Disclosure of Invention
Aiming at the technical problems in the prior art, the invention provides an image de-jitter method and device in the LED screen correction process.
An image de-jitter method in the process of LED screen correction comprises the following steps:
acquiring a picture of the LED screen, and selecting a part of the picture of the LED screen as a fuzzy core estimation sample;
carrying out frequency domain conversion on the fuzzy kernel estimation sample, and acquiring the distribution characteristics of high-frequency components;
quantizing the distribution characteristics, acquiring a fuzzy angle and determining a fuzzy direction;
performing iterative estimation according to the fuzzy angle and the fuzzy direction to obtain a fuzzy length;
constructing a fuzzy kernel according to the fuzzy angle and the fuzzy length;
and deblurring the picture of the LED screen by using a fuzzy core.
Further, the blur kernel estimates the samples as 1/16 through 1/2 of the picture of the LED screen.
Further, performing frequency domain conversion on the fuzzy core estimation sample, and acquiring the distribution characteristics of the high-frequency components, including:
converting the fuzzy kernel estimation sample from a space domain into a frequency domain through Fourier transform to obtain a frequency domain image;
and carrying out binarization processing on the frequency domain image to obtain the distribution characteristics of the high-frequency components.
Further, quantizing the distribution characteristics to obtain a fuzzy angle, including:
in the distribution characteristics, quantification is performed by the central moment;
estimating a blur angle from the quantization result, the blur angle being θ, an
Wherein the content of the first and second substances,
x
cand y
cIs the centroid coordinate of the distribution feature, and M is the component of the central moment.
Further, performing iterative estimation in a blur direction according to the blur angle to obtain a blur length, including:
determining a fuzzy direction according to the fuzzy angle;
setting integer interval [0, n]And continuously calculating K ═ r in the integer interval
1/r
2(ii) a Wherein
And H
1=M
20+M
02,
And selecting the interval value [0, l ] corresponding to the minimum value K, and taking l as the fuzzy length.
Further, constructing a blur kernel according to the blur angle and the blur length, comprising:
calculating a point spread function psf (theta, l) according to the fuzzy length and the fuzzy angle, and constructing a fuzzy core;
taking the distance from each point in the fuzzy kernel to the fuzzy length as the weight of the point;
and carrying out normalization processing on the fuzzy core.
Further, constructing a fuzzy core, comprising:
calculating the rotation radius of the fuzzy core according to the fuzzy length;
calculating an x component and a y component according to the rotation radius and the fuzzy angle;
calculating a coordinate stepping value according to the fuzzy angle;
generating two-dimensional grid point coordinates through the x component, the y component and the coordinate stepping value;
calculating the distance from the grid point to the fuzzy direction;
calculating the distance from the grid point to the origin;
screening abnormal grid points, wherein the distance from the grid points to the origin is larger than the rotation radius, and the distance from the grid points to the fuzzy direction is smaller than the fuzzy length;
calculating the distance from the abnormal grid point to the fuzzy direction;
and establishing a matrix as a fuzzy core.
Further, the LED screen picture is deblurred by adopting a fuzzy core, and the deblurring processing method comprises the following steps:
according to an iterative algorithm
The picture of the LED screen is deblurred, wherein,
for the current image of the iteration,
and P is a fuzzy core and Y is a picture of the LED screen as a result of the previous iteration.
Further, the picture of the LED screen is a single-channel image under an R channel, a G channel or a B channel.
The invention also provides a device for removing image shake in the process of correcting the LED screen, which comprises the following components: the image deblurring method comprises a fuzzy kernel estimation sample acquisition module, a fuzzy angle acquisition module, a fuzzy length acquisition module, a fuzzy kernel construction module and an image deblurring module, wherein:
the fuzzy kernel estimation sample acquisition module is connected with the fuzzy angle acquisition module and used for acquiring the pictures of the LED screen and selecting partial pictures of the LED screen as fuzzy kernel estimation samples;
the fuzzy angle acquisition module is connected with the fuzzy kernel estimation sample acquisition module, the fuzzy length acquisition module and the fuzzy kernel construction module and is used for carrying out frequency domain conversion on the fuzzy kernel estimation sample, acquiring the distribution characteristics of high-frequency components, quantizing the distribution characteristics, acquiring a fuzzy angle and determining a fuzzy direction;
the fuzzy length acquisition module is connected with the fuzzy angle acquisition module and the fuzzy kernel construction module and used for carrying out iterative estimation according to the fuzzy angle and the fuzzy direction to acquire the fuzzy length;
the fuzzy kernel construction module is connected with the fuzzy angle acquisition module, the fuzzy length acquisition module and the picture deblurring processing module and is used for constructing a fuzzy kernel according to the fuzzy angle and the fuzzy length;
and the picture deblurring processing module is connected with the fuzzy core construction module and is used for deblurring the picture of the LED screen by adopting the fuzzy core.
The image de-jittering method and device in the LED screen correction process are applied to a specific application scene of LED screen correction, the images shot by camera jittering during the LED screen correction are de-jittered, the time domain and the frequency domain of the images are transformed, the distribution characteristics of the images are convenient to extract, further analysis is carried out, and the parameter accurate estimation of high-resolution images is realized; compared with the existing jitter removal methods such as inverse filtering and wiener filtering, the method has the advantages of high parameter estimation accuracy, clearer deblurred image, no ripple noise, local pixel degradation and the like, improves the image jitter removal effect, and improves the quality and efficiency of LED screen correction.
Detailed Description
The technical solutions in the embodiments of the present invention will be described clearly and completely with reference to the accompanying drawings in 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.
An image de-jittering method in an LED screen correction process according to an embodiment of the present invention, as shown in fig. 1, includes the following steps:
step S10: and acquiring the picture of the LED screen, and selecting a part of the picture of the LED screen as a fuzzy core estimation sample.
In the method of the embodiment of the invention, the picture of the LED screen is obtained by shooting the LED screen to be detected by the camera, and the image debouncing process of the method is carried out on the picture of the LED screen. The acquisition mode of the picture of the LED screen can be realized by connecting the LED screen with a camera in a wired communication manner or in a wireless communication manner. The method of the embodiment can be used in the single-box correction process of the LED screen, and can also be used in the whole screen correction process.
Before the entire LED screen picture is subjected to debounce, a part of the picture is selected as a blur kernel estimation sample, wherein the blur kernel estimation sample is 1/16-1/2 of the LED screen picture, such as 1/16, 1/12, 1/9, 1/6, 1/4, 1/2 and the like, and is specifically determined according to the size of the picture of the LED screen.
Step S20: and carrying out frequency domain conversion on the fuzzy kernel estimation sample, and acquiring the distribution characteristics of the high-frequency components.
Specifically, as shown in fig. 2, in step S20 in this embodiment, performing frequency domain conversion on the fuzzy core estimation sample, and acquiring a distribution characteristic of the high-frequency component specifically includes:
step S201: and converting the fuzzy kernel estimation sample from a space domain into a frequency domain through Fourier transform to obtain a frequency domain image.
The frequency domain conversion formula of the present embodiment can refer to:
wherein f (x, y) represents a matrix of size M × N, x being 0, 1,2,.. said., M-1, y being 0, 1,2,.. said., N-1. F (u, v) represents a fourier transform of F (x, y), a coordinate system in which F (u, v) is located is called a frequency domain, and a matrix of M × N defined by u ═ 0, 1, 2.. 9., M-1 and v ═ 0, 1, 2.. once, N-1 is called a frequency domain matrix; the coordinate system where f (x, y) is located is called a spatial domain, and a matrix of M × N defined by x 0, 1, 2.. the matrix of M-1 and y 0, 1, 2.. the matrix of N-1 is called a spatial domain matrix, and obviously, the size of the frequency domain matrix is the same as the size of the original spatial domain matrix.
Step S202: and carrying out binarization processing on the frequency domain image to obtain the distribution characteristics of the high-frequency components.
The distribution characteristics are the characteristics of the fourier spectrum, such as the contour, the morphology and the like, shown on the frequency domain image after the binarization processing, the binarization image is a black-and-white image, wherein the brighter region contour is the distribution characteristics, as shown in fig. 3. The form of the high frequency component reflects the actual degree of blurring of the image picture, and is therefore used as a feature of the blur parameter estimation.
Step S30: and quantizing the distribution characteristics, acquiring a fuzzy angle and determining a fuzzy direction.
Specifically, as shown in fig. 4, in step S30 of the embodiment of the present invention, the quantifying the distribution characteristic to obtain the blur angle specifically includes the following steps:
step S301: in the distribution feature, quantification is performed by the central moment.
The central moment is obtained by counting all pixels in the image, and the calculation method is as follows:
second moment:
M20=∑I∑Ji2·V(i,j) (2)
M02=∑I∑Jj2·V(i,j) (3)
M11=∑I∑Ji·j·V(i,j) (4)
first moment:
M10=∑I∑Ji·V(i,j) (5)
M01=∑I∑Jj·V(i,j) (6)
zero order moment:
M00=∑I∑JV(i,j) (7)
where V (i, j) represents a gray scale value with coordinates (i, j) on the image.
Step S302: and estimating the fuzzy angle according to the quantization result.
The blurring angle is theta, and
wherein:
wherein the content of the first and second substances,
x
cand y
cIs the centroid coordinate of the distribution feature.
Step S40: and carrying out iterative estimation according to the fuzzy angle and the fuzzy direction to obtain the fuzzy length.
Specifically, as shown in fig. 5, the step S40 performs iterative estimation in the blur direction according to the blur angle to obtain the blur length, which specifically includes:
step S401: and determining the blurring direction according to the blurring angle.
After the blurring angle θ is obtained through the above steps, the direction of the blurring angle θ can be known through the numerical value of the blurring angle θ, that is, the blurring direction. For example, θ is 45 °, and the blurring direction is the result of a 45 ° counterclockwise rotation with 0 ° horizontal to the right as the x-axis
Step S402: setting integer interval [0, n]And continuously calculating K ═ r in the integer interval1/r2。
The embodiment adopts an iteration mode of continuous calculation, can obtain the optimal value, solves the problem of estimation deviation in an actual application scene compared with the existing single parameter estimation mode, and reduces system errors and noise possibly introduced in the traditional estimation.
Step S403: and selecting the interval value [0, l ] corresponding to the minimum value K, and taking l as the fuzzy length.
For example, if the set integer interval [0, n ] is [0, 50] and K is the minimum value when the interval value is [0, 10], the corresponding l is 10, which is the blurring length.
Step S50: and constructing a fuzzy kernel according to the fuzzy angle and the fuzzy length.
Specifically, as shown in fig. 6, the constructing a blur kernel according to the blur angle and the blur length in step S50 specifically includes:
step S501: and calculating a point spread function psf (theta, l) according to the fuzzy length and the fuzzy angle, and constructing a fuzzy core.
The point spread function is a mathematical description function of the blurring degree of pixels in the image, and the jittered image can be represented as a convolution of a sharp image and the point spread function, so that the sharp image can be obtained by solving the point spread function in a deconvolution mode.
The blur length l and the blur angle θ have been obtained in the above calculation step, and mathematical modeling is performed on the two parameters to calculate the point spread function psf (θ, l) and construct a blur kernel.
Taking the blurring angle θ being 45 ° and the blurring length l being 10 as examples, the blurring kernel construction is realized by the following steps:
in the embodiment of the present invention, for convenience of correspondence when calculating the blur kernel, the embodiment uniformly corresponds the angle to the fourth quadrant for calculation. And after the calculation is finished, matrix inversion is carried out according to the cos value. For example, if the blur angle is 45 °, the inversion is performed once, and if the blur angle is 135 °, the inversion is not performed.
As shown in fig. 7, in step S501, a blur kernel is constructed, specifically:
step S5011: and calculating the rotation radius of the fuzzy core according to the fuzzy length.
The radius of rotation is (10-1)/2-4.5.
Step S5012: the x-component and y-component are calculated from the radius of rotation and the blur angle.
The x component is calculated to be 4.5 × cos (45 °) +1 ≈ 4.18198 according to the rotation radius and the blur angle, the integer is 4, and the y component can be calculated to be 4 in the same way.
Step S5013: and calculating a coordinate stepping value according to the fuzzy angle.
The step of the point coordinates is determined to be 1 from the sign of cos (45 °).
Step S5014: two-dimensional grid point coordinates are generated by the x-component, y-component, and coordinate step values.
From the x-component, y-component, corresponding 5 x 5 two-dimensional grid point coordinates may be generated, as shown in tables 1 and 2 below, with table 1 representing the x-coordinate values for each grid point and table 2 representing the y-coordinate values for each grid point.
TABLE 1 grid points x coordinate values
0
|
1
|
2
|
3
|
4
|
0
|
1
|
2
|
3
|
4
|
0
|
1
|
2
|
3
|
4
|
0
|
1
|
2
|
3
|
4
|
0
|
1
|
2
|
3
|
4 |
TABLE 2 y coordinate values of grid points
Step S5015: the distances of the grid points to the blur direction are calculated.
Assuming that the fuzzy straight line passes through the origin, the slope of the fuzzy straight line can be determined by the known fuzzy angle, so that the distance from the grid point to the fuzzy direction, i.e. the distance from the point to the line, can be calculated, and the calculation result is shown in the following table 3:
TABLE 3 distances of grid points to blur direction
0
|
-0.7071
|
-1.4142
|
-2.1213
|
-2.8284
|
0.7071
|
1.1102e-16 |
-0.7071
|
-1.4142
|
-2.1213
|
1.4142
|
0.7071
|
2.2204e-16 |
-0.7071
|
-1.4142
|
2.1213
|
1.4142
|
0.7071
|
4.4409e-16 |
-0.7071
|
2.8284
|
2.1213
|
1.4142
|
0.7071
|
4.4409e-16 |
Step S5016: the distances of the grid points to the origin are calculated.
The results of calculating the distances from the grid points to the origin are shown in table 4 below:
TABLE 4 distances of grid points to origin
0
|
1
|
2
|
3
|
4
|
1
|
1.4142
|
2.2361
|
3.1623
|
4.1231
|
2
|
2.2361
|
2.8284
|
3.6056
|
4.4724
|
3
|
3.1623
|
3.6056
|
4.2426
|
5
|
4
|
4.1231
|
4.4721
|
5
|
5.6569 |
Step S5017: and screening abnormal grid points, wherein the distance from the grid points to the origin is larger than the rotation radius, and the distance from the grid points to the fuzzy direction is smaller than the fuzzy length.
The method comprises the steps of firstly screening three grid points with the distance values equal to 5 and 5.6569 according to the condition that the distance from the grid point to the origin in the table 4 is larger than the rotation radius, then screening the grid points with the distance from the grid point to the fuzzy direction smaller than the fuzzy length according to the three grid points and the table 3, wherein the three grid points are all abnormal grid points.
Step S5018: the distance of the abnormal grid point to the blurring direction is calculated.
As shown in fig. 8, the grid points are placed in the coordinate system, the abnormal grid points are T, Z and a1, and the distance from Z to the blur direction is the length of a line segment from Z2 to Z3, taking abnormal grid point Z as an example.
Step S5019: and establishing a matrix as a fuzzy core.
The step constructs a matrix with the size of (2 m-1,2 n-1) as a fuzzy core, and only initializes the size of the fuzzy core matrix and does not assign the size. The values of m and n refer to the radius of rotation in the previous example, the radius of rotation is 4.5, rounded to 5, so that m and n both take the value of 5. A matrix with a blur kernel size of 9 x 9 is created.
Step S502: and taking the distance from each point in the fuzzy core to the fuzzy length as the weight of the point.
The weight value uses the formula:
a calculation is performed where width represents the width set by the blur kernel, for example, taking the
value 1, and abs (dist (i, j)) represents the distance of the point from the origin. The weight matrix is calculated with the previous example as follows:
TABLE 5 weight matrix
0
|
0.161983
|
0
|
0
|
0
|
0.161983
|
1
|
0.292893
|
0
|
0
|
0
|
0.292893
|
1
|
0.292893
|
0
|
0
|
0
|
0.292893
|
1
|
0.292893
|
0
|
0
|
0
|
0.292893
|
1 |
Step S503: and carrying out normalization processing on the fuzzy core.
The diagonal angle of the weight is rotated by 180 degrees and filled into the fuzzy kernel, and the other adjacent blocks are supplemented with 0, so that the product of all numerical values in the fuzzy kernel satisfies
Where K represents the blur kernel of size (2 m-1,2 n-1) and K (i, j) represents the blur kernel coordinates (i, j), so the normalized blur kernel matrix is shown in Table 6:
TABLE 6 fuzzy core matrix
Step S60: and deblurring the picture of the LED screen by using a fuzzy core.
In the step, a clear image is finally output through deconvolution operation. As shown in fig. 9 and 10, the images before and after the deblurring process are shown, respectively.
The image dithering removing method in the LED screen correction process is applied to a specific application scene of LED screen correction, the dithering of a picture shot by camera dithering during the LED screen correction is removed, time domain and frequency domain transformation is carried out on the picture, the distribution characteristics of the picture are convenient to extract, further analysis is carried out, and accurate parameter estimation of a high-resolution picture is realized; compared with the existing jitter removal methods such as inverse filtering and wiener filtering, the method has the advantages of high parameter estimation accuracy, clearer deblurred image, no ripple noise, local pixel degradation and the like, improves the image jitter removal effect, and improves the quality and efficiency of LED screen correction.
Specifically, in step S60, the deblurring processing is performed on the LED screen picture by using a blur kernel, which specifically includes: according to an iterative algorithm
The picture of the LED screen is deblurred, wherein,
for the current image of the iteration,
and P is a fuzzy core and Y is a picture of the LED screen as a result of the previous iteration. In the embodiment of the invention, an iterative calculation mode is adopted during deblurring calculation, so that the output image is clearer and the deblurring effect is better.
Specifically, the pictures of the LED screens in all the embodiments of the present invention are single-channel images in an R channel, a G channel, or a B channel. Therefore, after the LED screen picture with the single channel is subjected to deblurring processing by the method, RGB three channels are merged, and the final clear colorful LED screen picture is output, so that the purpose of removing the shake of the original picture shot by the camera is realized.
The invention also discloses an image debouncing device 100 in the LED screen correction process, as shown in fig. 11, including a blur kernel estimation sample obtaining module 101, a blur angle obtaining module 102, a blur length obtaining module 103, a blur kernel constructing module 104 and a picture deblurring module 105, wherein:
and the fuzzy kernel estimation sample acquisition module 101 is connected with the fuzzy angle acquisition module 102 and is used for acquiring the pictures of the LED screen and selecting part of the pictures of the LED screen as the fuzzy kernel estimation samples.
The fuzzy angle obtaining module 102 is connected to the fuzzy kernel estimation sample obtaining module 101, the fuzzy length obtaining module 103 and the fuzzy kernel constructing module 104, and is configured to perform frequency domain conversion on the fuzzy kernel estimation sample, obtain distribution characteristics of the high frequency component, quantize the distribution characteristics, obtain a fuzzy angle, and determine a fuzzy direction.
The blur length obtaining module 103 is connected to the blur angle obtaining module 102 and the blur kernel constructing module 104, and is configured to perform iterative estimation according to the blur angle and the blur direction to obtain the blur length.
The blur kernel constructing module 104 is connected with the blur angle acquiring module 102, the blur length acquiring module 103 and the picture deblurring processing module 105, and is configured to construct a blur kernel according to the blur angle and the blur length.
And the picture deblurring processing module 105 is connected with the blur kernel constructing module 104 and is used for deblurring the picture of the LED screen by using the blur kernel.
The method for removing the jitter of the picture by the device 100 of the embodiment of the invention is specifically realized according to the following steps:
step S10: and acquiring the picture of the LED screen, and selecting a part of the picture of the LED screen as a fuzzy core estimation sample.
Step S20: and carrying out frequency domain conversion on the fuzzy kernel estimation sample, and acquiring the distribution characteristics of the high-frequency components.
Wherein, step S20 specifically includes:
step S201: and converting the fuzzy kernel estimation sample from a space domain into a frequency domain through Fourier transform to obtain a frequency domain image.
Step S202: and carrying out binarization processing on the frequency domain image to obtain the distribution characteristics of the high-frequency components.
Step S30: and quantizing the distribution characteristics, acquiring a fuzzy angle and determining a fuzzy direction.
Wherein, step S30 specifically includes:
step S301: in the distribution feature, quantification is performed by the central moment.
Step S302: and estimating the fuzzy angle according to the quantization result.
Step S40: and carrying out iterative estimation according to the fuzzy angle and the fuzzy direction to obtain the fuzzy length.
Wherein, step S40 specifically includes:
step S401: and determining the blurring direction according to the blurring angle.
Step S402: setting integer interval [0, n]And continuously calculating K ═ r in the integer interval1/r2。
Step S403: and selecting the interval value [0, l ] corresponding to the minimum value K, and taking l as the fuzzy length.
Step S50: and constructing a fuzzy kernel according to the fuzzy angle and the fuzzy length.
Wherein, step S50 specifically includes:
step S501: and calculating a point spread function psf (theta, l) according to the fuzzy length and the fuzzy angle, and constructing a fuzzy core.
Wherein, step S501 specifically includes:
step S5011: and calculating the rotation radius of the fuzzy core according to the fuzzy length.
Step S5012: the x-component and y-component are calculated from the radius of rotation and the blur angle.
Step S5013: and calculating a coordinate stepping value according to the fuzzy angle.
Step S5014: two-dimensional grid point coordinates are generated by the x-component, y-component, and coordinate step values.
Step S5015: the distances of the grid points to the blur direction are calculated.
Step S5016: the distances of the grid points to the origin are calculated.
Step S5017: and screening abnormal grid points, wherein the distance from the grid points to the origin is larger than the rotation radius, and the distance from the grid points to the fuzzy direction is smaller than the fuzzy length.
Step S5018: the distance of the abnormal grid point to the blurring direction is calculated.
Step S5019: and establishing a matrix as a fuzzy core.
Step S502: and taking the distance from each point in the fuzzy core to the fuzzy length as the weight of the point.
Step S503: and carrying out normalization processing on the fuzzy core.
Step S60: and deblurring the picture of the LED screen by using a fuzzy core.
For the description of the execution of each step, reference may be directly made to the embodiments of the foregoing method, which are not described herein again.
The image debouncing device in the LED screen correction process is applied to a specific application scene of LED screen correction, and is used for debouncing a picture shot by camera shake during the LED screen correction, and converting the time domain and the frequency domain of the image, so that the distribution characteristics of the image can be conveniently extracted, further analysis is carried out, and the parameter accurate estimation of a high-resolution image is realized; compared with the existing jitter removal methods such as inverse filtering and wiener filtering, the method has the advantages of high parameter estimation accuracy, clearer deblurred image, no ripple noise, local pixel degradation and the like, improves the image jitter removal effect, and improves the quality and efficiency of LED screen correction.
The embodiment of the invention also provides a memory, wherein the memory is stored with computer instructions, and the computer instructions are executed by the processor to realize the image de-jittering method in the embodiment.
As shown in fig. 12, the electronic device 200 according to an embodiment of the present invention includes a memory 201 and a processor 202, where the memory 201 stores computer instructions; the processor 202 is configured to execute computer instructions to implement the image de-jittering method in the above embodiments. The electronic device 200 in this embodiment is a computer.
The image de-jittering method is realized by the following steps:
step S10: and acquiring the picture of the LED screen, and selecting a part of the picture of the LED screen as a fuzzy core estimation sample.
Step S20: and carrying out frequency domain conversion on the fuzzy kernel estimation sample, and acquiring the distribution characteristics of the high-frequency components.
Wherein, step S20 specifically includes:
step S201: and converting the fuzzy kernel estimation sample from a space domain into a frequency domain through Fourier transform to obtain a frequency domain image.
Step S202: and carrying out binarization processing on the frequency domain image to obtain the distribution characteristics of the high-frequency components.
Step S30: and quantizing the distribution characteristics, acquiring a fuzzy angle and determining a fuzzy direction.
Wherein, step S30 specifically includes:
step S301: in the distribution feature, quantification is performed by the central moment.
Step S302: and estimating the fuzzy angle according to the quantization result.
Step S40: and carrying out iterative estimation according to the fuzzy angle and the fuzzy direction to obtain the fuzzy length.
Wherein, step S40 specifically includes:
step S401: and determining the blurring direction according to the blurring angle.
Step S402: setting integer interval [0, n]And continuously calculating K ═ r in the integer interval1/r2。
Step S403: and selecting the interval value [0, l ] corresponding to the minimum value K, and taking l as the fuzzy length.
Step S50: and constructing a fuzzy kernel according to the fuzzy angle and the fuzzy length.
Wherein, step S50 specifically includes:
step S501: and calculating a point spread function psf (theta, l) according to the fuzzy length and the fuzzy angle, and constructing a fuzzy core.
Wherein, step S501 specifically includes:
step S5011: and calculating the rotation radius of the fuzzy core according to the fuzzy length.
Step S5012: the x-component and y-component are calculated from the radius of rotation and the blur angle.
Step S5013: and calculating a coordinate stepping value according to the fuzzy angle.
Step S5014: two-dimensional grid point coordinates are generated by the x-component, y-component, and coordinate step values.
Step S5015: the distances of the grid points to the blur direction are calculated.
Step S5016: the distances of the grid points to the origin are calculated.
Step S5017: and screening abnormal grid points, wherein the distance from the grid points to the origin is larger than the rotation radius, and the distance from the grid points to the fuzzy direction is smaller than the fuzzy length.
Step S5018: the distance of the abnormal grid point to the blurring direction is calculated.
Step S5019: and establishing a matrix as a fuzzy core.
Step S502: and taking the distance from each point in the fuzzy core to the fuzzy length as the weight of the point.
Step S503: and carrying out normalization processing on the fuzzy core.
Step S60: and deblurring the picture of the LED screen by using a fuzzy core.
The present invention has been further described with reference to specific embodiments, but it should be understood that the detailed description should not be construed as limiting the spirit and scope of the present invention, and various modifications made to the above-described embodiments by those of ordinary skill in the art after reading this specification are within the scope of the present invention.