixels have higher priority.
Have in four neighbor pixels of the pixel of Current high resolution one or more pixel be edge pixel when,
By using and filtered pixel set of the different method of above method to determine this pixel in pixel.Described above
After anchor point and anchor straight line all determines in step, upward filtering is walked into control as much as possible perpendicular to edge, keeps its smaller the better,
Carry out the pixel value that high-resolution pixel is accurately determined along the filtering of anchor straight line simultaneously.In this way, base can be reduced to the greatest extent
The defect for having many edge blurry parts in the conventional method of interpolation.At the fringe region of image, the edge of usual image
Content in image is divided into two relatively uniform regions of content.Picture for each content consistency region, in region
Plain value reaches unanimity.So herein, it is first determined the Uniform Domains around Current high resolution pixel.If current edge
Anchor point pixel is p1=(ai,aj), then increase by two anchor point p of selection2=(ai-1,aj- 1) and p3=(ai+1,aj+1).For this
Two anchor points of multiselect centered on the two anchor points will do two straight lines more, and the direction of the two straight lines and existing anchor are straight
Line direction is parallel.In this way, including existing anchor straight line, one shares three straight lines.On every straight line, select close to straight line and this
5 pixels of the anchor point that straight line passes through, that is, minimize following function:
α*||pi-xi||+β*l(xi) (7)
Here, | | pi-xi| | it is pixel xiTo pixel piThe distance between, l (xi) it is pixel xiTo by anchor point pi
The distance between straight line of generation, α and β are two parameters in method.When encountering the pixel for having same functions value, use
With identical method described above, the preferential pixel for selecting upper left angular direction.After completing this operation, include by generating 3
The set omega of the Uniform Domains of pixel1,Ω2, and Ω3, and each set includes 5 pixels.Then, it calculates each
The average value of the pixel value of pixel, set omega in set1,Ω2, and Ω3In average pixel value use m respectively1,m2, and m3Come
It indicates.If | m2-m1| > | m3-m1|, then use set omega3Substitute Ω2；Otherwise, Ω2Do not change.
The determination of 1.3 high-resolution pixel values:
For being determined as the high-resolution pixel of smooth pixel above, its pixel value will be determined using following formula:
Here, (u, v) is the position of current high-resolution pixel, and Ω is the pixel with its nearest low resolution, f (x) and g
(x) it is two functions, takes f (x)=x in the present invention3, g (x)=x2, m (u, v) is the pixel nearest with current high-resolution
The average value of pixel value, w are normalized parameters, are had simultaneously:
For edge pixel, sets of pixel values Ω is used first1The formula (8) of Ω use or more is substituted, to obtain I1(u,v)。
Then, with sets of pixel values Ω2The formula (8) of Ω use or more is substituted, to obtain I2(u,v).Then, will to the two values into
Row combination is to obtain high-resolution pixel value I (u, v), it may be assumed that
I (u, v)=w1*I1(u,v)+w2*I2(u,v) (10)
Here, parameter wiWith pixel collection ΩiThe variance of middle pixel value is inversely proportional.
Application effect of the invention is explained in detail below with reference to experiment.
The comparison of the PSNR value of the method for the invention of table 1 and bi-cubic interpolation method
In the experiment to the method for the present invention, parameter value α and β are set individually into 2 and 1.Threshold T1It is configured to
200.The image tested from website " http://sipi.usc.edu/database/database.php? volume=
Image data base on misc ".In an experiment, original image is reduced first and down-sampling, the factor of down-sampling are 4 × 4.
Then, super-resolution processing is carried out to the image of obtained down-sampling using method of the invention and bi-cubic interpolation method, obtained
To the reconstruction image of big amplification as original image.Finally, calculating the peak value noise between original image and reconstruction image
Than (PSNR) with the performance of the various methods of determination.Table 1 shows experimental result.From this table, it is easy to see that the method proposed
Performance will be much better than the method for bi-cubic interpolation.
In present aspect, a new interpolation super-resolution method is proposed, is carried out in the Uniform Domains of image slotting
Value, to avoid the interpolation for crossing over two inconsistent regions.With the method, the high-resolution of reconstruction can be reduced to a certain extent
In the image of rate the problem of edge blurry.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Made any modifications, equivalent replacements, and improvements etc., should all be included in the protection scope of the present invention within mind and principle.
Claims (2)
1. a kind of super-resolution method based on Uniform Domains, which is characterized in that the super-resolution based on Uniform Domains
Rate method the following steps are included:
Carry out detection of edge pixels point first with Sobel operator；
Secondly the set for participating in the pixel of estimation and filtering of high-resolution pixel value is determined；
It will finally be filtered that non-edge pixels is selected and full-resolution picture vegetarian refreshments in this area to determine in Uniform Domains
Pixel value；
The determination method of the Sobel operator and edge pixel includes:
It is expressed as in the template of the Sobel operator in x-axis and y-axis direction:
Meanwhile the modulus value and phase angle of gradient indicate are as follows:
The convolution algorithm of operator representation second order；Using a threshold T1, gradient modulus value be greater than thresholding pixel will be confirmed as figure
Edge pixel as in；To each edge pixel, the pixel of four low resolution nearest with it is found, is indicated with (i, j)
The coordinate of the pixel of low resolution on the image；
For the high-resolution pixel intermediate at (i, j) and (i, j+1), low-resolution pixel (i, j), (i, j+1), (i-1, j),
(i-1, j+1) will be determined four neighbor pixels of high-resolution pixel thus；
For the high-resolution pixel between (i, j) and (i+1, j), low-resolution pixel (i, j), (i+1, j), (i, j-1),
(i+1, j-1) will be determined four neighbor pixels of high-resolution pixel thus；
For four neighbor pixels, if pixel (i, j) is edge pixel, a straight line will be done by this point, the angle of straight line and this
The angle of the gradient of pixelVertically；For the straight line done near the pixel of Current high resolution, this pixel is calculated to respectively
The distance between straight line；The straight line nearest from this pixel is the anchor straight line of current high-resolution pixel, generates the low resolution of this straight line
Rate pixel is referred to as anchor point；
Determine the Uniform Domains around Current high resolution pixel；Current edge anchor point pixel is p1=(ai,aj), then increase
Select two anchor point p2=(ai-1,aj- 1) and p3=(ai+1,aj+1)；It is centered on two anchor points to do for two anchor points more
Two straight lines, the direction of two straight lines are parallel with existing anchor rectilinear direction；On every straight line, select straight with this close to straight line
5 pixels of the anchor point that line passes through, that is, minimize following function:
α*||pi-xi||+β*l(xi)；
||pi-xi| | it is pixel xiTo pixel piThe distance between, l (xi) it is pixel xiTo the straight line generated by anchor point pi
The distance between, α and β are two parameters in method；3 consistency set omegas including pixel will be generated1,Ω2, and
Ω3, and each set includes 5 pixels；The average value of the pixel value of pixel in each consistency set is calculated, is gathered
Ω1,Ω2, and Ω3In average pixel value use m respectively1,m2, and m3To indicate；|m2-m1| > | m3-m1|, then use set omega3It replaces
For Ω2；Otherwise, Ω2Do not change；
Then, respectively in Uniform Domains set omega1And Ω2In be filtered, find out two values, then group is carried out to the two values
Conjunction obtains high-resolution pixel value.
2. as described in claim 1 based on the super-resolution method of Uniform Domains, which is characterized in that high-resolution pixel value
Determination method include:
Its pixel value is determined using formula:
(u, v) is the position of current high-resolution pixel, and Ω is the set of pixels with the low resolution in its nearest Uniform Domains
It closes, f (x) and g (x) are two functions, take f (x)=x3, g (x)=x2, m (u, v) is the picture nearest with Current high resolution pixel
Average value in the sets of pixel values Ω of vegetarian refreshments, w are normalized parameters, are had simultaneously:
For edge pixel, sets of pixel values Ω is used first1Substitute the formula of Ω use or moreTo obtain I1(u,v)；So
Afterwards, with sets of pixel values Ω2Substitute the formula of Ω use or moreTo obtain I2(u,v)；It is right
Two values are combined to obtain high-resolution pixel value I (u, v), it may be assumed that
I (u, v)=w1*I1(u,v)+w2*I2(u,v)；
Parameter wiWith pixel collection ΩiThe variance of middle pixel value is inversely proportional.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611055298.1A CN106447614B (en) | 2016-11-25 | 2016-11-25 | A kind of super-resolution method based on Uniform Domains |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611055298.1A CN106447614B (en) | 2016-11-25 | 2016-11-25 | A kind of super-resolution method based on Uniform Domains |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106447614A CN106447614A (en) | 2017-02-22 |
| CN106447614B true CN106447614B (en) | 2019-06-07 |
Family
ID=58219640
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611055298.1A Active CN106447614B (en) | 2016-11-25 | 2016-11-25 | A kind of super-resolution method based on Uniform Domains |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106447614B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109345490B (en) * | 2018-11-20 | 2021-09-21 | 湖南快乐阳光互动娱乐传媒有限公司 | Method and system for enhancing real-time video image quality of mobile playing terminal |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102609900A (en) * | 2012-01-16 | 2012-07-25 | 北京环星基业数码科技发展有限公司 | Image interpolation algorithm based on region consistency |
| CN103810675A (en) * | 2013-09-09 | 2014-05-21 | 深圳市华星光电技术有限公司 | Image super-resolution reconstruction system and method |
| CN104732491A (en) * | 2015-03-05 | 2015-06-24 | 北京空间机电研究所 | Edge priority guide single-frame remote sensing image super-resolution processing method |
| EP2927864A1 (en) * | 2012-11-29 | 2015-10-07 | NEC Corporation | Image processing device and image processing method |
| CN105678700A (en) * | 2016-01-11 | 2016-06-15 | 苏州大学 | Image interpolation method and system based on prediction gradient |
-
2016
- 2016-11-25 CN CN201611055298.1A patent/CN106447614B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102609900A (en) * | 2012-01-16 | 2012-07-25 | 北京环星基业数码科技发展有限公司 | Image interpolation algorithm based on region consistency |
| EP2927864A1 (en) * | 2012-11-29 | 2015-10-07 | NEC Corporation | Image processing device and image processing method |
| CN103810675A (en) * | 2013-09-09 | 2014-05-21 | 深圳市华星光电技术有限公司 | Image super-resolution reconstruction system and method |
| CN104732491A (en) * | 2015-03-05 | 2015-06-24 | 北京空间机电研究所 | Edge priority guide single-frame remote sensing image super-resolution processing method |
| CN105678700A (en) * | 2016-01-11 | 2016-06-15 | 苏州大学 | Image interpolation method and system based on prediction gradient |
Non-Patent Citations (2)
| Title |
|---|
| 《基于Sobel算子的自适应图像缩放算法》;江雯等;《计算机工程》;20100430;第214-216页 |
| 《基于索贝尔算子的高分辨率遥感影像分割技术研究》;王瀚闻等;《科技创新导报》;20121231;第18页 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106447614A (en) | 2017-02-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhao et al. | Simultaneous color-depth super-resolution with conditional generative adversarial networks | |
| CN105517671B (en) | Video frame interpolation method and system based on optical flow method | |
| JP5342068B2 (en) | Multiple frame approach and image upscaling system | |
| JP5058341B2 (en) | Image processing method | |
| JP4658223B2 (en) | Image generating method, apparatus, program thereof, and recording medium recording program | |
| Wei et al. | Contrast-guided image interpolation | |
| TWI359617B (en) | Low-complexity and high-quality error concealment | |
| JP2011504682A (en) | Resize image sequence | |
| TW200937946A (en) | Full-frame video stabilization with a polyline-fitted camcorder path | |
| CN107895345A (en) | A kind of method and apparatus for improving facial image resolution ratio | |
| JP2015522198A (en) | Depth map generation for images | |
| JP2013225740A (en) | Image formation device, image display device, and image formation method and image formation program | |
| DE102010046259A1 (en) | Image frequency conversion using motion estimation and compensation | |
| KR101987079B1 (en) | Method for removing noise of upscaled moving picture with dynamic parameter based on machine learning | |
| JP2011060282A (en) | Method and system for motion detection using nonlinear smoothing of motion field | |
| CN104537627B (en) | A kind of post-processing approach of depth image | |
| Kim et al. | Joint-adaptive bilateral depth map upsampling | |
| CN106447614B (en) | A kind of super-resolution method based on Uniform Domains | |
| Koo et al. | An image resolution enhancing technique using adaptive sub-pixel interpolation for digital still camera system | |
| TWI288896B (en) | Method and apparatus of deinterlacing | |
| Al Ismaeil et al. | Dynamic super resolution of depth sequences with non-rigid motions | |
| Zhang et al. | Video Superresolution Reconstruction Using Iterative Back Projection with Critical‐Point Filters Based Image Matching | |
| Nguyen et al. | Depth image-based rendering with low resolution depth | |
| Ye et al. | Arobust approach to super-resolution sprite generation | |
| Man et al. | Image Super-Resolution Using Image Registration and Neural Network Based Interpolation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20230621 Address after: 509 Kangrui Times Square, Keyuan Business Building, 39 Huarong Road, Gaofeng Community, Dalang Street, Longhua District, Shenzhen, Guangdong Province, 518000 Patentee after: Shenzhen lizhuan Technology Transfer Center Co.,Ltd. Address before: 321004 688 Yingbin Road, Zhejiang, Jinhua, Zhejiang Normal University Patentee before: ZHEJIANG NORMAL University |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20230816 Address after: Room 406, Unit B, Building 1, No. 13 Beiyuan Road, Chaoyang District, Beijing, 100107 Patentee after: Beijing More Vision Technology Co.,Ltd. Address before: 509 Kangrui Times Square, Keyuan Business Building, 39 Huarong Road, Gaofeng Community, Dalang Street, Longhua District, Shenzhen, Guangdong Province, 518000 Patentee before: Shenzhen lizhuan Technology Transfer Center Co.,Ltd. |
|
| TR01 | Transfer of patent right |