CN102479380A - Image interpolation method capable of improving image resolution and device utilizing image interpolation method - Google Patents

Image interpolation method capable of improving image resolution and device utilizing image interpolation method Download PDF

Info

Publication number
CN102479380A
CN102479380A CN2010105587451A CN201010558745A CN102479380A CN 102479380 A CN102479380 A CN 102479380A CN 2010105587451 A CN2010105587451 A CN 2010105587451A CN 201010558745 A CN201010558745 A CN 201010558745A CN 102479380 A CN102479380 A CN 102479380A
Authority
CN
China
Prior art keywords
resolution
vegetarian refreshments
pixel
selected window
partial structurtes
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
Application number
CN2010105587451A
Other languages
Chinese (zh)
Other versions
CN102479380B (en
Inventor
任杰
刘家瑛
郭宗明
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
New Founder Holdings Development Co ltd
Peking University
Beijing Founder Electronics Co Ltd
Original Assignee
Peking University
Peking University Founder Group Co Ltd
Beijing Founder Electronics Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Peking University, Peking University Founder Group Co Ltd, Beijing Founder Electronics Co Ltd filed Critical Peking University
Priority to CN201010558745.1A priority Critical patent/CN102479380B/en
Publication of CN102479380A publication Critical patent/CN102479380A/en
Application granted granted Critical
Publication of CN102479380B publication Critical patent/CN102479380B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Image Processing (AREA)

Abstract

The invention discloses an image interpolation method capable of improving image resolution and a device utilizing the image interpolation method, aiming to overcome defects at boundaries of interpolated images in the prior art. The image interpolation method capable of improving image resolution includes steps of determining similarity probability weight between all pixels in a local window and the reference pixel, modulating contribution of each pixel to the integral error in optimization estimation according to the similarity probability weight, and finally calculating all the pixel values to be estimated in the local window by estimating according to the weighted least square process. Since the similarity probability weight is introduced to serve as the weight for estimation of high-resolution pixels, estimation error is minimized, and defects at the boundaries are reduced.

Description

A kind of image interpolation method and device that improves image resolution ratio
Technical field
The invention belongs to technical field of image processing, particularly a kind of image interpolation method and device that improves image resolution ratio.
Background technology
Image is one of the mankind's main carrier of obtaining outside objective world information.Along with human society gets into the information age, computer technology, modern communications technology and the information processing technology have obtained development fast, and consumption electronic product has spread all over each corner of human lives.Along with the appearance of digital picture and application widely; People are also increasingly high to the requirement of picture quality; Improving the image capture device hardware capabilities under the situation near the limit, adopting software engineering to improve original image resolution ratio becomes one of hot issue of digital image processing field gradually.
Image interpolation is traditional one of method that improves image resolution ratio that is used for.The interpolation method of traditional digital picture has a variety of; Wherein fastest is Nearest Neighbor (nearest-neighbor) method; This method through copy be inserted into the pixel value of a little adjacent nearest pixel value as the insertion point; This method speed is fast, realizes simply, but in interpolation image, the zigzag flaw occurs through regular meeting; Bilinear (bilinearity) interpolation method is through getting each mean value that is inserted into four immediate pixel numerical value a little as the insertion point pixel value; This method has been eliminated the zigzag phenomenon; But computing time is long slightly, and blurring effect appears in boundary in image easily.Bicubic (bicubic) convolution method with the interpolation point (i+u, j+v) 16 neighbor pixel points are considered to come on every side, through following interpolation formula can obtain this interpolation point pixel value f (i+u, j+v):
f(i+u，j+v)＝A*B*C
A＝[s(u+1)s(u+0)s(u-1)s(u-2)]
B = f ( i - 1 , j - 1 ) f ( i - 1 , j + 0 ) f ( i - 1 , j + 1 ) f ( i - 1 , j + 2 ) f ( i + 0 , j - 1 ) f ( i + 0 , j + 0 ) f ( i + 0 , j + 1 ) f ( i + 0 , j + 2 ) f ( i + 1 , j - 1 ) f ( i + 1 , j + 0 ) f ( i + 1 , j + 1 ) f ( i + 1 , j + 2 ) f ( i + 2 , j - 1 ) f ( i + 2 , j + 0 ) f ( i + 2 , j + 1 ) f ( i + 2 , j + 2 )
C＝[s(v+1)s(v+0)s(v-1)s(v-2)] T
s ( x ) = 1 - 2 | x | 2 + | x | 3 0 &le; | x | < 1 4 - 8 | x | + 5 | x | 2 - | x | 3 1 &le; | x | < 2 0 | x | &GreaterEqual; 2
Compare arest neighbors and bilinear interpolation, the bicubic convolution method has bigger improvement to the image interpolation effect, and interpolation place numerical value and derivative are all continuous, and the details performance is clearer, has overcome the image flaw of above two kinds of methods to a certain extent.But owing to relate to a cube computing in the calculating formula, and the matrix convolution is arranged, so calculated amount is compared, and the above two are bigger.
Comparatively level and smooth zone in image, it is fine that above classic method such as Bilinear and Bicubic can handle.But for image-region at boundary, more than traditional interpolation method then have tangible flaw, as fuzzy, sawtooth etc.And the human visual system is very responsive for boundary characteristic, therefore makes this flaw cause bigger harmful effect to image visual effect.Therefore, realize a main target of better image interpolation is when interpolation, how to keep the characteristic of image at boundary better, avoid producing the various flaws of traditional interpolation method.In order to realize such target, need carry out comparatively reasonably modeling to picture signal.The difficult point of picture signal modeling is the unstable state characteristic of himself.Simultaneously, natural image often has geometrical rule property, and promptly image changes on boundary direction slowly, on perpendicular to the direction on border, changes rapidly.This character has reflected the segmentation statistic steady state characteristic of picture signal.According to a such characteristic, people such as Li have proposed a kind of covariance information that utilizes and have come implicit expression to comprise the adaptive interpolation method of boundary information.The high resolving power covariance information estimates through the low resolution covariance, the low resolution covariance information then is that statistic keeps stable in the hypothesis window, in a local window, samples and estimates.This method can be handled the border condition than large scale in the image well; But can't handle some border and textural characteristics well than small scale; Simultaneously because its method has only been utilized the neighbor pixel dot information on the diagonal; The interpolation sharpness on border is compared original image and is also had a certain distance, and has some false narrow and small edge effect image visual effects in the interpolation image, seems to lack the nature and the sense of reality.Zhang and Wu introduce into more pixel space correlationship; Not only utilized the correlationship of diagonal; And considered the correlationship on level and the vertical direction; Keep stable through as if statistics amount in a local window, this correlationship restriction is forced on the pixels all in window, estimate finally to obtain the value of one group of pixel to be inserted through piece optimization.This method has made full use of the correlationship between high low-resolution image, has adaptability preferably for the structural information of natural image, therefore bigger improvement people's such as Li algorithm.
More than two kinds of algorithms all be that hypothesis image statistics in selected window keeps stable; When image boundary characteristic dimension during less than window size; This hypothesis also is false; Therefore the estimation for statistic information in the window is inaccurate, causes the error of system model, thereby causes interpolation image flaw to occur at boundary.
Summary of the invention
Occur the problem of flaw in order to solve the prior art interpolation image at boundary, the embodiment of the invention provides a kind of image interpolation method that improves image resolution ratio, comprising:
Similarity probability between the partial structurtes of preset reference image vegetarian refreshments in the partial structurtes of each pixel and the selected window in the selected window of computed image;
Pixel value according to known low-resolution pixel point in the selected window is estimated first parameter and second parameter; First parameter is to be used for portraying in the window model parameter of neighbor pixel point regression relation on each pixel and its diagonal, and second parameter is the model parameter that is used for portraying neighbor pixel point regression relation on the perpendicular and horizontal line direction of interior each pixel of window;
Utilize the pixel value of low-resolution pixel point, the estimated result of first parameter and the estimated result of second parameter, set up the pixel value of interpolation full-resolution picture vegetarian refreshments in the selected window and the correlationship of matching error;
Set up the optimization aim function according to similarity probability and matching error, the optimization aim function is carried out the pixel value that optimum estimate obtains the full-resolution picture vegetarian refreshments;
The pixel value of the full-resolution picture vegetarian refreshments that obtains by optimum estimate inserts selected window with the similarity probability greater than the full-resolution picture vegetarian refreshments of setting threshold.
The embodiment of the invention also provides a kind of image interpolation device that improves image resolution ratio simultaneously, comprising:
The probability generation module is used for computed image and selectes the similarity probability between the partial structurtes of the partial structurtes of each pixel in the window and the reference image vegetarian refreshments that selected window is preset;
First estimation module; Be used for first parameter and second parameter being estimated according to the pixel value of known low-resolution pixel point in the selected window; First parameter is to be used for portraying in the window model parameter of neighbor pixel point regression relation on each pixel and its diagonal, and second parameter is the model parameter that is used for portraying neighbor pixel point regression relation on the perpendicular and horizontal line direction of interior each pixel of window;
Correlationship is set up module, is used to utilize the pixel value of low-resolution pixel point, the estimated result of first parameter and the estimated result of second parameter, sets up the pixel value of interpolation full-resolution picture vegetarian refreshments in the selected window and the correlationship of matching error;
Second estimation module is used for setting up the optimization aim function according to similarity probability and matching error, and the optimization aim function is carried out the pixel value that optimum estimate obtains the full-resolution picture vegetarian refreshments;
Insert module, the pixel value of the full-resolution picture vegetarian refreshments that is used for obtaining by optimum estimate inserts selected window with the similarity probability greater than the full-resolution picture vegetarian refreshments of setting threshold.
Can find out by specific embodiments provided by the invention, the weight when having introduced the similarity probability as the high-resolution pixel point estimation, and then make evaluated error minimize, reduced the flaw that boundary occurs.
Description of drawings
Fig. 1 is twice interpolation synoptic diagram provided by the invention;
Fig. 2 is the first embodiment method flow diagram provided by the invention;
Fig. 3 is the local window synoptic diagram with different similarity probability provided by the invention;
Fig. 4 is two right synoptic diagram of pixel with different similarity probability provided by the invention;
Fig. 5 a is that the low resolution neighbours on 4 diagonals utilizing each full-resolution picture vegetarian refreshments provided by the invention put pixel value and constitute partial structurtes vector synoptic diagram;
Fig. 5 b is that the low resolution neighbours on 4 diagonals utilizing each low-resolution pixel point provided by the invention put pixel value and constitute partial structurtes vector synoptic diagram;
Fig. 6 a has consistent model parameter synoptic diagram on level and the vertical direction between high low resolution pixel provided by the invention;
Fig. 6 b has consistent model parameter synoptic diagram on the diagonal between high low resolution pixel provided by the invention;
Fig. 7 is the configuration schematic diagram of the space correlation relation between interpolation neighbours' point provided by the invention;
Fig. 8 is the second embodiment system construction drawing provided by the invention.
Embodiment
In order to solve flaw appears in interpolation image in the prior art at boundary problem; The embodiment of the invention provides a kind of image interpolation method that improves image resolution ratio; In the scheme of present embodiment, the ground that is without loss of generality, regarding in the high-definition picture pixel in the low-resolution image as through rule the factor is that 2 down-sampling obtains; Therefore these pixels can be regarded the part in the original high resolution image as, and Fig. 1 has provided the interpolation synoptic diagram twice.The task of interpolation is selected 10 interpolation from existing low-resolution pixel exactly and is gone out first's full-resolution picture vegetarian refreshments 11 and second portion full-resolution picture vegetarian refreshments 12 so.Whole Interpolation Process was divided into for two steps.The first step is that interpolation obtains first's full-resolution picture vegetarian refreshments 11.Second step was to select 10 according to first's full-resolution picture vegetarian refreshments 11 that first step interpolation obtains with existing low-resolution pixel to come interpolation to obtain second portion full-resolution picture vegetarian refreshments 12.The algorithm that two steps handled is the same, only at rotation that 45 degree are arranged on the direction and the scaling on the yardstick.Therefore, the core objective of interpolation algorithm can be regarded as from existing low-resolution pixel and selects the process that 10 interpolation go out first's full-resolution picture vegetarian refreshments 11.
The image interpolation method of raising image resolution ratio provided by the invention sees also Fig. 2, comprises the steps:
Step 101, the similarity probability P between the partial structurtes of center pixel to be inserted in the partial structurtes of each pixel and the selected window in the selected window of computed image LAnd P H
For the ease of explanation; Present embodiment with center pixel to be inserted as reference image vegetarian refreshments preset in the selected window; But this is not limited, the reference image vegetarian refreshments also can be any pixel that comprises first pixel and last pixel in the selected window of image.Present embodiment is to be in the local window at center with center pixel to be inserted, and the pixel value of high-resolution pixel in this window is estimated.The local window with different similarity probability that is provided with is like this shown by Fig. 3, comprises first 21, second portion 22 and third part 23.In a such local window; Not all pixel all has the similar topography's characteristic of concentricity pixel to be inserted 20; Be partial structurtes and dissimilar; There are some possibly also have bigger difference; Wherein the pixel of first 21 is similar with local window center pixel partial structurtes to be inserted, and the pixel of second portion 22 and third part 23 and local window center pixel 20 partial structurtes to be inserted are dissimilar, therefore need the similarity difference that a kind of index is reacted partial structurtes between each pixel.This similarity difference can be measured by the similarity probability between them.
In the window in the partial structurtes of j pixel and the selected window computing formula as the similarity probability between the partial structurtes of d pixel of reference image vegetarian refreshments be
P (d wherein; J) for being the similarity probability between the partial structurtes of the center pixel to be inserted in the present embodiment as d pixel of reference image vegetarian refreshments in the partial structurtes of j pixel in the selected window and the selected window;
Figure BSA00000359809600062
Figure BSA00000359809600063
is the vector of the pixel value formation of order of the neighbours on four diagonals of center pixel to be inserted in the selected window; D is a positive integer; D≤m+n; N is the number of selected window middle high-resolution pixel; M is the number of low-resolution pixel point; ε is a positive number of avoiding removing zero overflow error;
Figure BSA00000359809600064
Figure BSA00000359809600065
is the vector of the pixel value formation of ordering of the neighbours on four diagonals of j pixel in the selected window; J is a positive integer; J≤m+n; H is used for the distribution of shapes parameter of control characteristic function, and Fig. 4 is two right synoptic diagram of pixel with different similarity probability.
According to above formula
Figure BSA00000359809600066
Can the similarity probability calculation between the concentricity pixel to be inserted of all pixels in the window be come out, form two probability vector P LAnd P H, wherein P L = ( p 1 L , . . . , p i L , . . . , p m L ) T , P H = ( p 1 H , . . . , p k H , . . . , p n H ) T .
Specific as follows; Utilize the low resolution neighbours' point on four diagonals of low-resolution pixel point; Similarity probability in partial structurtes that each known low-resolution pixel is selected in the selected window of calculating and the selected window between the partial structurtes of preset reference image vegetarian refreshments, and form low-resolution pixel point similarity probability vector P L, wherein
Figure BSA00000359809600071
Figure BSA00000359809600072
Represent in the 1st partial structurtes that low-resolution pixel is selected and the selected window similarity probability between the partial structurtes of preset reference image vegetarian refreshments,
Figure BSA00000359809600073
Represent the similarity probability i≤m between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes that i low-resolution pixel selected and the selected window,
Figure BSA00000359809600074
Represent the similarity probability between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes that m low-resolution pixel selected and the selected window.Utilize the low resolution neighbours' point on four diagonals of full-resolution picture vegetarian refreshments; Similarity probability between the partial structurtes of preset reference image vegetarian refreshments in the partial structurtes of each full-resolution picture vegetarian refreshments and the selected window in the selected window of calculating, and form full-resolution picture vegetarian refreshments probability vector P H, wherein
Figure BSA00000359809600075
Figure BSA00000359809600076
Represent the similarity probability between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes and the selected window of the 1st full-resolution picture vegetarian refreshments,
Figure BSA00000359809600077
Represent the similarity probability k≤n between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes and the selected window of k full-resolution picture vegetarian refreshments,
Figure BSA00000359809600078
Represent the similarity probability between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes and the selected window of n full-resolution picture vegetarian refreshments.When the similarity probability that calculates between full-resolution picture vegetarian refreshments and the center pixel to be inserted; Can directly utilize low resolution neighbours on 4 diagonals of each full-resolution picture vegetarian refreshments to select pixel value and constitute low resolution neighbours on 4 diagonals that partial structurtes vector
Figure BSA00000359809600079
utilizes each full-resolution picture vegetarian refreshments and put pixel value and constitute the partial structurtes vector shown in Fig. 5 a, directly utilize formula
Figure BSA000003598096000710
to calculate then.When the similarity probability that calculates between low-resolution pixel point and the center pixel to be inserted; Because it is the full-resolution picture vegetarian refreshments of the unknown that low-resolution pixel select that neighbours on the diagonal select; Can be according to the structural similarity between high low resolution hypothesis, utilize low resolution neighbours on its diagonal to put low resolution neighbours on 4 diagonals that pixel value constitutes
Figure BSA000003598096000711
utilizes each low-resolution pixel point and put pixel value and constitute the partial structurtes vector shown in Fig. 5 b.
Step 102; Pixel value according to known low-resolution pixel point in the selected window is estimated parameter a and parameter b; Parameter a is used for portraying in the window model parameter of neighbor pixel point regression relation on each pixel and its diagonal, and parameter b is the model parameter that is used for portraying neighbor pixel point regression relation on the perpendicular and horizontal line direction of interior each pixel of window.
In the present embodiment, utilize known low-resolution pixel to select and estimate to carry out modeling to the relation between them for unknown full-resolution picture vegetarian refreshments.Adopt two group model parameter a, b to come to portray respectively this space correlation relation.Represent the regression relation on the diagonal with
Figure BSA00000359809600081
; With the regression relation on
Figure BSA00000359809600082
expression level and the vertical direction; With t ∈ { 1; 2; 3,4} representes t neighbours on the respective direction.For example
Figure BSA00000359809600083
With
Figure BSA00000359809600084
Difference remarked pixel x iAnd y iT neighbours on diagonal;
Figure BSA00000359809600085
With
Figure BSA00000359809600086
Difference remarked pixel x iAnd y iT neighbours' point in the horizontal and vertical directions.Through supposing that the parameter between high low-resolution image is consistent, can estimate parameter a and b according to the pixel value of low-resolution image pixel in the local window.Between high low resolution pixel level with have consistent model parameter on the vertical direction shown in Fig. 6 a, parameter b can directly be carried out weighted least-squares and estimated to try to achieve its optimal value.Has consistent model parameter between high low resolution pixel on the diagonal shown in Fig. 6 b; According to the conforming hypothesis of model parameter between the high low resolution; Yardstick put be twice, adopt
Figure BSA00000359809600087
and come parameter a is estimated.
To parameter a, when parameter b is estimated, directly pass through Parameter a is carried out weighted least-squares estimate, wherein
Figure BSA00000359809600089
Represent the first parameter optimal value estimated result, a tRepresent a vectorial composition value with the parameter a of vector representation, t ∈ 1,2,3,4}, f (x i) i low-resolution pixel point x of expression iPixel value,
Figure BSA000003598096000810
Remarked pixel x iT low resolution neighbours point on diagonal
Figure BSA000003598096000811
Pixel value.Directly pass through
Figure BSA000003598096000812
Second parameter is carried out weighted least-squares method estimate, wherein
Figure BSA000003598096000813
Represent the second parameter optimal value estimated result, b tRepresent a vectorial composition value with second parameter of vector representation,
Figure BSA000003598096000814
Expression low-resolution pixel x iT neighbours' point on vertical and horizontal line direction
Figure BSA000003598096000815
Pixel value.
Step 103 is utilized the pixel value of low-resolution pixel point, the estimated result of parameter a and the estimated result of parameter b, sets up the pixel value of interpolation full-resolution picture vegetarian refreshments in the selected window and the correlationship of matching error.
Come the correlationship on the local diagonal of picture engraving through parameter
Figure BSA000003598096000816
, can get
Figure BSA00000359809600091
Figure BSA00000359809600092
portrays the level of topography and the correlationship on the vertical direction through parameter, can get and
Figure BSA00000359809600094
Describe in the integrating step 102, adopt two group model parameter a, b to come to portray respectively this space correlation relation.Represent the regression relation on the diagonal with
Figure BSA00000359809600095
; With the regression relation on expression level and the vertical direction; With t ∈ { 1; 2; 3; 4} representes t neighbours on the respective direction, and the configuration schematic diagram of space correlation relation is as shown in Figure 7 between interpolation neighbours' point.
Through the correlationship that use a model parameter a and b come the picture engraving local space, the matching error of above-mentioned expression can be expressed as following form:
Figure BSA00000359809600097
Figure BSA00000359809600098
Expression is passed through
Figure BSA00000359809600099
Portrayal full-resolution picture vegetarian refreshments y K 'With neighbor pixel point on its diagonal
Figure BSA000003598096000910
Matching error during regression relation, Q representes to have the neighbor pixel point on the selected window inner opposite angle line direction
Figure BSA000003598096000911
Full-resolution picture vegetarian refreshments y K 'Number,
Figure BSA000003598096000912
Figure BSA000003598096000913
Expression is passed through Portrayal low-resolution pixel point x I 'With neighbor pixel point on its diagonal
Figure BSA000003598096000915
Matching error during regression relation, P representes to have the neighbor pixel point on the selected window inner opposite angle line direction
Figure BSA000003598096000916
Low-resolution pixel point x I 'Number,
Figure BSA000003598096000917
Expression is passed through
Figure BSA000003598096000919
Portrayal full-resolution picture vegetarian refreshments y K "With neighbor pixel point on its diagonal
Figure BSA000003598096000920
Matching error during regression relation, K representes vertically and on the horizontal line direction have the neighbor pixel point in the selected window
Figure BSA000003598096000921
Full-resolution picture vegetarian refreshments y K "Number,
Figure BSA000003598096000922
Figure BSA000003598096000923
Expression is passed through
Figure BSA000003598096000924
Portrayal low-resolution pixel point x I "With neighbor pixel point on its diagonal
Figure BSA000003598096000925
Matching error during regression relation, S representes vertically and on the horizontal line direction have the neighbor pixel point in the selected window
Figure BSA000003598096000926
Low-resolution pixel point x I "Number.
Step 104 is according to the similarity probability P LAnd P HSet up the optimization aim function F with matching error, the optimization aim function F is carried out the best estimate that optimum estimate obtains the full-resolution picture vegetarian refreshments in the group window.
Set up optimization aim function
Figure BSA00000359809600101
satisfy simultaneously following constraint condition
Figure BSA00000359809600102
wherein F be the optimization aim function; The pixel value of f (y) expression full-resolution picture vegetarian refreshments, w representes selected window.Through finding the solution this Estimation Optimization problem, finally can obtain the best estimate of the full-resolution picture vegetarian refreshments in the group window.
The best estimate of full-resolution picture vegetarian refreshments promptly carries out the pixel value that optimum estimate obtains the full-resolution picture vegetarian refreshments to the optimization aim function.
Step 105, the pixel value of the full-resolution picture vegetarian refreshments that obtains by optimum estimate, the best estimate of output center pixel to be inserted.
Wherein, center pixel to be inserted is the main interpolation point of paying close attention in the present embodiment, therefore only exports the numerical value of this point.In order to reduce the time complexity of total algorithm; Can consider that the value of the point that those and center pixel similarity to be inserted probability is higher also exports as its final pixel value, be about to the similarity probability and insert selected window greater than the full-resolution picture vegetarian refreshments of setting threshold.
Through with local window in the slip of carrying out on the image on horizontal direction and the vertical direction, we can finally accomplish the process of image interpolation in the hope of the optimal estimation value of all pixels to be inserted.
Second embodiment provided by the invention is a kind of image interpolation device that improves image resolution ratio, and its structure is as shown in Figure 8, comprising:
Probability generation module 201 is used for computed image and selectes the similarity probability between the partial structurtes of the partial structurtes of each pixel in the window and the reference image vegetarian refreshments that selected window is preset;
First estimation module 202; Be used for first parameter and second parameter being estimated according to the pixel value of known low-resolution pixel point in the selected window; First parameter is to be used for portraying in the window model parameter of neighbor pixel point regression relation on each pixel and its diagonal, and second parameter is the model parameter that is used for portraying neighbor pixel point regression relation on the perpendicular and horizontal line direction of interior each pixel of window;
Correlationship is set up module 203, is used to utilize the pixel value of low-resolution pixel point, the estimated result of first parameter and the estimated result of second parameter, sets up the pixel value of interpolation full-resolution picture vegetarian refreshments in the selected window and the correlationship of matching error;
Second estimation module 204 is used for setting up the optimization aim function according to similarity probability and matching error, and the optimization aim function is carried out the pixel value that optimum estimate obtains the full-resolution picture vegetarian refreshments;
Insert module 205, the pixel value of the full-resolution picture vegetarian refreshments that is used for obtaining by optimum estimate inserts selected window with the similarity probability greater than the full-resolution picture vegetarian refreshments of setting threshold.
Further; Probability generation module 201; Also be used for carrying out the similarity probability calculation through
Figure BSA00000359809600111
; P (d wherein; J) in the partial structurtes of j pixel in the selected window and the selected window as the similarity probability between the partial structurtes of d pixel of reference image vegetarian refreshments;
Figure BSA00000359809600113
is the vector of the pixel value formation of ordering of the neighbours on four diagonals of d pixel in the selected window; D is a positive integer; D≤m+n; N is the number of selected window middle high-resolution pixel; M is the number of low-resolution pixel point; ε is a positive number of avoiding removing zero overflow error, be the vector of the pixel value formation of ordering of the neighbours on four diagonals of j pixel in the selected window, j is a positive integer; J≤m+n, h are used for the distribution of shapes parameter of control characteristic function.
Further; Probability generation module 201; Also be used to utilize the low resolution neighbours' point on four diagonals of low-resolution pixel point; Similarity probability in partial structurtes that each known low-resolution pixel is selected in the selected window of calculating and the selected window between the partial structurtes of preset reference image vegetarian refreshments, and form low-resolution pixel point similarity probability vector P L, wherein
Figure BSA00000359809600115
Represent in the 1st partial structurtes that low-resolution pixel is selected and the selected window similarity probability between the partial structurtes of preset reference image vegetarian refreshments,
Figure BSA00000359809600117
Represent the similarity probability i≤m between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes that i low-resolution pixel selected and the selected window,
Figure BSA00000359809600118
Represent the similarity probability between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes that m low-resolution pixel selected and the selected window; And
Utilize the low resolution neighbours' point on four diagonals of full-resolution picture vegetarian refreshments; Similarity probability between the partial structurtes of preset reference image vegetarian refreshments in the partial structurtes of each full-resolution picture vegetarian refreshments and the selected window in the selected window of calculating, and form full-resolution picture vegetarian refreshments probability vector P H, wherein
Figure BSA000003598096001110
Represent the similarity probability between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes and the selected window of the 1st full-resolution picture vegetarian refreshments,
Figure BSA00000359809600121
Represent the similarity probability k≤n between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes and the selected window of k full-resolution picture vegetarian refreshments,
Figure BSA00000359809600122
Represent the similarity probability between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes and the selected window of n full-resolution picture vegetarian refreshments.
Further, first estimation module 202 also is used for directly passing through
Figure BSA00000359809600123
First parameter is carried out weighted least-squares estimate, wherein
Figure BSA00000359809600124
Represent the first parameter optimal value estimated result, a tRepresent a vectorial composition value with first parameter of vector representation, t ∈ 1,2,3,4}, f (x i) i low-resolution pixel point x of expression iPixel value,
Figure BSA00000359809600125
Remarked pixel x iT low resolution neighbours point on diagonal
Figure BSA00000359809600126
Pixel value;
Directly pass through
Figure BSA00000359809600127
Second parameter is carried out weighted least-squares method estimate, wherein Represent the second parameter optimal value estimated result, b tRepresent a vectorial composition value with second parameter of vector representation,
Figure BSA00000359809600129
Expression low-resolution pixel x iT neighbours' point on vertical and horizontal line direction
Figure BSA000003598096001210
Pixel value.
Further, second estimation module 204 also is used to set up the optimization aim function
Figure BSA000003598096001211
Satisfy following constraint condition simultaneously
Figure BSA000003598096001212
Wherein F is the optimization aim function, the pixel value of f (y) expression full-resolution picture vegetarian refreshments, and w representes selected window,
Figure BSA000003598096001213
Expression is passed through
Figure BSA000003598096001215
Portrayal full-resolution picture vegetarian refreshments y K 'With neighbor pixel point on its diagonal Matching error during regression relation, Q representes to have the neighbor pixel point on the selected window inner opposite angle line direction
Figure BSA000003598096001217
Full-resolution picture vegetarian refreshments y K 'Number,
Figure BSA000003598096001218
Expression is passed through Portrayal low-resolution pixel point x I 'With neighbor pixel point on its diagonal
Figure BSA000003598096001221
Matching error during regression relation, P representes to have the neighbor pixel point on the selected window inner opposite angle line direction
Figure BSA00000359809600131
Low-resolution pixel point x I 'Number,
Figure BSA00000359809600133
Expression is passed through Portrayal full-resolution picture vegetarian refreshments y K "With neighbor pixel point on its diagonal Matching error during regression relation, K representes vertically and on the horizontal line direction have the neighbor pixel point in the selected window
Figure BSA00000359809600136
Full-resolution picture vegetarian refreshments y K "Number,
Figure BSA00000359809600137
Figure BSA00000359809600138
Expression is passed through Portrayal low-resolution pixel point x I "With neighbor pixel point on its diagonal
Figure BSA000003598096001310
Matching error during regression relation, S representes vertically and on the horizontal line direction have the neighbor pixel point in the selected window
Figure BSA000003598096001311
Low-resolution pixel point x I "Number.
Further, insert module 205 also is used for if the reference image vegetarian refreshments is the full-resolution picture vegetarian refreshments, will insert selected window as the full-resolution picture vegetarian refreshments of reference image vegetarian refreshments.
Obviously, those skilled in the art can carry out various changes and modification to the present invention and not break away from the spirit and scope of the present invention.Like this, belong within the scope of claim of the present invention and equivalent technologies thereof if of the present invention these are revised with modification, then the present invention also is intended to comprise these changes and modification interior.

Claims (12)

1. an image interpolation method that improves image resolution ratio is characterized in that, comprising:
Similarity probability between the partial structurtes of preset reference image vegetarian refreshments in the partial structurtes of each pixel and the selected window in the selected window of computed image;
Pixel value according to known low-resolution pixel point in the selected window is estimated first parameter and second parameter; First parameter is to be used for portraying in the window model parameter of neighbor pixel point regression relation on each pixel and its diagonal, and second parameter is the model parameter that is used for portraying neighbor pixel point regression relation on the perpendicular and horizontal line direction of interior each pixel of window;
Utilize the pixel value of low-resolution pixel point, the estimated result of first parameter and the estimated result of second parameter, set up the pixel value of interpolation full-resolution picture vegetarian refreshments in the selected window and the correlationship of matching error;
Set up the optimization aim function according to similarity probability and matching error, the optimization aim function is carried out the pixel value that optimum estimate obtains the full-resolution picture vegetarian refreshments;
The pixel value of the full-resolution picture vegetarian refreshments that obtains by optimum estimate inserts selected window with the similarity probability greater than the full-resolution picture vegetarian refreshments of setting threshold.
2. the method for claim 1 is characterized in that, the similarity probability between the partial structurtes of the reference image vegetarian refreshments of presetting in the partial structurtes of each pixel and the selected window in the selected window of computed image is specially:
Carry out the similarity probability calculation through ; P (d wherein; J) in the partial structurtes of j pixel in the selected window and the selected window as the similarity probability between the partial structurtes of d pixel of reference image vegetarian refreshments;
Figure FSA00000359809500012
Figure FSA00000359809500013
is the vector of the pixel value formation of ordering of the neighbours on four diagonals of d pixel in the selected window; D is a positive integer; D≤m+n; N is the number of selected window middle high-resolution pixel; M is the number of low-resolution pixel point; ε is a positive number of avoiding removing zero overflow error;
Figure FSA00000359809500014
Figure FSA00000359809500015
is the vector of the pixel value formation of ordering of the neighbours on four diagonals of j pixel in the selected window; J is a positive integer; J≤m+n, h are used for the distribution of shapes parameter of control characteristic function.
3. method as claimed in claim 2 is characterized in that, carries out the similarity probability calculation and is specially:
Utilize the low resolution neighbours' point on four diagonals of low-resolution pixel point; Similarity probability in partial structurtes that each known low-resolution pixel is selected in the selected window of calculating and the selected window between the partial structurtes of preset reference image vegetarian refreshments, and form low-resolution pixel point similarity probability vector P L, wherein
Figure FSA00000359809500021
Figure FSA00000359809500022
Represent in the 1st partial structurtes that low-resolution pixel is selected and the selected window similarity probability between the partial structurtes of preset reference image vegetarian refreshments,
Figure FSA00000359809500023
Represent the similarity probability i≤m between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes that i low-resolution pixel selected and the selected window,
Figure FSA00000359809500024
Represent the similarity probability between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes that m low-resolution pixel selected and the selected window; And
Utilize the low resolution neighbours' point on four diagonals of full-resolution picture vegetarian refreshments; Similarity probability between the partial structurtes of preset reference image vegetarian refreshments in the partial structurtes of each full-resolution picture vegetarian refreshments and the selected window in the selected window of calculating, and form full-resolution picture vegetarian refreshments probability vector P H, wherein
Figure FSA00000359809500025
Figure FSA00000359809500026
Represent the similarity probability between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes and the selected window of the 1st full-resolution picture vegetarian refreshments,
Figure FSA00000359809500027
Represent the similarity probability k≤n between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes and the selected window of k full-resolution picture vegetarian refreshments,
Figure FSA00000359809500028
Represent the similarity probability between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes and the selected window of n full-resolution picture vegetarian refreshments.
4. method as claimed in claim 3 is characterized in that, according to the pixel value of known low-resolution pixel point in the selected window first parameter and second parameter is estimated to be specially:
Directly pass through First parameter is carried out weighted least-squares estimate, wherein
Figure FSA000003598095000210
Represent the first parameter optimal value estimated result, a iRepresent a vectorial composition value with first parameter of vector representation, t ∈ 1,2,3,4}, f (x i) i low-resolution pixel point x of expression iPixel value,
Figure FSA000003598095000211
Remarked pixel x iT low resolution neighbours point on diagonal
Figure FSA000003598095000212
Pixel value;
Directly pass through
Figure FSA00000359809500031
Second parameter is carried out weighted least-squares method estimate, wherein
Figure FSA00000359809500032
Represent the second parameter optimal value estimated result, b iRepresent a vectorial composition value with second parameter of vector representation,
Figure FSA00000359809500033
Expression low-resolution pixel x jT neighbours' point on vertical and horizontal line direction
Figure FSA00000359809500034
Pixel value.
5. method as claimed in claim 4 is characterized in that, sets up the optimization aim function according to similarity probability and matching error and is specially:
Set up the optimization aim function
Figure FSA00000359809500035
Satisfy following constraint condition simultaneously
Figure FSA00000359809500036
Wherein F is the optimization aim function, the pixel value of f (y) expression full-resolution picture vegetarian refreshments, and w representes selected window,
Figure FSA00000359809500038
Expression is passed through
Figure FSA00000359809500039
Portrayal full-resolution picture vegetarian refreshments y K 'With neighbor pixel point on its diagonal
Figure FSA000003598095000310
Matching error during regression relation, Q representes to have the neighbor pixel point on the selected window inner opposite angle line direction
Figure FSA000003598095000311
Full-resolution picture vegetarian refreshments y K 'Number,
Figure FSA000003598095000312
Expression is passed through
Figure FSA000003598095000314
Portrayal low-resolution pixel point x I 'With neighbor pixel point on its diagonal
Figure FSA000003598095000315
Matching error during regression relation, P representes to have the neighbor pixel point on the selected window inner opposite angle line direction
Figure FSA000003598095000316
Low-resolution pixel point x I 'Number,
Figure FSA000003598095000317
Figure FSA000003598095000318
Expression is passed through
Figure FSA000003598095000319
Portrayal full-resolution picture vegetarian refreshments y K "With neighbor pixel point on its diagonal
Figure FSA000003598095000320
Matching error during regression relation, K representes vertically and on the horizontal line direction have the neighbor pixel point in the selected window
Figure FSA000003598095000321
Full-resolution picture vegetarian refreshments y K "Number,
Figure FSA000003598095000322
Figure FSA000003598095000323
Expression is passed through
Figure FSA000003598095000324
Portrayal low-resolution pixel point x I "With neighbor pixel point on its diagonal
Figure FSA000003598095000325
Matching error during regression relation, S representes vertically and on the horizontal line direction have the neighbor pixel point in the selected window
Figure FSA000003598095000326
Low-resolution pixel point x I "Number.
6. the method for claim 1 is characterized in that, the pixel value of the full-resolution picture vegetarian refreshments that obtains by optimum estimate inserts selected window with the similarity probability greater than the full-resolution picture vegetarian refreshments of setting threshold and is specially:
If the reference image vegetarian refreshments is the full-resolution picture vegetarian refreshments, will insert selected window as the full-resolution picture vegetarian refreshments of reference image vegetarian refreshments.
7. an image interpolation device that improves image resolution ratio is characterized in that, comprising:
The probability generation module is used for computed image and selectes the similarity probability between the partial structurtes of the partial structurtes of each pixel in the window and the reference image vegetarian refreshments that selected window is preset;
First estimation module; Be used for first parameter and second parameter being estimated according to the pixel value of known low-resolution pixel point in the selected window; First parameter is to be used for portraying in the window model parameter of neighbor pixel point regression relation on each pixel and its diagonal, and second parameter is the model parameter that is used for portraying neighbor pixel point regression relation on the perpendicular and horizontal line direction of interior each pixel of window;
Correlationship is set up module, is used to utilize the pixel value of low-resolution pixel point, the estimated result of first parameter and the estimated result of second parameter, sets up the pixel value of interpolation full-resolution picture vegetarian refreshments in the selected window and the correlationship of matching error;
Second estimation module is used for setting up the optimization aim function according to similarity probability and matching error, and the optimization aim function is carried out the pixel value that optimum estimate obtains the full-resolution picture vegetarian refreshments;
Insert module, the pixel value of the full-resolution picture vegetarian refreshments that is used for obtaining by optimum estimate inserts selected window with the similarity probability greater than the full-resolution picture vegetarian refreshments of setting threshold.
8. device as claimed in claim 7; It is characterized in that; The probability generation module; Also be used for carrying out the similarity probability calculation through
Figure FSA00000359809500041
; P (d wherein; J) in the partial structurtes of j pixel in the selected window and the selected window as the similarity probability between the partial structurtes of d pixel of reference image vegetarian refreshments;
Figure FSA00000359809500042
Figure FSA00000359809500043
is the vector of the pixel value formation of ordering of the neighbours on four diagonals of d pixel in the selected window; D is a positive integer; D≤m+n; N is the number of selected window middle high-resolution pixel; M is the number of low-resolution pixel point; ε is a positive number of avoiding removing zero overflow error;
Figure FSA00000359809500044
Figure FSA00000359809500045
is the vector of the pixel value formation of ordering of the neighbours on four diagonals of j pixel in the selected window; J is a positive integer, and j≤m+n, h are used for the distribution of shapes parameter of control characteristic function.
9. device as claimed in claim 8; It is characterized in that; The probability generation module; Also be used to utilize the low resolution neighbours' point on four diagonals of low-resolution pixel point, calculate the similarity probability between the partial structurtes of reference image vegetarian refreshments preset in each known low-resolution pixel select in the selected window partial structurtes and the selected window, and formation low-resolution pixel point similarity probability vector P L, wherein
Figure FSA00000359809500052
Represent in the 1st partial structurtes that low-resolution pixel is selected and the selected window similarity probability between the partial structurtes of preset reference image vegetarian refreshments,
Figure FSA00000359809500053
Represent the similarity probability i≤m between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes that i low-resolution pixel selected and the selected window,
Figure FSA00000359809500054
Represent the similarity probability between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes that m low-resolution pixel selected and the selected window; And
Utilize the low resolution neighbours' point on four diagonals of full-resolution picture vegetarian refreshments; Similarity probability between the partial structurtes of preset reference image vegetarian refreshments in the partial structurtes of each full-resolution picture vegetarian refreshments and the selected window in the selected window of calculating, and form full-resolution picture vegetarian refreshments probability vector P H, wherein Represent the similarity probability between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes and the selected window of the 1st full-resolution picture vegetarian refreshments,
Figure FSA00000359809500057
Represent the similarity probability k≤n between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes and the selected window of k full-resolution picture vegetarian refreshments,
Figure FSA00000359809500058
Represent the similarity probability between the partial structurtes of reference image vegetarian refreshments preset in partial structurtes and the selected window of n full-resolution picture vegetarian refreshments.
10. device as claimed in claim 9 is characterized in that, first estimation module also is used for directly passing through
Figure FSA00000359809500059
First parameter is carried out weighted least-squares estimate, wherein
Figure FSA000003598095000510
Represent the first parameter optimal value estimated result, a tRepresent a vectorial composition value with first parameter of vector representation, t ∈ 1,2,3,4}, f (x i) i low-resolution pixel point x of expression iPixel value, Remarked pixel x iT low resolution neighbours point on diagonal
Figure FSA000003598095000512
Pixel value;
Directly pass through
Figure FSA00000359809500061
Second parameter is carried out weighted least-squares method estimate, wherein
Figure FSA00000359809500062
Represent the second parameter optimal value estimated result, b tRepresent a vectorial composition value with second parameter of vector representation,
Figure FSA00000359809500063
Expression low-resolution pixel x iT neighbours' point on vertical and horizontal line direction
Figure FSA00000359809500064
Pixel value.
11. device as claimed in claim 10 is characterized in that, second estimation module also is used to set up the optimization aim function
Figure FSA00000359809500065
Satisfy following constraint condition simultaneously
Figure FSA00000359809500066
Wherein F is the optimization aim function, the pixel value of f (y) expression full-resolution picture vegetarian refreshments, and w representes selected window,
Figure FSA00000359809500067
Figure FSA00000359809500068
Expression is passed through Portrayal full-resolution picture vegetarian refreshments y K 'With neighbor pixel point on its diagonal
Figure FSA000003598095000610
Matching error during regression relation, Q representes to have the neighbor pixel point on the selected window inner opposite angle line direction
Figure FSA000003598095000611
Full-resolution picture vegetarian refreshments y K 'Number,
Figure FSA000003598095000612
Figure FSA000003598095000613
Expression is passed through
Figure FSA000003598095000614
Portrayal low-resolution pixel point x I 'With neighbor pixel point on its diagonal
Figure FSA000003598095000615
Matching error during regression relation, P representes to have the neighbor pixel point on the selected window inner opposite angle line direction
Figure FSA000003598095000616
Low-resolution pixel point x I 'Number,
Figure FSA000003598095000617
Expression is passed through
Figure FSA000003598095000619
Portrayal full-resolution picture vegetarian refreshments y K "With neighbor pixel point on its diagonal
Figure FSA000003598095000620
Matching error during regression relation, K representes vertically and on the horizontal line direction have the neighbor pixel point in the selected window
Figure FSA000003598095000621
Full-resolution picture vegetarian refreshments y K "Number,
Figure FSA000003598095000622
Figure FSA000003598095000623
Expression is passed through
Figure FSA000003598095000624
Portrayal low-resolution pixel point x I "With neighbor pixel point on its diagonal
Figure FSA000003598095000625
Matching error during regression relation, S representes vertically and on the horizontal line direction have the neighbor pixel point in the selected window
Figure FSA000003598095000626
Low-resolution pixel point x I "Number.
12. device as claimed in claim 11 is characterized in that, insert module also is used for if the reference image vegetarian refreshments is the full-resolution picture vegetarian refreshments, will insert selected window as the full-resolution picture vegetarian refreshments of reference image vegetarian refreshments.
CN201010558745.1A 2010-11-25 2010-11-25 Image interpolation method capable of improving image resolution and device utilizing image interpolation method Expired - Fee Related CN102479380B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201010558745.1A CN102479380B (en) 2010-11-25 2010-11-25 Image interpolation method capable of improving image resolution and device utilizing image interpolation method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201010558745.1A CN102479380B (en) 2010-11-25 2010-11-25 Image interpolation method capable of improving image resolution and device utilizing image interpolation method

Publications (2)

Publication Number Publication Date
CN102479380A true CN102479380A (en) 2012-05-30
CN102479380B CN102479380B (en) 2014-01-15

Family

ID=46092013

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201010558745.1A Expired - Fee Related CN102479380B (en) 2010-11-25 2010-11-25 Image interpolation method capable of improving image resolution and device utilizing image interpolation method

Country Status (1)

Country Link
CN (1) CN102479380B (en)

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102946522A (en) * 2012-10-22 2013-02-27 华为技术有限公司 Super-resolution method and device based on non-local means
CN103179424A (en) * 2013-03-26 2013-06-26 深圳市掌网立体时代视讯技术有限公司 Method and device for synthesising three-dimensional image
WO2014082523A1 (en) * 2012-11-28 2014-06-05 北京大学 Image interpolation method and device based on autoregressive model
CN104715450A (en) * 2013-12-16 2015-06-17 方正国际软件(北京)有限公司 Method and system for extracting icons from picture
CN105069747A (en) * 2015-07-13 2015-11-18 中国科学院遥感与数字地球研究所 Image interpolation method and device
CN105096247A (en) * 2014-05-09 2015-11-25 北京大学 Image interpolation method and device
CN105719238A (en) * 2016-01-15 2016-06-29 电子科技大学 Image interpolation method based on non-local geometrical similarities
CN106558024A (en) * 2015-09-28 2017-04-05 北京大学 Carry out the autoregression interpolation method and device of self-adapting window expansion
CN112416915A (en) * 2020-11-11 2021-02-26 北京大学 Data correction method, device, storage medium and equipment
CN113643341A (en) * 2021-10-12 2021-11-12 四川大学 Different-scale target image registration method based on resolution self-adaptation
CN113850719A (en) * 2021-06-16 2021-12-28 北京大学 RGB image guided depth map super-resolution method based on joint implicit image function

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020012464A1 (en) * 2000-06-12 2002-01-31 Samsung Electronics Co., Ltd. Two-dimensional non-linear interpolation system based on edge information and two-dimensional mixing interpolation system using the same
CN101226634A (en) * 2008-01-30 2008-07-23 哈尔滨工业大学 Method for reestablishment of single frame image quick super-resolution based on nucleus regression
CN101668196A (en) * 2009-09-25 2010-03-10 西安电子科技大学 Low code rate image compression method based on down sampling and interpolation
CN101674478A (en) * 2009-10-19 2010-03-17 浙江大学 Video super-resolution method based on compressed domain

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020012464A1 (en) * 2000-06-12 2002-01-31 Samsung Electronics Co., Ltd. Two-dimensional non-linear interpolation system based on edge information and two-dimensional mixing interpolation system using the same
CN101226634A (en) * 2008-01-30 2008-07-23 哈尔滨工业大学 Method for reestablishment of single frame image quick super-resolution based on nucleus regression
CN101668196A (en) * 2009-09-25 2010-03-10 西安电子科技大学 Low code rate image compression method based on down sampling and interpolation
CN101674478A (en) * 2009-10-19 2010-03-17 浙江大学 Video super-resolution method based on compressed domain

Cited By (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102946522A (en) * 2012-10-22 2013-02-27 华为技术有限公司 Super-resolution method and device based on non-local means
CN102946522B (en) * 2012-10-22 2015-08-19 华为技术有限公司 Based on super-resolution method and the equipment of non-local mean
CN103854254B (en) * 2012-11-28 2016-12-21 北京大学 A kind of arbitrary proportion image interpolation method based on autoregression model
WO2014082523A1 (en) * 2012-11-28 2014-06-05 北京大学 Image interpolation method and device based on autoregressive model
CN103854254A (en) * 2012-11-28 2014-06-11 北京大学 Any ratio image interpolation method based on autoregression model
US9672590B2 (en) 2012-11-28 2017-06-06 Peking University Founder Group Co. Ltd. Image interpolation method and device based on autoregressive model
CN103179424A (en) * 2013-03-26 2013-06-26 深圳市掌网立体时代视讯技术有限公司 Method and device for synthesising three-dimensional image
CN103179424B (en) * 2013-03-26 2015-03-11 深圳市掌网立体时代视讯技术有限公司 Method and device for synthesising three-dimensional image
CN104715450A (en) * 2013-12-16 2015-06-17 方正国际软件(北京)有限公司 Method and system for extracting icons from picture
CN105096247A (en) * 2014-05-09 2015-11-25 北京大学 Image interpolation method and device
CN105096247B (en) * 2014-05-09 2018-02-06 北京大学 Image interpolation method and device
CN105069747A (en) * 2015-07-13 2015-11-18 中国科学院遥感与数字地球研究所 Image interpolation method and device
CN105069747B (en) * 2015-07-13 2018-07-20 中国科学院遥感与数字地球研究所 Image interpolation method and device
CN106558024A (en) * 2015-09-28 2017-04-05 北京大学 Carry out the autoregression interpolation method and device of self-adapting window expansion
CN106558024B (en) * 2015-09-28 2019-02-01 北京大学 Carry out the autoregression interpolation method and device of self-adapting window expansion
CN105719238A (en) * 2016-01-15 2016-06-29 电子科技大学 Image interpolation method based on non-local geometrical similarities
CN105719238B (en) * 2016-01-15 2018-11-02 电子科技大学 A kind of image interpolation method based on non local geometric similarity
CN112416915A (en) * 2020-11-11 2021-02-26 北京大学 Data correction method, device, storage medium and equipment
CN113850719A (en) * 2021-06-16 2021-12-28 北京大学 RGB image guided depth map super-resolution method based on joint implicit image function
CN113850719B (en) * 2021-06-16 2024-06-18 北京大学 RGB image guided depth map super-resolution method based on joint implicit image function
CN113643341A (en) * 2021-10-12 2021-11-12 四川大学 Different-scale target image registration method based on resolution self-adaptation

Also Published As

Publication number Publication date
CN102479380B (en) 2014-01-15

Similar Documents

Publication Publication Date Title
CN102479380B (en) Image interpolation method capable of improving image resolution and device utilizing image interpolation method
CN108648161B (en) Binocular vision obstacle detection system and method of asymmetric kernel convolution neural network
US9524539B2 (en) Method for edge-directed adaptive image interpolation and VLSI implementation device for the same
CN101901343B (en) Remote sensing image road extracting method based on stereo constraint
CN102006425B (en) Method for splicing video in real time based on multiple cameras
Lee et al. Local disparity estimation with three-moded cross census and advanced support weight
US20150097827A1 (en) Target Region Fill Utilizing Transformations
CN111899295B (en) Monocular scene depth prediction method based on deep learning
CN103455991A (en) Multi-focus image fusion method
CN106408513A (en) Super-resolution reconstruction method of depth map
CN109472267A (en) Picture similarity alignment algorithm
EP2466901B1 (en) Depth data upsampling
Hua et al. Extended guided filtering for depth map upsampling
CN107578430A (en) A kind of solid matching method based on adaptive weight and local entropy
CN106934806A (en) It is a kind of based on text structure without with reference to figure fuzzy region dividing method out of focus
CN109741358B (en) Superpixel segmentation method based on adaptive hypergraph learning
CN103679680B (en) Solid matching method and system
CN106530336A (en) Stereo matching algorithm based on color information and graph-cut theory
CN106257537A (en) A kind of spatial depth extracting method based on field information
CN115631223A (en) Multi-view stereo reconstruction method based on self-adaptive learning and aggregation
CN110415169A (en) A kind of depth map super resolution ratio reconstruction method, system and electronic equipment
CN111369435B (en) Color image depth up-sampling method and system based on self-adaptive stable model
CN106056575B (en) A kind of image matching method based on like physical property proposed algorithm
Damjanović et al. Local stereo matching using adaptive local segmentation
CN108364255A (en) Remote sensing image amplification method based on rarefaction representation and Partial Differential Equation Model

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
TR01 Transfer of patent right

Effective date of registration: 20220623

Address after: 100871 No. 5, the Summer Palace Road, Beijing, Haidian District

Patentee after: Peking University

Patentee after: New founder holdings development Co.,Ltd.

Patentee after: BEIJING FOUNDER ELECTRONICS Co.,Ltd.

Address before: 100871 No. 5, the Summer Palace Road, Beijing, Haidian District

Patentee before: Peking University

Patentee before: PEKING UNIVERSITY FOUNDER GROUP Co.,Ltd.

Patentee before: BEIJING FOUNDER ELECTRONICS Co.,Ltd.

TR01 Transfer of patent right
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20140115

CF01 Termination of patent right due to non-payment of annual fee