CN106920213B - Method and system for acquiring high-resolution image - Google Patents
Method and system for acquiring high-resolution image Download PDFInfo
- Publication number
- CN106920213B CN106920213B CN201710036911.3A CN201710036911A CN106920213B CN 106920213 B CN106920213 B CN 106920213B CN 201710036911 A CN201710036911 A CN 201710036911A CN 106920213 B CN106920213 B CN 106920213B
- Authority
- CN
- China
- Prior art keywords
- resolution image
- pixel
- image
- low
- resolution
- 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 35
- 238000012545 processing Methods 0.000 claims abstract description 33
- 238000005070 sampling Methods 0.000 claims description 35
- 230000003190 augmentative effect Effects 0.000 claims description 26
- 239000011159 matrix material Substances 0.000 claims description 21
- 238000010276 construction Methods 0.000 claims description 3
- 230000007547 defect Effects 0.000 abstract description 6
- 238000012805 post-processing Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000014509 gene expression Effects 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 101100161752 Mus musculus Acot11 gene Proteins 0.000 description 1
- OAICVXFJPJFONN-UHFFFAOYSA-N Phosphorus Chemical compound [P] OAICVXFJPJFONN-UHFFFAOYSA-N 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
Images
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
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a method and a system for acquiring a high-resolution image. The method comprises the following steps: acquiring a plurality of low-resolution images; carrying out interpolation processing on the plurality of low resolutions to obtain an interpolated high resolution image; establishing a denoising model of an L1 fidelity term with total variation regularization; and processing the interpolated high-resolution image by adopting the denoising model to obtain a denoised high-resolution image. The method and the system for acquiring the high-resolution image can overcome the defect of fuzzy boundary after image interpolation and effectively improve the resolution of the image.
Description
Technical Field
The invention relates to the field of image processing, in particular to a method and a system for acquiring a high-resolution image.
Background
High resolution imaging is an important subject of study in remote sensing images, but is limited by the size of the optical system, such as the imaging aperture, which results in a limitation of obtaining higher resolution images by increasing the focal length of the optical system. The SPOT5 remote sensing satellite launched in 2002 adopts a sub-pixel sampling technology for the first time, a series of low-resolution images are generated by utilizing staggered sensors, and images with higher resolution are obtained through a high-resolution reconstruction algorithm. On the basis, various oblique sampling modes are researched to carry out sub-pixel level sampling.
On the other hand, a series of low-resolution images which can be obtained by the oblique sampling system can obtain a spatial high-resolution image with higher resolution through a proper image super-resolution reconstruction algorithm, so that a high-quality image is provided for practical application. The current common methods are as follows: an image domain or wavelet domain interpolation method, a convex set projection iteration method, Landweber iteration and an acceleration algorithm thereof, a maximum posterior probability method and the like.
Due to the particularities of the oblique sampling mode, in which the geometrical relationship between the images is known, it can be assumed that the images obtained by the oblique sampling system are low resolution images that have been registered with high accuracy, without considering the errors of the sampling mode itself, and the noise level is low. The main idea of the image domain interpolation method is to select one image in the sequence images as a reference image, map other sequence images to a high-resolution grid corresponding to the reference image according to motion parameters obtained by registration to obtain non-uniformly distributed space sampling, and then perform non-uniform interpolation processing on the obtained non-uniformly distributed space sampling image so as to obtain pixel values of all integer grid points. The image domain non-uniform interpolation method has the advantages of simple algorithm, suitability for on-line operation and the like, but the obtained high-resolution image can cause partial blurring at the boundary of the image (namely, the place with large gray value difference between adjacent pixel points), and the blurring does not have a specific blurring kernel, so that the general blurring kernel estimation method is not applicable and influences the quality of the high-resolution image.
Disclosure of Invention
The invention aims to provide a method and a system for acquiring a high-resolution image, which can overcome the defect of fuzzy boundary after image interpolation and effectively improve the resolution of the image.
In order to achieve the purpose, the invention provides the following scheme:
a method of high resolution image acquisition, the method comprising:
acquiring a plurality of low-resolution images;
carrying out interpolation processing on the plurality of low resolutions to obtain an interpolated high resolution image;
establishing a denoising model of an L1 fidelity term with total variation regularization;
and processing the interpolated high-resolution image by adopting the denoising model to obtain a denoised high-resolution image.
Optionally, the acquiring a plurality of low-resolution images specifically includes:
performing oblique sampling on the image to obtain an oblique sampling image;
and acquiring a plurality of oblique sampling images.
Optionally, the interpolating the plurality of low-resolution images to obtain an interpolated high-resolution image specifically includes:
constructing a high-resolution image grid;
and calculating the pixel value of each grid point in the high-resolution image by adopting an interpolation algorithm according to the low-resolution image to obtain an interpolated high-resolution image.
Optionally, the calculating, according to the low-resolution image, a pixel value of each grid point in the high-resolution image by using an interpolation algorithm to obtain an interpolated high-resolution image specifically includes:
selecting a pixel point in one low-resolution image near the grid point, and marking the pixel point as a first adjacent pixel point;
selecting three pixel points adjacent to the adjacent pixel points in the low pixel image where the adjacent pixel points are located, marking the three pixel points as second adjacent pixel points, and enabling the first adjacent pixel points and the second adjacent pixel points to form a square area;
obtaining regional pixel points in the low-resolution image except the low-resolution image where the first adjacent pixel point is located, wherein the regional pixel points are pixel points located in the square region;
acquiring pixel values of the first adjacent pixel point, the second adjacent pixel point and the regional pixel point;
and calculating the pixel values of the grid points by adopting a bilinear interpolation function according to the pixel values of the first adjacent pixel points, the second adjacent pixel points and the regional pixel points and the positions of the grid points.
Optionally, the establishing a denoising model with a total variation regularization L1 fidelity term specifically includes:
establishing a denoising model of an L1 fidelity term with total variation regularization
Wherein X is a pixel value matrix of the high-resolution image to be solved, Y is a pixel value matrix of the interpolated high-resolution image,
α is a regularization parameter, # Xi,jIs a value for matrix ^ X row ith column jth, where ^ represents the gradient operator.
Optionally, the processing the interpolated high-resolution image by using the denoising model specifically includes:
determining an augmented Lagrangian function of the denoising model;
obtaining an iterative formula of the augmented Lagrangian function;
and substituting the pixel value of the interpolated high-resolution image into the iterative formula of the augmented Lagrange function to obtain the pixel value of the denoised high-resolution image.
The invention also provides a system for improving image resolution, comprising:
the low-resolution image acquisition module is used for acquiring a plurality of low-resolution images;
the interpolation processing module is used for carrying out interpolation processing on a plurality of low resolutions to obtain an interpolated high resolution image;
the denoising model establishing module is used for establishing a denoising model with a total variation regularized L1 fidelity term;
and the denoising module is used for processing the interpolated high-resolution image by adopting the denoising model to obtain a denoised high-resolution image.
Optionally, the low-resolution image obtaining module specifically includes:
the oblique sampling unit is used for performing oblique sampling on the image to obtain an oblique sampling image;
and the oblique sampling image acquisition unit is used for acquiring a plurality of oblique sampling images.
Optionally, the interpolation processing module specifically includes:
a high resolution image grid construction unit for constructing a high resolution image grid;
and the high-resolution image determining unit is used for calculating the pixel value of each grid point in the high-resolution image by adopting an interpolation algorithm according to the low-resolution image to obtain an interpolated high-resolution image.
Optionally, the high-resolution image determining unit specifically includes:
a first adjacent pixel point selecting subunit, configured to select a pixel point in one of the low-resolution images near the grid point, and mark the pixel point as a first adjacent pixel point;
a second adjacent pixel point selecting subunit, configured to select, in the low-pixel image where the adjacent pixel point is located, three pixel points adjacent to the adjacent pixel point, and mark the three pixel points as second adjacent pixel points, where the first adjacent pixel point and the second adjacent pixel point form a square region;
a regional pixel point obtaining subunit, configured to obtain a regional pixel point in the low-resolution image except the low-resolution image where the first adjacent pixel point is located, where the regional pixel point is a pixel point located in the square region;
the pixel value acquisition subunit is used for acquiring the pixel values of the first adjacent pixel point, the second adjacent pixel point and the regional pixel point;
and the grid point pixel value operator unit is used for calculating the pixel values of the grid points by adopting a bilinear interpolation function according to the pixel values of the first adjacent pixel points, the second adjacent pixel points and the regional pixel points and the positions of the grid points.
Optionally, the denoising model establishing module specifically includes:
a denoising model establishing unit for establishing a denoising model with a total variation regularized L1 fidelity term
Wherein X is a pixel value matrix of the high-resolution image to be solved, Y is a pixel value matrix of the interpolated high-resolution image,
α is a regularization parameter, # Xi,jIs a value for matrix ^ X row ith column jth, where ^ represents the gradient operator.
Optionally, the denoising module specifically includes:
the augmented Lagrange function determining unit is used for determining the augmented Lagrange function of the denoising model;
the iterative formula obtaining unit is used for obtaining an iterative formula of the augmented Lagrangian function;
and the high-resolution image pixel value unit is used for substituting the pixel value of the interpolated high-resolution image into the iterative formula of the augmented Lagrange function to obtain the pixel value of the denoised high-resolution image.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects: according to the invention, the interpolation processing is carried out on the low-resolution image, and the denoising model with the L1 fidelity term of total variation regularization is adopted to carry out the denoising processing on the interpolated high-resolution image, so that the defect of fuzzy boundary of the interpolated high-resolution image in the prior art is overcome, and the resolution of the image is effectively improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
FIG. 1 is a schematic flow chart of a method for obtaining a high resolution image according to an embodiment of the present invention;
FIG. 2 is five low resolution sequence target images in an embodiment of the present invention;
FIG. 3 is an image obtained by an interpolation algorithm in an embodiment of the present invention;
FIG. 4 is an image obtained by the ADMM post-processing algorithm in an embodiment of the invention;
FIG. 5 is an enlarged image of a portion of a low resolution target image, an interpolated image, and an ADMM post-processing image in accordance with an embodiment of the present invention;
fig. 6 is a schematic structural diagram of a system for acquiring a high-resolution image according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention aims to provide a method and a system for acquiring a high-resolution image, which can overcome the defect of fuzzy boundary after image interpolation and effectively improve the resolution of the image.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Fig. 1 is a schematic flow chart of a method for acquiring a high-resolution image according to an embodiment of the present invention, and as shown in fig. 1, the method for acquiring a high-resolution image includes the following steps:
step 101: acquiring a plurality of low-resolution images;
step 102: carrying out interpolation processing on the plurality of low resolutions to obtain an interpolated high resolution image;
step 103: establishing a denoising model of an L1 fidelity term with total variation regularization;
step 104: and processing the interpolated high-resolution image by adopting the denoising model to obtain a denoised high-resolution image.
Step 101 of acquiring a plurality of low-resolution images specifically includes:
performing oblique sampling on the image to obtain an oblique sampling image;
and acquiring a plurality of oblique sampling images.
constructing a high-resolution image grid;
calculating the pixel value of each grid point in the high-resolution image by adopting an interpolation algorithm according to the low-resolution image to obtain an interpolated high-resolution image, wherein the method specifically comprises the following steps:
selecting a pixel point in one low-resolution image near the grid point, and marking the pixel point as a first adjacent pixel point;
selecting three pixel points adjacent to the adjacent pixel points in the low pixel image where the adjacent pixel points are located, marking the three pixel points as second adjacent pixel points, and enabling the first adjacent pixel points and the second adjacent pixel points to form a square area;
obtaining regional pixel points in the low-resolution image except the low-resolution image where the first adjacent pixel point is located, wherein the regional pixel points are pixel points located in the square region;
acquiring pixel values of the first adjacent pixel point, the second adjacent pixel point and the regional pixel point;
and calculating the pixel values of the grid points by adopting a bilinear interpolation function according to the pixel values of the first adjacent pixel points, the second adjacent pixel points and the regional pixel points and the positions of the grid points.
establishing a denoising model of an L1 fidelity term with total variation regularization
Wherein X is a pixel value matrix of the high-resolution image to be solved, Y is a pixel value matrix of the interpolated high-resolution image,
α is a regularization parameter, # Xi,jIs a value for matrix ^ X row ith column jth, where ^ represents the gradient operator. The image edge blurring caused by the interpolation algorithm can be effectively removed, and the resolution of the image can be better improved.
Wherein, the L1 fidelity term is based on the L1 norm, and the most part of the consistency is kept between the interpolated image X and the image Y to be obtained. Embodied in a model
This term.
determining an augmented Lagrangian function of the denoising model;
obtaining an iterative formula of the augmented Lagrangian function;
and substituting the pixel value of the interpolated high-resolution image into the iterative formula of the augmented Lagrange function to obtain the pixel value of the denoised high-resolution image.
As a specific embodiment of the present invention, five given low resolution sequence target images (oblique sampling images) are first selected as experimental images, fig. 2 is five low resolution sequence target images in the embodiment of the present invention, as shown in fig. 2, with fig. 2(a) being a 400 × 400 reference diagram, fig. 2(b) being shifted by 0.4 pixels in the x direction and 0.2 pixels in the y direction relative to fig. 2 (a); FIG. 2(c) is shifted by 0.8 picture elements in the x-direction and 0.4 picture elements in the y-direction relative to FIG. 2 (a); the 2(d) picture is shifted by 1.2 picture elements in the x-direction and 0.6 picture elements in the y-direction relative to the picture of fig. 2 (a); FIG. 2(e) is shifted by 1.6 picture elements in the x-direction and 0.8 picture elements in the y-direction relative to FIG. 2 (a). Fig. 3 is an image obtained by an interpolation algorithm in the embodiment of the present invention, and as shown in fig. 3, five given low-resolution sequence target images are interpolated to obtain an interpolated image, an L1 fidelity denoising model with TV regularization is then established, an auxiliary variable is introduced to obtain an augmented lagrangian function of the model, and finally an iterative format of the lagrangian function is derived, so that a post-processing image is solved, as shown in fig. 4, and fig. 4 is an image obtained by an ADMM post-processing algorithm in the embodiment of the present invention. The method mainly comprises the steps of calculating an interpolation image, establishing a denoising model, and establishing and solving an augmented Lagrange function.
The specific process is as follows:
calculating an interpolation image:
given 5 low resolution images with a resolution of 400 x 400 at known relative positions, a super resolution image grid with a resolution of 800 x 800 is constructed. For each grid point in the high-resolution image, selecting a low-resolution image pixel point near the grid point and three adjacent pixel points of the image corresponding to the pixel in the low-pixel image, wherein the four points form a square area containing the high-resolution grid point. Due to the characteristic of the oblique sampling mode, in the square region, a pixel point in other low-resolution images is also included, so that 8 pixel points at known positions (pixel points corresponding to the vertexes of 4 square regions and a point in each of the other 4 images are in the square region) and grid points at known positions exist in the square region. The region is approximated by a bilinear function to calculate the pixel values of the grid points. Assuming that the points in the region satisfy a bilinear function of a + b x + c y + d x y, the values and relative positions of the known 8 pixels are respectively recorded as:
(x1，y1，z1)，(x2，y2，z2)，(x3，y3，z3)，(x4，y4，z4)，(x3，y5，z5)，(x6，y6，z6)，(x7，y7，z7)，(x8，y8，z8)
and the values and positions of the high resolution image grid points, noted as (x, y, z), where x, y are known. To find the parameters (a, b, c, d) and thus obtain the pixel value of the z point, it can be expressed in the form of a matrix
The above problem is equivalent to solving the overdetermined equation Ax ═ b; its least squares solution satisfies ATAx＝ATb; thereby obtaining an interpolated image.
Establishing a denoising model:
the basic form of the model is:
x is the high resolution image to be solved and Y is the image obtained by the interpolation algorithm. Where the TV mode of X is defined as follows:
the regularization parameter α is used to balance the data fidelity term and the regularization term. Here, the
As a fidelity term, alpha | | | X | | non-woven phosphorTVIs a regular term.
Is a matrix
Row i, column j, where ^ represents the gradient operator.
Establishing an augmented Lagrange function:
introducing auxiliary variables Z ═ X-Y and W ═ X, to obtain an augmented lagrange function for the model, as follows:
where ρ is1，ρ2Being arbitrary normal numbers, λ, μ is the lagrange multiplier corresponding to the equality constraint.
Solving the augmented Lagrange function:
deriving an iterative format of a Lagrangian function, thereby solving for X;
the iterative format of the method can be written as follows:
here the X-subproblem is a quadratic minimization problem, whose solution can be directly found as:
the W-sub problem and the Z-sub problem also have explicit expressions:
here, the function T is a soft threshold operator, whose expression is:
fig. 5 is an image obtained by partially amplifying and comparing a low-resolution target image, an interpolated image, and an ADMM post-processed image in an embodiment of the present invention, and as shown in fig. 5, fig. 5(a) is the low-resolution target image, fig. 5(b) is the image after interpolation processing, and fig. 5(c) is the image obtained by processing the low-resolution image by using the method provided by the present invention.
According to the method for acquiring the high-resolution image, the low-resolution image is subjected to interpolation processing, and the denoising model with the L1 fidelity term of total variation regularization is adopted to perform denoising processing on the interpolated high-resolution image, so that the defect of fuzzy boundary of the interpolated high-resolution image in the prior art is overcome, and the resolution of the image is effectively improved.
The present invention further provides a system for acquiring a high resolution image, fig. 6 is a schematic structural diagram of the system for acquiring a high resolution image according to the embodiment of the present invention, as shown in fig. 6, the system includes:
a low resolution image obtaining module 601, configured to obtain a plurality of low resolution images;
an interpolation processing module 602, configured to perform interpolation processing on multiple low resolutions to obtain an interpolated high resolution image;
the denoising model establishing module 603 is configured to establish a denoising model with a total variation regularized L1 fidelity term;
and a denoising module 604, configured to process the interpolated high-resolution image by using the denoising model to obtain a denoised high-resolution image.
The low-resolution image obtaining module 601 specifically includes:
the oblique sampling unit is used for performing oblique sampling on the image to obtain an oblique sampling image;
and the oblique sampling image acquisition unit is used for acquiring a plurality of oblique sampling images.
The interpolation processing module 602 specifically includes:
a high resolution image grid construction unit for constructing a high resolution image grid;
and the high-resolution image determining unit is used for calculating the pixel value of each grid point in the high-resolution image by adopting an interpolation algorithm according to the low-resolution image to obtain an interpolated high-resolution image.
The high-resolution image determination unit specifically includes:
a first adjacent pixel point selecting subunit, configured to select a pixel point in one of the low-resolution images near the grid point, and mark the pixel point as a first adjacent pixel point;
a second adjacent pixel point selecting subunit, configured to select, in the low-pixel image where the adjacent pixel point is located, three pixel points adjacent to the adjacent pixel point, and mark the three pixel points as second adjacent pixel points, where the first adjacent pixel point and the second adjacent pixel point form a square region;
a regional pixel point obtaining subunit, configured to obtain a regional pixel point in the low-resolution image except the low-resolution image where the first adjacent pixel point is located, where the regional pixel point is a pixel point located in the square region;
the pixel value acquisition subunit is used for acquiring the pixel values of the first adjacent pixel point, the second adjacent pixel point and the regional pixel point;
and the grid point pixel value operator unit is used for calculating the pixel values of the grid points by adopting a bilinear interpolation function according to the pixel values of the first adjacent pixel points, the second adjacent pixel points and the regional pixel points and the positions of the grid points.
The denoising model establishing module 603 specifically includes:
denoising model constructionA vertical unit for establishing a denoising model with a total variation regularized L1 fidelity term
Wherein X is a pixel value matrix of the high-resolution image to be solved, Y is a pixel value matrix of the interpolated high-resolution image,
α is a regularization parameter, # Xi,jIs a value for matrix ^ X row ith column jth, where ^ represents the gradient operator.
The denoising module 604 specifically includes:
the augmented Lagrange function determining unit is used for determining the augmented Lagrange function of the denoising model;
the iterative formula obtaining unit is used for obtaining an iterative formula of the augmented Lagrangian function;
and the high-resolution image pixel value unit is used for substituting the pixel value of the interpolated high-resolution image into the iterative formula of the augmented Lagrange function to obtain the pixel value of the denoised high-resolution image.
According to the system for acquiring the high-resolution image, the low-resolution image is subjected to interpolation processing, and the denoising model with the L1 fidelity term of total variation regularization is adopted to perform denoising processing on the interpolated high-resolution image, so that the defect of fuzzy boundary of the interpolated high-resolution image in the prior art is overcome, and the resolution of the image is effectively improved.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (8)
1. A method for obtaining a high resolution image, the method comprising:
acquiring a plurality of low-resolution images;
carrying out interpolation processing on the plurality of low resolutions to obtain an interpolated high resolution image;
establishing a denoising model of an L1 fidelity term with total variation regularization;
processing the interpolated high-resolution image by using the denoising model to obtain a denoised high-resolution image;
the interpolating the plurality of low-resolution images to obtain an interpolated high-resolution image specifically includes:
constructing a high-resolution image grid;
calculating the pixel value of each grid point in the high-resolution image by adopting an interpolation algorithm according to the low-resolution image to obtain an interpolated high-resolution image;
the calculating, according to the low-resolution image, a pixel value of each grid point in the high-resolution image by using an interpolation algorithm to obtain an interpolated high-resolution image specifically includes:
selecting a pixel point in one low-resolution image near a certain grid point in the high-resolution image, and marking the pixel point as a first adjacent pixel point;
selecting three pixel points adjacent to the adjacent pixel points in the low pixel image where the adjacent pixel points are located, marking the three pixel points as second adjacent pixel points, and enabling the first adjacent pixel points and the second adjacent pixel points to form a square area;
obtaining regional pixel points in the low-resolution image except the low-resolution image where the first adjacent pixel point is located, wherein the regional pixel points are pixel points located in the square region;
acquiring pixel values of the first adjacent pixel point, the second adjacent pixel point and the regional pixel point;
and calculating the pixel values of the grid points by adopting a bilinear interpolation function according to the pixel values of the first adjacent pixel points, the second adjacent pixel points and the regional pixel points and the positions of the grid points.
2. The method according to claim 1, wherein said acquiring a plurality of low resolution images comprises:
performing oblique sampling on the image to obtain an oblique sampling image;
and acquiring a plurality of oblique sampling images.
3. The method according to claim 1, wherein the establishing of the denoising model with the L1 fidelity term of total variation regularization specifically comprises:
establishing a denoising model of an L1 fidelity term with total variation regularization
Wherein X is a pixel value matrix of the high-resolution image to be solved, Y is a pixel value matrix of the interpolated high-resolution image,
alpha is a regularization parameter that is,
is a matrix
The ith row and the jth column, wherein
A gradient operator is represented.
4. The method according to claim 1, wherein the processing the interpolated high resolution image using the denoising model specifically includes:
determining an augmented Lagrangian function of the denoising model;
obtaining an iterative formula of the augmented Lagrangian function;
and substituting the pixel value of the interpolated high-resolution image into the iterative formula of the augmented Lagrange function to obtain the pixel value of the denoised high-resolution image.
5. A system for increasing the resolution of an image, the system comprising:
the low-resolution image acquisition module is used for acquiring a plurality of low-resolution images;
the interpolation processing module is used for carrying out interpolation processing on a plurality of low resolutions to obtain an interpolated high resolution image;
the denoising model establishing module is used for establishing a denoising model with a total variation regularized L1 fidelity term;
the denoising module is used for processing the interpolated high-resolution image by adopting the denoising model to obtain a denoised high-resolution image;
the interpolation processing module specifically includes:
a high resolution image grid construction unit for constructing a high resolution image grid;
the high-resolution image determining unit is used for calculating the pixel value of each grid point in the high-resolution image by adopting an interpolation algorithm according to the low-resolution image to obtain an interpolated high-resolution image;
the high-resolution image determination unit specifically includes:
a first adjacent pixel point selecting subunit, configured to select a pixel point in one of the low-resolution images near the grid point, and mark the pixel point as a first adjacent pixel point;
a second adjacent pixel point selecting subunit, configured to select, in the low-pixel image where the adjacent pixel point is located, three pixel points adjacent to the adjacent pixel point, and mark the three pixel points as second adjacent pixel points, where the first adjacent pixel point and the second adjacent pixel point form a square region;
a regional pixel point obtaining subunit, configured to obtain a regional pixel point in the low-resolution image except the low-resolution image where the first adjacent pixel point is located, where the regional pixel point is a pixel point located in the square region;
the pixel value acquisition subunit is used for acquiring the pixel values of the first adjacent pixel point, the second adjacent pixel point and the regional pixel point;
and the grid point pixel value operator unit is used for calculating the pixel values of the grid points by adopting a bilinear interpolation function according to the pixel values of the first adjacent pixel points, the second adjacent pixel points and the regional pixel points and the positions of the grid points.
6. The system according to claim 5, wherein the low resolution image acquisition module specifically includes:
the oblique sampling unit is used for performing oblique sampling on the image to obtain an oblique sampling image;
and the oblique sampling image acquisition unit is used for acquiring a plurality of oblique sampling images.
7. The system of claim 5, wherein the denoising model establishing module specifically comprises:
a denoising model establishing unit for establishing a denoising model with a total variation regularized L1 fidelity term
Wherein X is a pixel value matrix of the high-resolution image to be solved, and Y is the interpolated high-resolution imageA matrix of pixel values is formed by a matrix of pixel values,
alpha is a regularization parameter that is,
is a matrix
The ith row and the jth column, wherein
A gradient operator is represented.
8. The system of claim 5, wherein the denoising module specifically comprises:
the augmented Lagrange function determining unit is used for determining the augmented Lagrange function of the denoising model;
the iterative formula obtaining unit is used for obtaining an iterative formula of the augmented Lagrangian function;
and the high-resolution image pixel value unit is used for substituting the pixel value of the interpolated high-resolution image into the iterative formula of the augmented Lagrange function to obtain the pixel value of the denoised high-resolution image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710036911.3A CN106920213B (en) | 2017-01-19 | 2017-01-19 | Method and system for acquiring high-resolution image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710036911.3A CN106920213B (en) | 2017-01-19 | 2017-01-19 | Method and system for acquiring high-resolution image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106920213A CN106920213A (en) | 2017-07-04 |
| CN106920213B true CN106920213B (en) | 2021-02-12 |
Family
ID=59454440
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710036911.3A Active CN106920213B (en) | 2017-01-19 | 2017-01-19 | Method and system for acquiring high-resolution image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106920213B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109829874B (en) * | 2019-01-30 | 2023-06-30 | 西安电子科技大学 | SAR image fusion method based on frame theory |
| CN109889800B (en) * | 2019-02-28 | 2021-09-10 | 深圳市商汤科技有限公司 | Image enhancement method and device, electronic equipment and storage medium |
| CN110827224A (en) * | 2019-11-12 | 2020-02-21 | 南京工程学院 | High-resolution high-temperature forging shape detection method |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102222320B (en) * | 2011-05-24 | 2013-01-02 | 西安电子科技大学 | Single-frame image spatial resolution enhancing method based on total variation iteration back projection |
| US8923653B2 (en) * | 2011-06-27 | 2014-12-30 | Yutaka Yamamoto | Super-resolution processing method and system |
| CN103440619B (en) * | 2013-06-09 | 2018-04-17 | 清华大学 | A kind of tiltedly pattern sampling modeling and super resolution ratio reconstruction method |
| CN103473752A (en) * | 2013-08-22 | 2013-12-25 | 杨勇 | Super-resolution image reconstruction method based on coupled partial differential equation model |
| CN106127689B (en) * | 2016-06-30 | 2019-10-18 | 北京大学 | Image/video super-resolution method and device |
-
2017
- 2017-01-19 CN CN201710036911.3A patent/CN106920213B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN106920213A (en) | 2017-07-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107025632B (en) | Image super-resolution reconstruction method and system | |
| CN110827200B (en) | Image super-resolution reconstruction method, image super-resolution reconstruction device and mobile terminal | |
| CN102136144B (en) | Image registration reliability model and reconstruction method of super-resolution image | |
| CN101163224A (en) | Super-resolution device and method | |
| CN106169173B (en) | Image interpolation method | |
| US20110299795A1 (en) | Image processing system, image processing method, and image processing program | |
| JP5166156B2 (en) | Resolution conversion apparatus, method and program | |
| CN108961163A (en) | A kind of high-resolution satellite image super-resolution reconstruction method | |
| CN106920213B (en) | Method and system for acquiring high-resolution image | |
| Su et al. | Super-resolution without dense flow | |
| US20150071567A1 (en) | Image processing device, image processing method and non-transitory computer readable medium | |
| CN106169174B (en) | Image amplification method | |
| CN107292819A (en) | A kind of infrared image super resolution ratio reconstruction method protected based on edge details | |
| CN104063849A (en) | Video super-resolution reconstruction method based on image block self-adaptive registration | |
| CN104574338B (en) | Remote sensing image super-resolution reconstruction method based on multi-angle linear array CCD sensors | |
| CN115578255A (en) | Super-resolution reconstruction method based on inter-frame sub-pixel block matching | |
| Zhang et al. | Video super-resolution with 3D adaptive normalized convolution | |
| CN104376544B (en) | Non-local super-resolution reconstruction method based on multi-region dimension zooming compensation | |
| JP5566199B2 (en) | Image processing apparatus, control method therefor, and program | |
| KR101341617B1 (en) | Apparatus and method for super-resolution based on error model of single image | |
| JP2016164709A (en) | Image processing device, imaging apparatus, and image processing program | |
| CN107194874B (en) | Super-resolution imaging system and method based on bias image stabilization | |
| Al Ismaeil et al. | Dynamic super resolution of depth sequences with non-rigid motions | |
| Angelopoulou et al. | FPGA-based real-time super-resolution on an adaptive image sensor | |
| Mohan | Adaptive super-resolution image reconstruction with lorentzian error norm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |