CN113327206A - Image fuzzy processing method of intelligent power transmission line inspection system based on artificial intelligence - Google Patents
Image fuzzy processing method of intelligent power transmission line inspection system based on artificial intelligence Download PDFInfo
- Publication number
- CN113327206A CN113327206A CN202110618991.XA CN202110618991A CN113327206A CN 113327206 A CN113327206 A CN 113327206A CN 202110618991 A CN202110618991 A CN 202110618991A CN 113327206 A CN113327206 A CN 113327206A
- Authority
- CN
- China
- Prior art keywords
- image
- gradient
- average gray
- difference
- calculating
- 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
- 238000007689 inspection Methods 0.000 title claims abstract description 15
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 14
- 238000003672 processing method Methods 0.000 title claims abstract description 14
- 230000005540 biological transmission Effects 0.000 title claims abstract description 13
- 238000000034 method Methods 0.000 claims description 20
- 238000004364 calculation method Methods 0.000 claims description 14
- 230000000875 corresponding effect Effects 0.000 description 36
- 230000008569 process Effects 0.000 description 5
- 230000000694 effects Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000003384 imaging method Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 230000002708 enhancing effect Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000011084 recovery Methods 0.000 description 2
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000000717 retained effect Effects 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
- G06T5/73—Deblurring; Sharpening
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention relates to an image fuzzy processing method of an intelligent power transmission line inspection system based on artificial intelligence, which comprises the steps of obtaining the gradient distribution of a standard image and a fuzzy image, calculating the fuzziness of the image according to the difference of the gradient distribution of the standard image and the fuzzy image, and carrying out image enhancement processing on the fuzzy image to obtain a semi-clear image; then clustering the gray difference data of each pixel in the calculated blurred image and the semi-sharp image to obtain the average gray difference and weight of each type, calculating the sum of the average gray differences of the whole image, calculating the block size of the blurred image by combining the average gray difference of each type to obtain a sliding window, performing sliding window operation on the semi-sharp image to obtain an image block corresponding to a window equal to the sum of the average gray differences as a final Patch block, calculating a blur kernel, and deblurring the semi-sharp image to obtain the sharp image. Namely, the invention reserves the image information of the blurred image when deblurring is carried out, and can effectively avoid the subsequent over-processing of the partial area of the semi-clear image.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to an image fuzzy processing method of an intelligent power transmission line inspection system based on artificial intelligence.
Background
With the development of the artificial intelligence technology era, people have stronger dependence on power resources, so that the stability of voltage is particularly important to maintain. Under traditional operation mode, the power worker need regularly patrol high-voltage line, and this kind of mode has not only brought a series of safety risks for the power worker, and needs a large amount of workman to accomplish, and efficiency is also comparatively low when consuming great cost of labor.
At present, although a professional high-voltage wire inspection robot is used for replacing workers to carry out all-weather uninterrupted inspection, the labor cost is correspondingly saved while the efficiency is improved, but the environment influence is great when the robot runs. The robot is loaded with an industrial camera, and can travel along the high-voltage wire and shoot the video of the high-voltage wire after being fixed on the high-voltage wire. However, the high-voltage wire is generally located at a high position, and particularly when the wind force is strong, the machine is disturbed by the wind force during the traveling process to shake, so that the imaging picture is blurred, and the judgment is affected.
The commonly used technology for blurred images is to use wiener filtering or Richardson-Lucy algorithm, and perform inverse convolution operation by using the imaging of an actual lens and the diffraction limit PSF of the lens to obtain the imaging of an ideal lens. Although the method has ideal effect, the PSF parameters (fuzzy length and scale) of the lens need to be known in advance, and the PSF parameters cannot be measured in advance in many cases. Although there are a series of methods for estimating camera blur kernel to obtain PSF parameters laterally, the method still has the problems of large calculation amount, unsatisfactory result and the like, and because of the randomness of shaking, the blur degree of each part in the image is not the same, and the uniform parameter recovery of the image sometimes obtains a poor recovery effect.
Disclosure of Invention
The invention aims to provide an image blurring processing method of an artificial intelligence-based intelligent inspection system for a power transmission line, which is used for solving the problem that the effect is poor when the unified processing of images is carried out because the difference of blurring degrees of all parts in the images is not considered in the prior art.
The invention provides an image fuzzy processing method of an intelligent power transmission line inspection system based on artificial intelligence, which comprises the following steps:
acquiring a blurred image to be processed and a standard image, wherein the standard image is a clear image similar to the blurred image;
secondly, performing image enhancement processing on the blurred image to obtain an enhanced semi-sharp image;
thirdly, calculating gray difference data of each pixel in the blurred image and the semi-sharp image, clustering the gray difference data to obtain average gray differences of various types after clustering, setting corresponding weights for different types, and calculating the sum of the average gray differences of the whole image according to the average gray differences of various types and the set weights
Wherein the sum of the average gray differences of the whole image
Comprises the following steps:
wherein, deltaTIs the sum of the average gray differences of the F types;
representing the weight, wherein the smaller the average gray difference is, the larger the weight occupied correspondingly;
step four, calculating the size of the fuzzy image block according to the sum of the average gray difference of the whole image and the average gray difference of each type;
the calculation formula of the block size of the blurred image is as follows:
wherein G isjSynthesizing the sum of the average gray difference to the average gray difference for each cluster
The contribution value of (c).
Step five, the fuzzy graph is processedThe size of the image block is processed to be used as the size of a sliding window of the semi-clear image, the sliding window operation is carried out on the semi-clear image, the average gray difference of the image in the window is calculated, after the semi-clear image is traversed, the length and the width of the sliding window are changed, the average gray difference of the image in the window is continuously calculated, and the average gray difference of the image in the window is compared with the average gray difference of the image in the window
When the difference value between the two is smaller than a set value, the image block of the semi-clear image corresponding to the average gray difference of the image in the window is used as a final Patch block;
and step six, calculating a fuzzy core of the obtained Patch block, and performing deblurring processing on the semi-clear image according to the fuzzy core to obtain a clear image.
Further, the image enhancement processing process includes:
1) respectively calculating the gradient distribution of the blurred image and the standard image, acquiring gradient distribution curves of corresponding images, calculating the curvature of pixel points corresponding to the same gradient on each curve, and calculating the curvature difference of points corresponding to the same gradient on the two curves; calculating the variance of all the curvature differences according to the curvature differences to determine the similarity of the gradient distribution curves of the two images;
the similarity is A ═ e-δWherein δ is the variance of the calculated curvature difference;
2) respectively acquiring the number of pixel points of the fuzzy image and the standard image when the gradient is 0, and calculating the gradient curve similarity of the two images;
the gradient curve similarity B is as follows:
wherein N is the total gradient value corresponding to the image with the largest gradient range in the two gradient images; n isc，iThe number of pixel points corresponding to the ith gradient on the current real-time gradient image is determined; n iss，iRepresenting standard gradient imagesThe number of pixel points corresponding to the ith gradient; n isTThe total number of the pixel points on the current real-time gradient image is obtained;
3) calculating the image fuzziness according to the obtained similarity and the gradient curve similarity;
wherein the image blur degree is:
Z＝1-A·B
wherein A is similarity, B is gradient curve similarity; (ii) a
4) Optimizing pixels of the blurred image according to the calculated image blur degree to obtain pixel values of the optimized image; wherein the pixel values of the optimized image are:
wherein M isi(x, y) represents the gray value finally updated after the ith pixel integrates the horizontal direction and the vertical direction; gix，Giy is the pixel value G updated by the degree of blur Z corresponding to the horizontal and vertical directions, respectivelyix：
Gix＝gix+∈ix(1+Z)
Wherein, giRepresents the gray value of the ith pixel, ∈iIs the gradient between the ith pixel and its neighboring pixels
Further, in step three, the calculation of the sum of the average gray differences is also performed according to the areas of the different classes:
wherein S is the total area of the image, SjIs the area of the clustered jth class.
Further, the calculation formula of the block size of the blurred image is as follows:
wherein,
then G isjSynthesizing the average gray scale difference and the corresponding area-to-average gray scale difference for each cluster
The contribution value of (c).
Further, the size of the sliding window for performing the sliding window operation is:
wherein x and y are the length and width of the sliding window respectively.
The invention has the beneficial effects that:
the scheme of the image blurring processing method of the invention reserves the image information of the blurred image when deblurring is carried out, and can effectively avoid the subsequent over processing of the partial area of the semi-clear image.
Meanwhile, the method and the device utilize the characteristics of gradient distribution and the aggregation of pixel points during image blurring to obtain the blurring degree of the image, thereby optimizing and adjusting the blurred image and enhancing the definition of the blurred image.
According to the method, the final patch block is determined by introducing the factor of the area occupied by each type, so that the deblurring accuracy can be improved to a certain extent.
Drawings
In order to more clearly illustrate the technical solution of the embodiment of the present invention, the drawings needed to be used in the embodiment will be briefly described as follows:
fig. 1 is a flow chart of the method of the image fuzzy processing method of the intelligent power transmission line inspection system based on artificial intelligence.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be described below with reference to the drawings in the embodiments of the present invention.
The invention provides an image fuzzy processing method of an intelligent power transmission line inspection system based on artificial intelligence, which comprises the following steps as shown in figure 1:
acquiring a blurred image to be processed and a standard image, wherein the standard image is a clear image similar to the blurred image;
in this embodiment, the standard image is a clear image close to the blurred image, and the close clear image is a multi-frame image taken when the robot is stable in a good weather condition and in a windless environment, and 20 frames may be selected.
Secondly, performing image enhancement processing on the blurred image to obtain an enhanced semi-sharp image;
the image enhancement processing in the above steps comprises:
1) respectively calculating the gradient distribution of the blurred image and the standard image, acquiring gradient distribution curves of corresponding images, calculating the curvature of pixel points corresponding to the same gradient on each curve, and calculating the curvature difference of points corresponding to the same gradient on the two curves; calculating the variance of all the curvature differences according to the curvature differences to determine the similarity of the gradient distribution curves of the two images;
the similarity is A ═ e-δWhere δ is the variance of the calculated curvature difference.
In the above step, the obtained gradient distribution of the blurred image and the standard image is normalized to make the two distribution curves in the same range.
In this embodiment, the blurred image and the standard image are further processed, that is, after the captured image is converted into a grayscale image, a gradient corresponding to the image is calculated, where the gradient is a difference between two adjacent pixels in each image.
It should be noted that a clear image contains many smooth regions and is low in noise, so that pixels with a gradient close to zero occupy most of the clear image, and therefore, the gradient of the clear natural image conforms to a heavy tail distribution form. The image is clear, so the object boundary is obvious, the gradient of a plurality of pixels is large, the distribution diagram is approximately as shown in figure 1 and is the gradient distribution of the standard image acquired when the robot is stable; whereas blurred images tend to have more pixel gradients towards zero as the edges are blurred out.
2) Respectively acquiring the number of pixel points of the fuzzy image and the standard image when the gradient is 0, and calculating the gradient curve similarity of the two images;
the gradient curve similarity B is:
wherein N is the total gradient value corresponding to the image with the largest gradient range in the two gradient images; n isc，iThe number of pixel points corresponding to the ith gradient on the current real-time gradient image is determined; n iss，iExpressing the number of pixel points corresponding to the ith gradient on the standard gradient image; n isTThe total number of the pixel points on the current real-time gradient image is obtained.
3) Calculating the image ambiguity according to the obtained first similarity and the gradient curve similarity;
wherein the image blur degree is:
Z＝1-A·B
wherein A is similarity, B is gradient curve similarity;
the higher the degree of blur of the images in the above embodiments is, the lower the similarity between the two images is, and accordingly the blurred image is more blurred, otherwise the blurred image is clearer.
4) Optimizing pixels of the blurred image according to the calculated image blur degree to obtain pixel values of the optimized image; wherein the pixel values of the optimized image are:
wherein M isi(x, y) represents the i-th pixel after integrating the horizontal direction and the vertical directionThe newly obtained gray value; gix，Giy is the pixel value after the update of the degree of blur Z corresponding to the horizontal and vertical directions, Gix is:
Gix＝gix+∈ix(1+Z)
wherein, giRepresents the gray value of the ith pixel, ∈iIs the gradient between the ith pixel and its neighboring pixels.
Since the pixel values are integers, G calculated in the above formula is usediThe value of x is rounded down and when GiWhen x exceeds 255, let Gix＝255。
The ambiguity Z obtained by the method comprehensively reflects the ambiguity of the original blurred image, and the problem that the ambiguity is not accurately calculated from single factors such as curve similarity and corresponding gradient density is solved; and not only on the basis of the method, compared with the conventional image enhancement process, the method effectively improves the definition of the original image in the subsequent image enhancement process, but also further improves the definition of the image for obtaining the most ideal patch block subsequently.
A specific embodiment is given below, taking an image as an example, to perform a detailed description of image enhancement:
1) if a certain pixel value in the horizontal direction in the image is 90 and the adjacent pixels in the horizontal direction are 80,80 and 80 respectively, the corresponding gradients in the horizontal direction are 10,0 and 0 respectively;
the gradient in the above is calculated only by taking the absolute value of the change, indicating the magnitude of the change.
2) The gray value of the pixel in the new image after optimization is as follows: gix is 90+10(1+0.6) is 106, wherein the blurring degree Z of the image where the pixel is located in the first step is 0.6; then the updated horizontal direction corresponding pixel gray values are 106, 80,80, 80.
According to the optimization mode, the contrast of the image is enhanced, particularly the outline and the edge of an object in the image are greatly distinguished from the background, and the purpose of enhancing the definition of the image is achieved.
Because only the horizontal direction is calculated in the steps, the vertical direction can be obtained in the same way; since the square sum is open in practical application and results in large calculation amount, the operation of approximating the square and square root according to the absolute value in the conventional approximation method in the invention reduces the calculation amount, and the calculation formula is as follows:
Mi(x，y)＝|Gix|+|Giy for the above embodiment, it should be noted that the difference in gray scale value between the gray scale values of the pixels 90 and 80 is 10, which is not obvious, and the contour is necessarily blurred with a large group of consecutive pixels having a gray scale value of 90. Because the gray value of the adjacent pixel is changed, the corresponding gradient has a value, if the gray value of the adjacent pixel is not changed, the gradient is zero, therefore, the gray value and the corresponding gray value of the pixel are optimized through the characteristic value Z and then added, the gray value is not changed, the pixel value is not changed, and the gray value with the gradient value is increased;
thirdly, calculating gray difference data of each pixel in the blurred image and the semi-sharp image, clustering the gray difference data to obtain average gray differences of various types after clustering, setting corresponding weights for different types, and calculating the sum of the average gray differences of the whole image according to the average gray differences of various types and the set weights
Wherein the sum of the average gray differences of the whole image
Comprises the following steps:
wherein, deltaTIs the sum of the average gray differences of the F types;
represents a weight, alsoThat is, the smaller the average gray scale difference is, the larger the weight occupied by the average gray scale difference is.
For example, if F is five types, the average gray scale difference is 1, 2, 3, 4, 5 from small to large, and then delta isT15, each corresponding weight is
It should be noted that, because the gray level difference between the blurred image and the semi-sharp image is used instead of being calculated by the semi-sharp image and the standard image, the original blurred image is subjected to image enhancement processing, the contrast of the contour in the processed image is clearer and more obvious, and then the corresponding pixel gray level value is changed differently. At this time, if the processed semi-sharp image is compared with the standard image, even if a partial region of the image is already enhanced to be very sharp and even exceeds the definition of a corresponding region of the standard image, because the pixel gray values of the semi-sharp image and the standard image are not the same, a person may misunderstand that the enhanced region is still different from the standard region and still has a certain degree of blur, thereby affecting the subsequent determination result.
Therefore, by calculating the average gray level difference between the blurred image and the semi-sharp image, according to the principle that the more blurred the image is, the smaller the pixel gradient of the corresponding region is, and the smaller the gray level difference after compensation is, the smaller the gray level difference between the blurred image and the semi-sharp image is, the more blurred the region is, that is, the region is not effectively enhanced, so that in the subsequent calculation, the weight ratio corresponding to the average gray level difference of the region is larger, that is, when deblurring is performed, the image information of the blurred image is retained, and the subsequent over-processing of partial regions of the semi-sharp image can be effectively avoided; the average gray difference corresponding region reflects a region which can represent the most enhanced image blurring degree; it should be emphasized that the average gray level in the present invention is only used to find the area index with a larger degree of corresponding blur.
As other implementation modes, the invention also introduces the factor of area, namely, the area difference of various types of clustered images occupying the total blurred image is considered, and further the weight is generatedThe corresponding effect, i.e. the sum of the average gray differences of the whole image
Comprises the following steps:
wherein S is the total area of the image, SjIs the area of the clustered jth class.
Step four, calculating the size of the fuzzy image block according to the sum of the average gray difference of the whole image and the average gray difference of each type;
the calculation formula of the block size of the blurred image is as follows:
wherein,
then G isjThe sum of the average gray level difference and the corresponding area factor influenced average gray level difference is integrated for each cluster
The contribution value of (c).
The step is to determine the sliding window size W of the selected Patch block according to the contribution value of each cluster, namely to determine the size of the sliding window size area, and the sliding window area comprehensively considers the influence of each cluster on the average gray value, so that the obtained sliding window area and the corresponding average gray difference sum
And the close combination can determine a more accurate and effective Patch block.
As another embodiment, when the area factor is introduced, the calculation formula of the block size of the blurred image is as follows:
wherein,
then G isjThe sum of the average gray level difference and the corresponding area factor influenced average gray level difference is integrated for each cluster
The contribution value of (c).
Step five, the block size of the blurred image is processed to be used as the size of a sliding window of the semi-clear image, the sliding window operation is carried out on the semi-clear image, the average gray difference of the image in the window is calculated, after the semi-clear image is traversed, the length and the width of the sliding window are changed, the average gray difference of the image in the window is continuously calculated, and the average gray difference of the image in the window is compared with the average gray difference of the image in the window
When the difference value between the two is smaller than the set per unit value range, taking the image block of the semi-clear image corresponding to the average gray level difference of the image in the window as a final Patch block;
specifically, the blurred image block size W is converted into a common sliding window shape by previously setting the maximum length of the long side of the sliding window to twice the length of the short side of the sliding window
Wherein x and y are the length and width of the sliding window respectively.
It should be noted that, in the present invention, the window sliding operation with the step length of 1 is performed every time one window shape size is started from the first pixel on the enhanced image, the average gray level difference in the window is calculated, the size of the window is changed after the entire image is traversed, the step length is still one pixel for continuous traversal, and the finally obtained average gray level difference under each window is selected to be closest to the size of the window
The window (the difference between the two is within the range of the set per unit value, and the specific value of the per unit value can be set according to the actual situation) of (1) is used as the final Patch block.
And step six, calculating a fuzzy core of the obtained Patch block, and performing deblurring processing on the semi-clear image according to the fuzzy core to obtain a clear image.
In this embodiment, the estimation of the blur kernel is to estimate the convolution kernel by using a MAP method and a variational bayes method, and further calculate the blur kernel of the image.
The method specifically comprises the following steps:
1) considering the problem as a maximum a posteriori probability problem, a standard MAP objective function is used to solve the convolution kernel k, argmaxkp (k | p), where p is the blurred image block represented by the Patch block.
Where p (k | p) is the marginal probability with respect to the unknown resultant image L, i.e.
p(k|p)＝∫p(k，L|p)dL
2) Since the posterior probabilities are required for every possible k in the above steps and are added up to obtain p (k | p), the calculation amount for the marginal probability is very large.
The expression and the solution of the posterior probability are approximated according to the most general variational Bayes method in the invention, and the process is not described in detail because it is a known technology and is not the focus of the invention.
In the embodiment, the sharp image L can be reconstructed according to a standard deconvolution algorithm according to the obtained blur kernel and a known blurred image, and the final sharp image is obtained in an iterative manner by using the most classical Richardson-Lucy algorithm in the invention.
It should be noted that while the preferred embodiments of the present invention have been described, additional variations and modifications to these embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
Claims (5)
1. An image fuzzy processing method of an intelligent power transmission line inspection system based on artificial intelligence is characterized by comprising the following steps:
acquiring a blurred image to be processed and a standard image, wherein the standard image is a clear image similar to the blurred image;
calculating the gradient distribution of the standard image and the blurred image, calculating the blur degree of the image according to the difference of the gradient distribution of the standard image and the blurred image, and performing image enhancement processing on the blurred image according to the blur degree to obtain an enhanced semi-sharp image;
thirdly, calculating gray difference data of each pixel in the blurred image and the semi-sharp image, clustering the gray difference data to obtain average gray differences of various types after clustering, setting corresponding weights for different types, and calculating the sum of the average gray differences of the whole image according to the average gray of various types and the set weights
Wherein the sum of the average gray differences of the whole image
Comprises the following steps:
wherein, deltaTIs the sum of the average gray differences of the F types;
representing the weight, wherein the smaller the average gray difference is, the larger the weight occupied correspondingly;
step four, calculating the size of the fuzzy image block according to the sum of the average gray difference of the whole image and the average gray difference of each type;
the calculation formula of the block size of the blurred image is as follows:
wherein G isjSynthesizing the sum of the average gray difference to the average gray difference for each cluster
The contribution value of (c).
Step five, the block size of the blurred image is processed to be used as the size of a sliding window of the semi-clear image, the sliding window operation is carried out on the semi-clear image, the average gray difference of the image in the window is calculated, after the semi-clear image is traversed, the length and the width of the sliding window are changed, the average gray difference of the image in the window is continuously calculated, and the average gray difference of the image in the window is compared with the average gray difference of the image in the window
When the difference value between the two is smaller than a set value, the image block of the semi-clear image corresponding to the average gray difference of the image in the window is used as a final Patch block;
and step six, calculating a fuzzy core of the obtained Patch block, and performing deblurring processing on the semi-clear image according to the fuzzy core to obtain a clear image.
2. The image blurring processing method for the intelligent power transmission line inspection system based on the artificial intelligence as claimed in claim 1, wherein the image enhancement processing procedure is as follows:
1) respectively calculating the gradient distribution of the blurred image and the standard image, acquiring gradient distribution curves of corresponding images, calculating the curvature of pixel points corresponding to the same gradient on each curve, and calculating the curvature difference of points corresponding to the same gradient on the two curves; calculating the variance of all the curvature differences according to the curvature differences to determine the similarity of the gradient distribution curves of the two images;
the similarity is A ═ e-δWherein δ is the variance of the calculated curvature difference;
2) respectively acquiring the number of pixel points of the fuzzy image and the standard image when the gradient is 0, and calculating the gradient curve similarity of the two images;
the gradient curve similarity B is as follows:
wherein N is the total gradient value corresponding to the image with the largest gradient range in the two gradient images; n isc，iThe number of pixel points corresponding to the ith gradient on the current real-time gradient image is determined; n iss，iExpressing the number of pixel points corresponding to the ith gradient on the standard gradient image; n isTThe total number of the pixel points on the current real-time gradient image is determined;
3) calculating the image fuzziness according to the obtained similarity and the gradient curve similarity;
wherein the image blur degree is:
Z＝1-A·B
wherein A is similarity, B is gradient curve similarity;
4) optimizing pixels of the blurred image according to the calculated image blur degree to obtain pixel values of the optimized image;
wherein the pixel values of the optimized image are:
wherein M isi(x, y) represents the gray value finally updated after the ith pixel integrates the horizontal direction and the vertical direction; gix，Giy is the pixel value G updated by the degree of blur Z corresponding to the horizontal and vertical directions, respectivelyix：
Gix＝gix+∈ix(1+Z)
Wherein, giRepresents the gray value of the ith pixel, ∈iIs the gradient between the ith pixel and its neighboring pixels.
3. The image blurring processing method for the intelligent power transmission line inspection system based on the artificial intelligence as claimed in claim 1, wherein in step three, the calculation of the sum of the average gray level differences is further performed according to the areas of different classes:
wherein S is the total area of the image, SjIs the area of the clustered jth class.
4. The image blurring processing method for the intelligent inspection system for power transmission lines based on artificial intelligence as claimed in claim 3, wherein the calculation formula of the block size of the blurred image is as follows:
wherein,
then G isjSynthesizing the average gray scale difference and the corresponding area-to-average gray scale difference for each cluster
The contribution value of (c).
5. The image blurring processing method for the intelligent power transmission line inspection system based on the artificial intelligence as claimed in claim 1 or 4, wherein the size of the sliding window for performing the sliding window operation is as follows:
wherein x and y are the length and width of the sliding window respectively.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110618991.XA CN113327206B (en) | 2021-06-03 | 2021-06-03 | Image fuzzy processing method of intelligent power transmission line inspection system based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110618991.XA CN113327206B (en) | 2021-06-03 | 2021-06-03 | Image fuzzy processing method of intelligent power transmission line inspection system based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113327206A true CN113327206A (en) | 2021-08-31 |
| CN113327206B CN113327206B (en) | 2022-03-22 |
Family
ID=77419628
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110618991.XA Active CN113327206B (en) | 2021-06-03 | 2021-06-03 | Image fuzzy processing method of intelligent power transmission line inspection system based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113327206B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115049662A (en) * | 2022-08-16 | 2022-09-13 | 山东大拇指喷雾设备有限公司 | Sprayer flow control method based on image processing |
| CN115100304A (en) * | 2022-04-24 | 2022-09-23 | 南通海隆磁业有限公司 | Nuclear magnetic resonance image enhancement method based on image processing |
| CN116630220A (en) * | 2023-07-25 | 2023-08-22 | 江苏美克医学技术有限公司 | Fluorescent image depth-of-field fusion imaging method, device and storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104282003A (en) * | 2014-10-29 | 2015-01-14 | 哈尔滨工业大学 | Digital blurred image blind restoration method based on gradient screening |
| US20150110415A1 (en) * | 2013-10-23 | 2015-04-23 | Adobe Systems Incorporated | Deblurring images having spatially varying blur |
| CN107369134A (en) * | 2017-06-12 | 2017-11-21 | 上海斐讯数据通信技术有限公司 | A kind of image recovery method of blurred picture |
| CN108305230A (en) * | 2018-01-31 | 2018-07-20 | 上海康斐信息技术有限公司 | A kind of blurred picture integrated conduct method and system |
| CN108833770A (en) * | 2018-05-23 | 2018-11-16 | 释码融和（上海）信息科技有限公司 | Image definition calculation method, calculating equipment and focusing system for focusing |
| CN109242790A (en) * | 2018-08-22 | 2019-01-18 | 大连理工大学 | A kind of image deblurring appraisal procedure towards fields of measurement |
| CN111275626A (en) * | 2018-12-05 | 2020-06-12 | 深圳市炜博科技有限公司 | Video deblurring method, device and equipment based on ambiguity |
-
2021
- 2021-06-03 CN CN202110618991.XA patent/CN113327206B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150110415A1 (en) * | 2013-10-23 | 2015-04-23 | Adobe Systems Incorporated | Deblurring images having spatially varying blur |
| CN104282003A (en) * | 2014-10-29 | 2015-01-14 | 哈尔滨工业大学 | Digital blurred image blind restoration method based on gradient screening |
| CN107369134A (en) * | 2017-06-12 | 2017-11-21 | 上海斐讯数据通信技术有限公司 | A kind of image recovery method of blurred picture |
| CN108305230A (en) * | 2018-01-31 | 2018-07-20 | 上海康斐信息技术有限公司 | A kind of blurred picture integrated conduct method and system |
| CN108833770A (en) * | 2018-05-23 | 2018-11-16 | 释码融和（上海）信息科技有限公司 | Image definition calculation method, calculating equipment and focusing system for focusing |
| CN109242790A (en) * | 2018-08-22 | 2019-01-18 | 大连理工大学 | A kind of image deblurring appraisal procedure towards fields of measurement |
| CN111275626A (en) * | 2018-12-05 | 2020-06-12 | 深圳市炜博科技有限公司 | Video deblurring method, device and equipment based on ambiguity |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115100304A (en) * | 2022-04-24 | 2022-09-23 | 南通海隆磁业有限公司 | Nuclear magnetic resonance image enhancement method based on image processing |
| CN115100304B (en) * | 2022-04-24 | 2024-04-19 | 江苏中勤通信科技有限公司 | Nuclear magnetic resonance image enhancement method based on image processing |
| CN115049662A (en) * | 2022-08-16 | 2022-09-13 | 山东大拇指喷雾设备有限公司 | Sprayer flow control method based on image processing |
| CN116630220A (en) * | 2023-07-25 | 2023-08-22 | 江苏美克医学技术有限公司 | Fluorescent image depth-of-field fusion imaging method, device and storage medium |
| CN116630220B (en) * | 2023-07-25 | 2023-11-21 | 江苏美克医学技术有限公司 | Fluorescent image depth-of-field fusion imaging method, device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113327206B (en) | 2022-03-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113327206B (en) | Image fuzzy processing method of intelligent power transmission line inspection system based on artificial intelligence | |
| US10339643B2 (en) | Algorithm and device for image processing | |
| CN107507173B (en) | No-reference definition evaluation method and system for full-slice image | |
| CN107292842B (en) | Image deblurring method based on prior constraint and outlier suppression | |
| CN112734650A (en) | Virtual multi-exposure fusion based uneven illumination image enhancement method | |
| CN110147816B (en) | Method and device for acquiring color depth image and computer storage medium | |
| CN106097256B (en) | A kind of video image fuzziness detection method based on Image Blind deblurring | |
| CN108664839B (en) | Image processing method and device | |
| CN111598918B (en) | Video image stabilizing motion estimation method based on reference frame optimization and foreground and background separation | |
| CN115731146B (en) | Multi-exposure image fusion method based on color gradient histogram feature optical flow estimation | |
| CN110796616A (en) | Fractional order differential operator based L0Norm constraint and adaptive weighted gradient turbulence degradation image recovery method | |
| Srinivas et al. | Remote sensing image segmentation using OTSU algorithm | |
| Feng et al. | Low-light image enhancement algorithm based on an atmospheric physical model | |
| CN115631223A (en) | Multi-view stereo reconstruction method based on self-adaptive learning and aggregation | |
| CN113421210B (en) | Surface point Yun Chong construction method based on binocular stereoscopic vision | |
| US11625886B2 (en) | Storage medium storing program, training method of machine learning model, and image generating apparatus | |
| CN115660994B (en) | Image enhancement method based on regional least square estimation | |
| CN116758021A (en) | Installation standard detection method based on image analysis | |
| CN108364274B (en) | Nondestructive clear reconstruction method of optical image under micro-nano scale | |
| Xie et al. | DHD-Net: A novel deep-learning-based dehazing network | |
| CN115063679A (en) | Pavement quality assessment method based on deep learning | |
| CN108230276B (en) | Natural scene image deblurring method fusing non-parameter image quality evaluation | |
| CN113450380A (en) | Track calibration method based on airport runway scribed lines | |
| Cui et al. | Unsupervised Barcode Image Reconstruction Based on Knowledge Distillation | |
| CN112258434A (en) | Detail-preserving multi-exposure image fusion algorithm in static scene |
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 |