CN113962908A - Pneumatic optical effect large-view-field degraded image point-by-point correction restoration method and system - Google Patents
Pneumatic optical effect large-view-field degraded image point-by-point correction restoration method and system Download PDFInfo
- Publication number
- CN113962908A CN113962908A CN202111569294.6A CN202111569294A CN113962908A CN 113962908 A CN113962908 A CN 113962908A CN 202111569294 A CN202111569294 A CN 202111569294A CN 113962908 A CN113962908 A CN 113962908A
- Authority
- CN
- China
- Prior art keywords
- point
- fuzzy
- gradient
- image
- space
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 49
- 230000003287 optical effect Effects 0.000 title claims abstract description 41
- 238000012937 correction Methods 0.000 title claims abstract description 25
- 238000004364 calculation method Methods 0.000 claims abstract description 31
- 230000015556 catabolic process Effects 0.000 claims abstract description 28
- 238000006731 degradation reaction Methods 0.000 claims abstract description 28
- 239000011159 matrix material Substances 0.000 claims abstract description 21
- 238000001914 filtration Methods 0.000 claims description 17
- 238000004422 calculation algorithm Methods 0.000 claims description 16
- 230000000694 effects Effects 0.000 claims description 12
- 230000000877 morphologic effect Effects 0.000 claims description 9
- 238000012216 screening Methods 0.000 claims description 6
- 238000000926 separation method Methods 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 4
- 230000000007 visual effect Effects 0.000 abstract description 5
- 230000006870 function Effects 0.000 description 10
- 230000008569 process Effects 0.000 description 8
- 239000013598 vector Substances 0.000 description 6
- 230000008859 change Effects 0.000 description 4
- 230000007423 decrease Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000009499 grossing Methods 0.000 description 4
- 238000003384 imaging method Methods 0.000 description 4
- 238000001514 detection method Methods 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 150000001875 compounds Chemical class 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000036961 partial effect Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000002829 reductive effect Effects 0.000 description 2
- 230000002411 adverse Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000007850 degeneration Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000003628 erosive effect Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000012634 optical imaging Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a point-by-point correction restoration method for a large visual field degraded image with a pneumatic optical effect, which comprises the following steps: calculating the gradient of an input degraded image, selecting a plurality of large gradient areas, and calculating a fuzzy core of each local area; calculating the distance from each point to the central point of the nearest two local areas, and performing inverse distance weighted interpolation calculation on the fuzzy kernels of each point of the whole image according to the two distances of each pixel point to obtain initial values of the fuzzy kernels of each point of the whole image and form an initial fuzzy kernel matrix; establishing a space-variant degradation model according to the initial fuzzy kernel matrix, and adding a non-negativity constraint regularization term and a sparsity constraint regularization term based on a self-adaptive anisotropic variable coefficient to enable the target image and each point fuzzy kernel to have nonnegativity and spatial adaptivity; and solving the space-variant degradation model to obtain the fuzzy kernel of each point and the gray value of each point to realize point-by-point correction, and finally outputting the space-variant degradation restored image. The method can correct and restore the aerodynamic optical effect space-variant degraded image with a large view field.
Description
Technical Field
The invention relates to the field of aerospace image processing and pneumatic optical effect correction, in particular to a point-by-point correction restoration method and system for a large-field degraded image with a pneumatic optical effect.
Background
In recent years, the aerospace detection technology of the high-speed aircraft in China is rapidly developed, and the high-speed aircraft carrying the optical imaging system is widely applied to tasks such as monitoring, investigation and detection. However, as the imaging environment of the high-speed aircraft is severe in the high-speed flight process, a complex high-speed flow field is formed between the optical head cover and the incoming flow, the pneumatic optical effect is generated, the target imaging quality is degraded, and the signal-to-noise ratio are greatly reduced. Modern techniques for restoration of degraded images by aerodynamic optical effects have not only focused on the ability to achieve restoration, but also have severe requirements on the quality of restoration. In the fields of star surface detection, aircraft orbit survey and the like, high-quality clear images are required to be obtained, and the clearer images are more beneficial to information analysis. Therefore, how to counteract or mitigate the adverse effect of the aerodynamic optical effect under high-speed flight conditions becomes an urgent problem to be solved.
The existing restoration method for the degraded image of the pneumatic optical effect mainly comprises the following steps: 1. flow field control methods can reduce the effects of aerodynamic optical effects, but far from eliminating such effects; 2. the adaptive optics method can complete partial restoration of wavefront phase difference, and the digital image restoration method can solve restoration of a space invariant aerodynamic optical effect degraded image, but cannot realize restoration of a large field of view space variant degraded image of the aerodynamic optical effect. In addition, in practice, imaging environment interference of the space image is numerous and unavoidable, actual imaging fuzzy kernels are generally space-variant, the fuzzy kernels solved by the existing algorithm always have deviation, image restoration quality is sometimes reduced, and time consumption is too long. It is therefore necessary to design a dedicated aero-optical effect degradation image restoration algorithm for these application requirements.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a point-by-point correction restoration method for a large field-of-view degraded image with a pneumatic optical effect aiming at the defects in the prior art.
The technical scheme adopted by the invention for solving the technical problems is as follows:
the method for restoring the degraded image with the large visual field and the pneumatic optical effect through point-by-point correction is provided, and comprises the following steps:
s1, calculating the gradient of the input degraded image, selecting a plurality of large gradient areas, extracting a plurality of local areas according to the distribution of the large gradient on the degraded image, and calculating a fuzzy core of each local area;
s2, calculating the distance between the central points of the two nearest local areas point by point, and performing inverse distance weighted interpolation calculation on the fuzzy kernels of all points of the whole image according to the two distances of each pixel point to obtain initial values of the fuzzy kernels of all points of the whole image, so as to form an initial fuzzy kernel matrix;
s3, establishing a space-variant degradation model according to the initial fuzzy kernel matrix, and adding a nonnegativity constraint regularization term determined by gray values of the target image and the initial fuzzy kernels of each point and a sparsity constraint regularization term based on a self-adaptive anisotropic coefficient determined by gradient values of the target image and the initial fuzzy kernels of each point to enable the target image and the fuzzy kernels of each point to have nonnegativity and spatial adaptivity;
and S4, solving the space-variant degradation model, obtaining the fuzzy core of each point and the gray value of each point to realize point-by-point correction, and finally outputting the space-variant degradation restoration image.
In the above technical solution, the specific method of step S1 is as follows:
s11, obtaining the gradient of the degraded image by using a multi-scale morphological gradient operator;
s12, filtering a small structural gradient region by using a gradient usefulness index, selecting a large gradient local region with the length and width directions larger than a certain value on the gradient image after filtering, and extracting a plurality of local regions according to the distribution of the large gradient on the degraded image;
and S13, estimating the fuzzy core of each extracted local area by using a non-negative least square criterion algorithm based on space correlation constraint.
In the above technical solution, the specific method of step S2 is as follows:
s21, regarding the fuzzy core of each local area as the initial value of the fuzzy core of each point of the local area;
s22, calculating Euclidean distances from each pixel point to the central points of all local areas;
and S23, comparing to obtain Euclidean distance from each pixel point to the central points of the two nearest local areas, regarding the distance as a corresponding weight coefficient, and performing inverse distance weighted interpolation calculation on the fuzzy core of the corresponding pixel point according to the weight coefficient to obtain the initial value of the fuzzy core of the corresponding pixel point.
In step S4, the fuzzy kernel and the gray value of each point are solved by using Bregman multivariate separation solving algorithm and the hysteresis fixed point iteration method.
The invention also provides a point-by-point restoration system for the large-view-field space-variant degraded image with the pneumatic optical effect, which comprises the following steps:
the local region screening module is used for calculating the gradient of the input degraded image, selecting a plurality of large gradient regions, extracting a plurality of local regions according to the distribution of the large gradient on the degraded image, and calculating a fuzzy core of each local region;
the fuzzy kernel initial value calculation module is used for calculating the distance between the central points of the two nearest local areas point by point, and performing inverse distance weighted interpolation calculation on the fuzzy kernels of all points of the whole image according to the two distances of each pixel point to obtain fuzzy kernel initial values of all points of the whole image so as to form an initial fuzzy kernel matrix;
the space-variant degradation model building module is used for building a space-variant degradation model according to the initial fuzzy kernel matrix, and adding a non-negativity constraint regularization term and a sparsity constraint regularization term based on a self-adaptive anisotropic variable coefficient to enable a target image and each point fuzzy kernel to have nonnegativity and space self-adaptability;
and the model solving module is used for solving the space-variant degradation model, obtaining the fuzzy kernel of each point and the gray value of each point to realize point-by-point correction, and finally outputting the space-variant degradation restored image.
According to the technical scheme, the local area screening module specifically comprises:
the gradient calculation submodule is used for solving the gradient of the degraded image by utilizing a multi-scale morphological gradient operator;
the gradient filtering submodule is used for filtering a small structural gradient region by utilizing a gradient usefulness index, selecting a large gradient local region with the length and width directions larger than a certain value on a gradient image after filtering, and then extracting a plurality of local regions according to the distribution of a large gradient on a degraded image;
and the regional fuzzy kernel estimation submodule is used for estimating the fuzzy kernel of each extracted local region by utilizing a non-negative least square criterion algorithm based on space correlation constraint.
According to the technical scheme, the fuzzy core initial value calculation module specifically comprises:
each point fuzzy kernel determining submodule is used for regarding the fuzzy kernel of each local area as the initial value of the fuzzy kernel of each point of the local area;
the distance calculation submodule is used for calculating Euclidean distances from each pixel point to the central points of all the local areas;
and the weighted fuzzy kernel calculation submodule is used for comparing to obtain the Euclidean distance from each pixel point to the central points of the two nearest local areas, regarding the distance as a corresponding weight coefficient, and performing inverse distance weighted interpolation calculation on the fuzzy kernels of the corresponding pixel points according to the weight coefficient to obtain the initial values of the fuzzy kernels of the corresponding pixel points.
According to the technical scheme, the model solving module specifically utilizes a Bregman multivariate separation solving algorithm and a hysteresis fixed point iteration method to solve the fuzzy core of each point and the gray value of each point.
The invention also provides a computer storage medium which can be executed by a processor and in which a computer program is stored, wherein the computer program executes the point-by-point correction and restoration method for the large field-of-view degraded image with the pneumatic optical effect.
The invention provides a point-by-point correction restoration method for a large field-of-view degraded image with a pneumatic optical effect, which comprises the following steps of:
the invention has the following beneficial effects: the point-by-point restoration method for the large visual field degraded image with the pneumatic optical effect has the advantages that under the condition of a large visual field, an initial fuzzy kernel matrix is formed through interpolation to meet the continuous change rule of fuzzy kernels of all points, a non-negative sparse regularization term based on a self-adaptive anisotropic variable coefficient is further added to a space-variant degraded model, the obvious effects in the aspects of noise suppression and edge characteristic retention are achieved, the fuzzy kernels and the gray values of all points are iteratively solved to obtain a clear image, and the restoration effect is more accurate compared with that of a space-invariant restoration method.
Drawings
The invention will be further described with reference to the accompanying drawings and examples, in which:
FIG. 1 is a flow chart of a point-by-point correction restoration method for a large field-of-view degraded image with aerodynamic optical effect according to an embodiment of the present invention;
FIG. 2 is a schematic diagram illustrating a relationship between image coordinates and a blur kernel position according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a regularization scheme according to an embodiment of the present invention;
FIG. 4 shows an embodiment of the present invention
A function curve image;
FIG. 5 shows an embodiment of the present invention
A function curve image;
FIG. 6 is an original aero-optical effect degraded image of an embodiment of the present invention;
FIG. 7 is a large gradient profile image of an embodiment of the present invention;
FIG. 8 is a block diagram of a local area blur kernel image according to an embodiment of the present invention;
FIG. 9 is a diagram illustrating interpolation estimation of a space-variant blur kernel image for a representative point (1, 1) according to an embodiment of the present invention;
FIG. 10 is a comparison graph of the result of the restoration of the degraded image by the aerodynamic optical effect and the result of the restoration by the aerodynamic optical effect according to the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The point-by-point correction and restoration method for the large-field-of-view degraded image with the pneumatic optical effect can be realized by adopting a C + + program of VC6.0 and a Matlab platform, the running environment is Windows10, and the processor is Intel Core i 7.
As shown in fig. 1, the point-by-point correction restoration method for a large field-of-view degraded image with aerodynamic optical effect according to an embodiment of the present invention includes the following steps:
s1, inputting the degraded image, calculating the gradient of the degraded image, selecting a plurality of large gradient areas, extracting a plurality of local areas according to the distribution of the large gradient on the degraded image, and calculating a fuzzy core of each local area;
s2, calculating the distance between the central points of the two nearest local areas point by point, and performing inverse distance weighted interpolation calculation on the fuzzy kernels of all points of the whole image according to the two distances of each pixel point to obtain initial values of the fuzzy kernels of all points of the whole image, so as to form an initial fuzzy kernel matrix;
s3, establishing a space-variant degradation model according to the initial fuzzy kernel matrix, and adding a nonnegativity constraint regularization term determined by gray values of the target image and the initial fuzzy kernels of each point and a sparsity constraint regularization term based on a self-adaptive anisotropic coefficient determined by gradient values of the target image and the initial fuzzy kernels of each point to enable the target image and the fuzzy kernels of each point to have nonnegativity and spatial adaptivity;
and S4, solving the space-variant degradation model, obtaining the fuzzy core of each point and the gray value of each point to realize point-by-point correction, and finally outputting the space-variant degradation restoration image.
Further, the specific method of step S1 is:
s11, obtaining the gradient of the degraded image by using a multi-scale morphological gradient operator;
s12, filtering a small structural gradient region by using a gradient usefulness index, selecting a large gradient local region with the length and width directions larger than a certain value on the gradient image after filtering, and extracting a plurality of local regions according to the distribution of the large gradient on the degraded image;
and S13, estimating the fuzzy core of each extracted local area by using a non-negative least square criterion algorithm based on space correlation constraint.
The specific method of step S2 is:
s21, regarding the fuzzy core of each local area as the initial value of the fuzzy core of each point of the local area;
s22, calculating Euclidean distances from each pixel point to the central points of all local areas;
and S23, comparing to obtain Euclidean distance from each pixel point to the central points of the two nearest local areas, regarding the distance as a corresponding weight coefficient, and performing inverse distance weighted interpolation calculation on the fuzzy core of the corresponding pixel point according to the weight coefficient to obtain the initial value of the fuzzy core of the corresponding pixel point.
In a preferred embodiment of the present invention, in step S1, the degraded image is recorded as
As shown in fig. 6, the size of the image is
. Firstly, solving the gradient of a degraded image by using a multi-scale morphological gradient operator:
in the formula
In order to be a gradient of the magnetic field,
and
respectively representing the expansion and erosion operations,
for the coordinates of the points of the image,
，
the value is 3 for the scale,
is the radius of the morphological structural element,
is the side length of the morphological structural element.
Using gradient usefulness indicators
To filter out small structural gradient regions:
in the formula
Is composed of
The norm of the number of the first-order-of-arrival,
to be a pixel point
The length and width of the center are
The rectangular neighborhood of the light source is determined,
。
is an index of a pixel within the neighborhood,
；
can eliminate
Medium gradient sawtooth small peak, i.e. small structural gradients in the image gradient.
Is that
Sum of absolute values of medium gradient. Computing
Setting a threshold value
，
Value less than
Indication point
The neighborhood of (2) is a small structural gradient region, and filtering is carried out. Constants in examples of the inventioncValue of 0.1, threshold
Set to 0.5.
After filtering out the small structural gradient regions, large gradient regions with length and width directions larger than 10 are selected on the gradient image, as shown in fig. 7. If the region is not selected, the size of the region selection can be changed appropriately, or the image is already a clear image, and no recovery is needed. Finally, three local regions are extracted according to the distribution of the large gradient on the degraded image, and a fuzzy kernel is used
Is sized as
(usually set in the range of 3 to 27 and an odd number), wherein
And carrying out fuzzy kernel on the extracted three local regions by using a non-negative least square rule algorithm based on space correlation constraint on the basis of the prior art
：
Solving by using iterative minimization algorithm
Let the objective function derivative be zero-solved to obtain:
wherein
Is composed of
The norm of the number of the first-order-of-arrival,
the fuzzy point corresponds to an area in the original image,
a one-dimensional column vector formed by pixels from the degraded image,
is a blur kernel in the form of a one-dimensional column vector.
Represents the second in the blur kernel
Component and adjacent second
The specific values of the components, the three local area blur kernels solved and the 3D display are shown in fig. 8.
In step S2, the blur kernel of each local region can be regarded as the initial value of the blur kernel at each point of the local region. Because only the small-field-of-view image can be regarded as space invariant, and the problem of large-field-of-view space variant deblurring is solved, the invention estimates the initial value of each point blur kernel by interpolation according to the rule that the change of the aerodynamic optical effect space variant blur kernel in the large field of view has continuity. By point
For example, comparative points
To the first
Block fuzzy kernel estimation of region center point
Stored set of Euclidean distances
In which
。
Comparison
The size of each element in the table is obtained
The Euclidean distances to the central points of the two closest fuzzy core estimation areas are respectively recorded as
As shown in fig. 2 (a). The distance is taken as a weight coefficient, and the inverse distance weighted interpolation operation is carried out on the point fuzzy core. The calculation formula is as follows:
in the formula
Is to a point
A distance of
Local area center point of
The blur kernel of (a) is determined,
is to a point
A distance of
Local area center point of
The blur kernel of (a) is determined,
is a point
And (c) processing the blur kernel formed by interpolation, as shown in (b) of fig. 2. And analogizing in turn to obtain initial values of the fuzzy cores of all points of the whole graph. Representative point
The interpolation blur kernel and the 3D display are shown in fig. 9.
In step S3, a space-variant degradation model is established, and a sparsity constraint regularization term and a nonnegativity constraint regularization term based on the adaptive anisotropic regularization variable coefficient are added to constrain blur kernels at each point of the target image and the degraded image. The model construction process is as follows:
the degraded image forming process can be modeled in the following form:
wherein
Is a degraded image in which the image is degraded,
is a kernel of the blur and is,
is a clear image and is a color image,
is noise.
Under the condition of large field of view, the invention attaches importance to the fact that each point of the degraded image is fuzzy and different, and the invention will
Each point
Using column vectors
Piled up, the space-variant degeneration process can be expressed in a matrix-vector form as:
wherein
Are respectively as
，
，
Is used to accumulate the column vectors of the column,
in order to blur the kernel matrix, the kernel matrix is,
and sequentially forming the fuzzy kernel of each pixel point from left to right and from top to bottom. When the image is a space that is not blurred,
from left to right, from top to bottom, correspond to the same fuzzy kernel, i.e. for each point
Is stationary. When the image is a space-variant blur,
from left to right, from top to bottom, corresponding to the fuzzy kernels of different pixel points, i.e. of each point
Are variable.
Due to fuzzy kernel matrix of each pixel point
A certain calculation error is generated in the interpolation forming process, and the accuracy needs to be further improved. To solve the error problem, let
For a known blur kernel matrix containing errors, there are:
wherein
From left to right, sequentially from top to bottom for each pixel point
Blur kernel formed by interpolation
，
For error, the space-variant degradation model is established as follows:
formula of further direction
Reasonable regularization terms are added to enable the solution of the model to approach the real solution, and the smoothness degree of each pixel point in 4 directions is mostly determined by adopting fixed regularization parameters in the traditional space-variant degradation model
The same is true, and no good effect is achieved. Since the gradient has directionality, the smoothing should be different when the gradient magnitude is different for different directions of the point. Therefore, the size of the regularization parameter is related to the gradient value of each point, and the invention aims at the correction problem of the degraded image with large visual field according to the target image
And fuzzy core
The respective characteristics provide different anisotropic regularization parameters to perform different adjustments on 4 directions of each point respectively so as to achieve different smoothness on the different directions of each point, as shown in fig. 3.
Actually, a large view field image with an observed aerodynamic optical effect has many different edge features, and for the small gradient value and the relatively close gray value of pixels in a target and background region in a target image, a large degree of regularization is needed to smooth the partial region so as to suppress noise influence. Meanwhile, in order to notice the difference of the gray values of pixels at the boundary between the target and the background, the regularization of the region is weakened to keep the gradient difference, the regularization parameter is in a function form which monotonically decreases and sharply decreases in mathematical analysis, and in order to simplify the calculation, the regularization function in the following form is selected by the method
The regularization term coefficient is constrained as the sparsity of a target image so as to achieve the purpose of protecting edge characteristics while suppressing noise.
In the formula (I), the compound is shown in the specification,
is a target image
The gradient value of each point is initially obtained in step S1
The gradient value of each point.C 1The regularization constant coefficient of the target image is generally set to 1 when
When the value is too large, it is adjusted to be largerC 1The value of (c) increases the degree of smoothing and conversely decreases. When in useC 1When the ratio is not less than 1,
is in the range of 0 to 1, and the function curve is shown in FIG. 4, and is exponentialnThe value is between 0 and 5,nthe value is determined by the decay, the faster the decay,nthe value is larger, otherwise smaller. When in use
When the temperature of the water is higher than the set temperature,
being non-directional, all directions being the same, when
It is obvious that the gradient value can be changed
The adjustment is carried out, so that the target image has certain space self-adaptive capacity.
Aiming at the characteristics that a fuzzy kernel image has a Gaussian-like shape and is not steep, the gradient of each point rises and falls continuously, the gradient change is mainly reflected on the integral attenuation, and the transition between adjacent points is slow. The prior knowledge is integrated into the model, the regularization parameter should be a small value in a large gradient area to protect the fuzzy kernel peak value, and the regularization parameter should be a little larger value in a small gradient area to suppress noise. For convenient algorithm implementation, firstly, the gradient of each point interpolation forming fuzzy core is calculated
：
Selecting a regularization function with relatively gradual changes
Regularization term coefficients are constrained as a blur kernel sparsity to protect blur kernel peaks.
In the formula (I), the compound is shown in the specification,
the initial value being the gradient of the fuzzy kernel
The gradient value of each point in the middle is calculated,C 2the regularization constant for the blur kernel is generally set to 1 when
When the value is too large, it is adjusted to be largerC 2The value of (c) increases the degree of smoothing and conversely decreases. When in useC 2When the ratio is not less than 1,
the symmetric monotone decreasing function curve is shown in FIG. 5. Parameter(s)
For smoothing the control coefficients, in general
1 is taken.
When the attenuation is not 1, the attenuation is determined, and the faster the attenuation is,
the value of (a) is between 0 and 1, so that a large gradient is protected, otherwise, the value of (b) is taken to be large, and obviously, the gradient value can be selected
The method is adjusted to make the fuzzy core have space self-adaptability.
In addition, aiming at the characteristic that the target image and the fuzzy core have nonnegativity, when a negative value appears in the solution vector, punishment needs to be carried out on the negative value, and the solution is forced to develop towards the nonnegativity direction. In order to reduce the calculation amount, the invention selects the cost function with the following form according to the gray value of each point of the target image and the fuzzy core image
And as a target, fuzzy kernel nonnegative constraint regularization item.
In the formula
Is to
A constant coefficient for penalizing a medium negative value,
the value range of (a) is between 0 and 10, when
When the medium negative value is too large, the negative value,
the value is small, otherwise, the value is large, so that the change of each point of the target image and the fuzzy core image is smooth. In the present example, the value is set to 1,
is prepared by reacting with
The diagonal matrix related to each element in (1) is as follows:
wherein
Diagonal element
Values of 1 and 0. Diagonal element
Are respectively composed of
In response to
Value of individual element
Determine whether there is
Let variable quantity
Is related to the blur kernel, based on the above description, if noise is not taken into account, in the formula
The minimization model of the space-variant deblurring is established on the basis as follows:
the first term in the equation is the basic data term,
is composed of
The orthogonal Haar wavelet transform matrix plays a role in protecting image details and texture information, and the second term and the third term are sparsity constraint regularization terms of a target and a fuzzy kernel based on a self-adaptive anisotropic variable coefficient.
And (4) a target fuzzy kernel non-negativity constraint regularization item.
In step S3, the solving process of the constructed minimization model according to the present invention is as follows:
due to the formula
Therein contain
Norm, therefore the invention adopts Bregman multivariable separation solving method to solveFormula (II)
Let a variable of
And introducing Bregman auxiliary variable
，
To update the iterative process, i.e. the formula
The transformation can be as follows:
order to
，
Is as followsnStep (2) iteratively solving the obtained resultnThe +1 step iterative solution process is as follows:
the update of the Bregman auxiliary variable is:
until an iteration end condition is reached in the iteration process:
Formula (II)
In relation to
After the variable minimization is zero, FFT is used for the frequency domain to the above formula
And (3) carrying out quick solution, namely:
in the formula
Which represents the fourier transform of the signal,
which represents the inverse of the fourier transform,
a fourier complex conjugate operator is expressed to thereby realize a fast point-by-point correction, and a space-variant restored image is obtained, as shown in fig. 10 (b). As can be seen from fig. 10, the null-deformation restoration effect is better than the null-constant restoration effect.
The point-by-point restoration system for the large-view-field space-variant degraded image with the pneumatic optical effect is mainly used for realizing the method embodiment, and comprises the following steps:
the local region screening module is used for calculating the gradient of the input degraded image, selecting a plurality of large gradient regions, extracting a plurality of local regions according to the distribution of the large gradient on the degraded image, and calculating a fuzzy core of each local region;
the fuzzy kernel initial value calculation module is used for calculating the distance between the central points of the two nearest local areas point by point, and performing inverse distance weighted interpolation calculation on the fuzzy kernels of all points of the whole image according to the two distances of each pixel point to obtain fuzzy kernel initial values of all points of the whole image so as to form an initial fuzzy kernel matrix;
the space-variant degradation model building module is used for building a space-variant degradation model according to the initial fuzzy kernel matrix, and adding a non-negativity constraint regularization term and a sparsity constraint regularization term based on a self-adaptive anisotropic variable coefficient to enable a target image and each point fuzzy kernel to have nonnegativity and space self-adaptability;
and the model solving module is used for solving the space-variant degradation model, obtaining the fuzzy kernel of each point and the gray value of each point to realize point-by-point correction, and finally outputting the space-variant degradation restored image.
Wherein, local region screening module specifically includes:
the gradient calculation submodule is used for solving the gradient of the degraded image by utilizing a multi-scale morphological gradient operator;
the gradient filtering submodule is used for filtering a small structural gradient region by utilizing a gradient usefulness index, selecting a large gradient local region with the length and width directions larger than a certain value on a gradient image after filtering, and then extracting a plurality of local regions according to the distribution of a large gradient on a degraded image;
and the regional fuzzy kernel estimation submodule is used for estimating the fuzzy kernel of each extracted local region by utilizing a non-negative least square criterion algorithm based on space correlation constraint.
Further, the fuzzy core initial value calculation module specifically includes:
each point fuzzy kernel determining submodule is used for regarding the fuzzy kernel of each local area as the initial value of the fuzzy kernel of each point of the local area;
the distance calculation submodule is used for calculating Euclidean distances from each pixel point to the central points of all the local areas;
and the weighted fuzzy kernel calculation submodule is used for comparing to obtain the Euclidean distance from each pixel point to the central points of the two nearest local areas, regarding the distance as a corresponding weight coefficient, and performing inverse distance weighted interpolation calculation on the fuzzy kernels of the corresponding pixel points according to the weight coefficient to obtain the initial values of the fuzzy kernels of the corresponding pixel points.
Further, the model solving module specifically uses a Bregman multivariate separation solving algorithm to solve the fuzzy core of each point and the gray value of each point.
The implementation of each module is referred to the above method embodiment, and is not described herein again.
The present application also provides a computer-readable storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory (e.g., SD or DX memory, etc.), a Random Access Memory (RAM), a Static Random Access Memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, a server, an App application mall, etc., on which a computer program is stored, which when executed by a processor implements corresponding functions. The computer readable storage medium of this embodiment is used for implementing the point-by-point correction restoration method of the large field-of-view degraded image with aerodynamic optical effect of this embodiment when being executed by the processor.
The point-by-point restoration method for the large-view-field degraded image with the pneumatic optical effect can provide a processing method for fuzzy kernel estimation, fuzzy kernel optimization and restoration requirements of each point of the space-variant degraded image with the pneumatic optical effect; the method can restore the large-view-field space-variant degraded image with the pneumatic optical effect, and can meet the requirement of restoring the large-view-field space-variant degraded image with the pneumatic optical effect in the aerospace field.
It will be understood that modifications and variations can be made by persons skilled in the art in light of the above teachings and all such modifications and variations are intended to be included within the scope of the invention as defined in the appended claims.
Claims (9)
1. A point-by-point correction restoration method for a large field-of-view degraded image with a pneumatic optical effect is characterized by comprising the following steps:
s1, calculating the gradient of the input degraded image, selecting a plurality of large gradient areas, extracting a plurality of local areas according to the distribution of the large gradient on the degraded image, and calculating a fuzzy core of each local area;
s2, calculating the distance between the central points of the two nearest local areas point by point, and performing inverse distance weighted interpolation calculation on the fuzzy kernels of all points of the whole image according to the two distances of each pixel point to obtain initial values of the fuzzy kernels of all points of the whole image, so as to form an initial fuzzy kernel matrix;
s3, establishing a space-variant degradation model according to the initial fuzzy kernel matrix, and adding a nonnegativity constraint regularization term determined by gray values of the target image and the initial fuzzy kernels of each point and a sparsity constraint regularization term based on a self-adaptive anisotropic coefficient determined by gradient values of the target image and the initial fuzzy kernels of each point to enable the target image and the fuzzy kernels of each point to have nonnegativity and spatial adaptivity;
and S4, solving the space-variant degradation model, obtaining the fuzzy core of each point and the gray value of each point to realize point-by-point correction, and finally outputting the space-variant degradation restoration image.
2. The method for restoring a degraded image with a large field of view and an aerodynamic optical effect according to claim 1, wherein the step S1 is specifically performed by:
s11, obtaining the gradient of the degraded image by using a multi-scale morphological gradient operator;
s12, filtering a small structural gradient region by using a gradient usefulness index, selecting a large gradient local region with the length and width directions larger than a certain value on the gradient image after filtering, and extracting a plurality of local regions according to the distribution of the large gradient on the degraded image;
and S13, estimating the fuzzy core of each extracted local area by using a non-negative least square criterion algorithm based on space correlation constraint.
3. The method for restoring a degraded image with a large field of view and an aerodynamic optical effect according to claim 1, wherein the step S2 is specifically performed by:
s21, regarding the fuzzy core of each local area as the initial value of the fuzzy core of each point of the local area;
s22, calculating Euclidean distances from each pixel point to the central points of all local areas;
and S23, comparing to obtain Euclidean distance from each pixel point to the central points of the two nearest local areas, regarding the distance as a corresponding weight coefficient, and performing inverse distance weighted interpolation calculation on the fuzzy core of the corresponding pixel point according to the weight coefficient to obtain the initial value of the fuzzy core of the corresponding pixel point.
4. The method for point-by-point correction and restoration of the aero-optical effect large-field-of-view degraded image according to claim 1, wherein in step S4, the blur kernel of each point and the gray value of each point are solved by using a Bregman multivariate separation solving algorithm and a lag fixed point iteration method.
5. A point-by-point restoration system for large-field-of-view space-variant degraded images with pneumatic optical effect is characterized by comprising the following components:
the local region screening module is used for calculating the gradient of the input degraded image, selecting a plurality of large gradient regions, extracting a plurality of local regions according to the distribution of the large gradient on the degraded image, and calculating a fuzzy core of each local region;
the fuzzy kernel initial value calculation module is used for calculating the distance between the central points of the two nearest local areas point by point, and performing inverse distance weighted interpolation calculation on the fuzzy kernels of all points of the whole image according to the two distances of each pixel point to obtain fuzzy kernel initial values of all points of the whole image so as to form an initial fuzzy kernel matrix;
the space-variant degradation model building module is used for building a space-variant degradation model according to the initial fuzzy kernel matrix, and adding a non-negativity constraint regularization term and a sparsity constraint regularization term based on a self-adaptive anisotropic variable coefficient to enable a target image and each point fuzzy kernel to have nonnegativity and space self-adaptability;
and the model solving module is used for solving the space-variant degradation model, obtaining the fuzzy kernel of each point and the gray value of each point to realize point-by-point correction, and finally outputting the space-variant degradation restored image.
6. The system for restoring a large-field-of-view spatially-varying degraded image with a dynamic optical effect point-by-point method according to claim 5, wherein the local region screening module specifically comprises:
the gradient calculation submodule is used for solving the gradient of the degraded image by utilizing a multi-scale morphological gradient operator;
the gradient filtering submodule is used for filtering a small structural gradient region by utilizing a gradient usefulness index, selecting a large gradient local region with the length and width directions larger than a certain value on a gradient image after filtering, and then extracting a plurality of local regions according to the distribution of a large gradient on a degraded image;
and the regional fuzzy kernel estimation submodule is used for estimating the fuzzy kernel of each extracted local region by utilizing a non-negative least square criterion algorithm based on space correlation constraint.
7. The system for restoring a large-field-of-view spatially-varying degraded image with aerodynamic optical effects point-by-point as claimed in claim 5, wherein the blur kernel initial value calculating module specifically comprises:
each point fuzzy kernel determining submodule is used for regarding the fuzzy kernel of each local area as the initial value of the fuzzy kernel of each point of the local area;
the distance calculation submodule is used for calculating Euclidean distances from each pixel point to the central points of all the local areas;
and the weighted fuzzy kernel calculation submodule is used for comparing to obtain the Euclidean distance from each pixel point to the central points of the two nearest local areas, regarding the distance as a corresponding weight coefficient, and performing inverse distance weighted interpolation calculation on the fuzzy kernels of the corresponding pixel points according to the weight coefficient to obtain the initial values of the fuzzy kernels of the corresponding pixel points.
8. The system for restoring a large-field-of-view space-variant degraded image point-by-point through an aerodynamic optical effect according to claim 5, wherein the model solving module solves the blur kernel of each point and the gray value of each point by using a Bregman multivariate separation solving algorithm and a hysteresis fixed-point iteration method.
9. A computer storage medium executable by a processor and having stored therein a computer program for performing the method for point-by-point correction restoration of a degraded image with a large field of view of aero-optical effects as set forth in any one of claims 1 to 4.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111569294.6A CN113962908B (en) | 2021-12-21 | 2021-12-21 | Pneumatic optical effect large-visual-field degraded image point-by-point correction restoration method and system |
| PCT/CN2022/093807 WO2023115801A1 (en) | 2021-12-21 | 2022-05-19 | Point-by-point correction and restoration method and system for large field-of-view degraded image having aero-optical effect |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111569294.6A CN113962908B (en) | 2021-12-21 | 2021-12-21 | Pneumatic optical effect large-visual-field degraded image point-by-point correction restoration method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113962908A true CN113962908A (en) | 2022-01-21 |
| CN113962908B CN113962908B (en) | 2022-05-10 |
Family
ID=79473314
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111569294.6A Active CN113962908B (en) | 2021-12-21 | 2021-12-21 | Pneumatic optical effect large-visual-field degraded image point-by-point correction restoration method and system |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN113962908B (en) |
| WO (1) | WO2023115801A1 (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023115801A1 (en) * | 2021-12-21 | 2023-06-29 | 武汉工程大学 | Point-by-point correction and restoration method and system for large field-of-view degraded image having aero-optical effect |
| CN117173058A (en) * | 2023-11-03 | 2023-12-05 | 武汉工程大学 | Unified restoration method and system for space-variant blurred image |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117557626B (en) * | 2024-01-12 | 2024-04-05 | 泰安大陆医疗器械有限公司 | Auxiliary positioning method for spray head installation of aerosol sprayer |
| CN117726307B (en) * | 2024-02-18 | 2024-04-30 | 成都汇智捷成科技有限公司 | Data management method based on business center |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112365420A (en) * | 2020-11-12 | 2021-02-12 | 重庆邮电大学 | Blurred image blind restoration method based on non-local window gradient |
| CN113538374A (en) * | 2021-07-15 | 2021-10-22 | 中国科学院上海技术物理研究所 | Infrared image blur correction method for high-speed moving object |
| CN113793284A (en) * | 2021-11-17 | 2021-12-14 | 武汉工程大学 | Image restoration method for nonuniform blocking of aerodynamic optical effect space-variant blurred image |
| CN113793285A (en) * | 2021-11-17 | 2021-12-14 | 武汉工程大学 | Ultrafast restoration method and system for pneumatic optical effect target twin image |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9800866B2 (en) * | 2015-04-08 | 2017-10-24 | Algolux Inc. | Method for providing an estimation of a point spread function indicative of intrinsic camera blur |
| CN109919871A (en) * | 2019-03-05 | 2019-06-21 | 重庆大学 | Fuzzy core estimation method based on image and fuzzy core mixed constraints |
| CN113962908B (en) * | 2021-12-21 | 2022-05-10 | 武汉工程大学 | Pneumatic optical effect large-visual-field degraded image point-by-point correction restoration method and system |
-
2021
- 2021-12-21 CN CN202111569294.6A patent/CN113962908B/en active Active
-
2022
- 2022-05-19 WO PCT/CN2022/093807 patent/WO2023115801A1/en unknown
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112365420A (en) * | 2020-11-12 | 2021-02-12 | 重庆邮电大学 | Blurred image blind restoration method based on non-local window gradient |
| CN113538374A (en) * | 2021-07-15 | 2021-10-22 | 中国科学院上海技术物理研究所 | Infrared image blur correction method for high-speed moving object |
| CN113793284A (en) * | 2021-11-17 | 2021-12-14 | 武汉工程大学 | Image restoration method for nonuniform blocking of aerodynamic optical effect space-variant blurred image |
| CN113793285A (en) * | 2021-11-17 | 2021-12-14 | 武汉工程大学 | Ultrafast restoration method and system for pneumatic optical effect target twin image |
Non-Patent Citations (3)
| Title |
|---|
| HONG HANYU.ET.L: "A fast method for single image haze removal based on multiscale dark channel prior", 《2017: PARALLEL PROCESSING OF IMAGES AND OPTIMIZATION TECHNIQUES》 * |
| LIANG XU.ET.L: "High altitude aero-optic imaging deviation prediction for a hypersonic flying vehicle", 《2011 IEEE INTERNATIONAL CONFERENCE ON IMAGING SYSTEMS AND TECHNIQUES》 * |
| 洪汉玉等: "气动光学效应红外序列退化图像优化复原算法", 《红外与激光工程》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023115801A1 (en) * | 2021-12-21 | 2023-06-29 | 武汉工程大学 | Point-by-point correction and restoration method and system for large field-of-view degraded image having aero-optical effect |
| CN117173058A (en) * | 2023-11-03 | 2023-12-05 | 武汉工程大学 | Unified restoration method and system for space-variant blurred image |
| CN117173058B (en) * | 2023-11-03 | 2024-02-02 | 武汉工程大学 | Unified restoration method and system for space-variant blurred image |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2023115801A1 (en) | 2023-06-29 |
| CN113962908B (en) | 2022-05-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113962908B (en) | Pneumatic optical effect large-visual-field degraded image point-by-point correction restoration method and system | |
| Hyun Kim et al. | Dynamic scene deblurring | |
| CN106920220B (en) | The turbulent flow method for blindly restoring image optimized based on dark primary and alternating direction multipliers method | |
| US9692939B2 (en) | Device, system, and method of blind deblurring and blind super-resolution utilizing internal patch recurrence | |
| CN110675347B (en) | Image blind restoration method based on group sparse representation | |
| CN102326379B (en) | Method for removing blur from image | |
| US9589328B2 (en) | Globally dominant point spread function estimation | |
| CN112215773B (en) | Local motion deblurring method and device based on visual saliency and storage medium | |
| JP5890547B2 (en) | Image processing device | |
| Rabin et al. | Removing artefacts from color and contrast modifications | |
| WO2017100971A1 (en) | Deblurring method and device for out-of-focus blurred image | |
| CN110796616B (en) | Turbulence degradation image recovery method based on norm constraint and self-adaptive weighted gradient | |
| CN110874827A (en) | Turbulent image restoration method and device, terminal equipment and computer readable medium | |
| CN105574823B (en) | A kind of deblurring method and device of blurred picture out of focus | |
| Anantrasirichai et al. | Mitigating the effects of atmospheric distortion using DT-CWT fusion | |
| Hu et al. | Error tolerance and effects analysis of satellite vibration characteristics and measurement error on TDICCD image restoration | |
| CN112862729B (en) | Remote sensing image denoising method based on characteristic curve guidance | |
| Junjie et al. | An image defocus deblurring method based on gradient difference of boundary neighborhood | |
| Yu et al. | Research Status and Development Trend of Image Restoration Technology | |
| CN113822823B (en) | Point neighbor restoration method and system for aerodynamic optical effect image space-variant fuzzy core | |
| El Abbadi et al. | A Survey on Blind De-Blurring of Digital Image. | |
| James et al. | Fourier Based Pre-Processing For Seeing Through Water | |
| CN115082333B (en) | Image deblurring method based on normalized weighted total variation method and computer program product | |
| CN117173058B (en) | Unified restoration method and system for space-variant blurred image | |
| CN116703958B (en) | Edge contour detection method, system, equipment and storage medium for microscopic image |
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 |