CN112131924A - Transformer substation equipment image identification method based on density cluster analysis - Google Patents
Transformer substation equipment image identification method based on density cluster analysis Download PDFInfo
- Publication number
- CN112131924A CN112131924A CN202010664013.4A CN202010664013A CN112131924A CN 112131924 A CN112131924 A CN 112131924A CN 202010664013 A CN202010664013 A CN 202010664013A CN 112131924 A CN112131924 A CN 112131924A
- Authority
- CN
- China
- Prior art keywords
- image
- density
- gray level
- point
- segmentation
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 52
- 238000007621 cluster analysis Methods 0.000 title claims description 14
- 239000011159 matrix material Substances 0.000 claims abstract description 36
- 230000011218 segmentation Effects 0.000 claims abstract description 29
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 20
- 238000009826 distribution Methods 0.000 claims abstract description 17
- 238000003709 image segmentation Methods 0.000 claims abstract description 12
- 238000000605 extraction Methods 0.000 claims abstract description 11
- 230000007797 corrosion Effects 0.000 claims description 12
- 238000005260 corrosion Methods 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 7
- 230000000877 morphologic effect Effects 0.000 claims description 6
- 238000003672 processing method Methods 0.000 claims description 6
- 230000003628 erosive effect Effects 0.000 claims description 5
- 238000007781 pre-processing Methods 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 abstract description 2
- 238000007689 inspection Methods 0.000 description 8
- 238000011161 development Methods 0.000 description 5
- 230000018109 developmental process Effects 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 230000009466 transformation Effects 0.000 description 4
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 3
- 238000012423 maintenance Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 230000009194 climbing Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000000926 separation method Methods 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 229910018503 SF6 Inorganic materials 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000010339 dilation Effects 0.000 description 1
- 238000007599 discharging Methods 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000005389 magnetism Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000013450 outlier detection Methods 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- SFZCNBIFKDRMGX-UHFFFAOYSA-N sulfur hexafluoride Chemical compound FS(F)(F)(F)(F)F SFZCNBIFKDRMGX-UHFFFAOYSA-N 0.000 description 1
- 229960000909 sulfur hexafluoride Drugs 0.000 description 1
- 238000009827 uniform distribution Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration using local operators
- G06T5/30—Erosion or dilatation, e.g. thinning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/40—Analysis of texture
- G06T7/41—Analysis of texture based on statistical description of texture
- G06T7/45—Analysis of texture based on statistical description of texture using co-occurrence matrix computation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20112—Image segmentation details
- G06T2207/20152—Watershed segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30164—Workpiece; Machine component
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Probability & Statistics with Applications (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Quality & Reliability (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a transformer substation equipment image recognition method based on density clustering analysis, which is characterized in that the independent segmentation of each equipment in a field image containing complex scenes is realized by constructing an equipment image segmentation model based on an improved watershed, then an equipment image feature extraction model based on a gray level co-occurrence matrix is constructed, the brightness distribution condition and the texture feature of the gray level image are extracted, and finally the established model is solved by adopting a density clustering-based algorithm, so that the accuracy of transformer substation equipment image recognition is greatly improved. The problems that images with background noise cannot be processed, the requirement on template selection is high, and segmentation is excessive and the like possibly caused in the identification method in the prior art are solved.
Description
Technical Field
The invention relates to the technical field of image recognition, in particular to a transformer substation equipment image recognition method based on density cluster analysis.
Background
The construction and development of the intelligent power grid are based on informatization and intelligence of power generation, power transmission, power transformation, power distribution, power utilization and scheduling. The transformer substation is an important node in a power grid, and is connected with power grids at all levels through a transformer, so that the continuous development of the intellectualization of the transformer substation can provide a solid foundation for the safe operation of the intelligent power grid. The intelligent development trend of the transformer substation is that no or little person is on duty. The method is influenced by dangerous working areas, severe weather conditions and long time for routing inspection, and the manual routing inspection has large workload and low efficiency, thereby becoming a negative factor restricting the stable operation and production of the transformer substation equipment. The industrial robot has strong adaptability to severe environments, gradually replaces the traditional manual inspection, and provides reliable field visual data for the links of running state diagnosis and intelligent maintenance of substation equipment. In the face of a large amount of image information returned by the inspection robot, an intelligent image processing technology is needed to mine the image characteristics of target equipment, each piece of equipment in a field image containing a complex scene is divided independently, then the divided images are identified, the equipment type of each image is determined, whether the equipment has the phenomena of discharging, oil leakage and damage in the operation process is judged according to the image identification result, and the equipment state is correctly evaluated. Therefore, the accurate image segmentation and identification algorithm has important significance for monitoring the state of the substation equipment.
At present, research aiming at the image recognition method of the transformer substation equipment at home and abroad has a certain foundation. The edge detection method based on the differential operator utilizes the characteristic that the gray gradient of a target object in an image changes violently at the edge to carry out differential operation on image pixels to obtain the edge of the target, but the differential operator has unsatisfactory effect in processing the image with background noise; the identification method based on the template makes a corresponding template according to the shape and contour characteristics of the device image, matches the target device in the original image with the known template to obtain the final classification identification result, but the method has higher requirement on the selection of the template, and the identification method based on the template can not meet the actual requirement due to the situations of affine transformation and mutual shielding of the field image target; the region growing-based method selects some representative pixels in the segmentation region as the starting points of the algorithm, and combines pixels with the same or similar characteristics with the starting points together around the starting points to form a region until all the pixels are combined, but a group of representative starting points is difficult to determine, and the segmentation is excessive due to the excessive number of the starting points.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a transformer substation equipment image identification method based on density cluster analysis, which aims to solve the problems that the traditional method is difficult to accurately identify the transformer substation equipment image, and the image transmitted back by a transformer substation inspection robot is complex in scene and difficult to identify, realize accurate segmentation and identification of the transformer substation equipment image, monitor the running state of equipment in a transformer substation in real time, prevent transformer substation faults and facilitate the realization of intelligent development of the transformer substation, thereby achieving the effects of improving the operation and maintenance efficiency of the transformer substation, improving the running safety of a power system and improving the power supply reliability of a power distribution network.
In order to achieve the purpose, the invention provides a transformer substation equipment image identification method based on density cluster analysis, which comprises the following steps:
(1) constructing an improved watershed-based device image segmentation model, and realizing independent segmentation of each device in a scene image containing a complex scene;
(2) constructing an equipment image feature extraction model based on the gray level co-occurrence matrix, and extracting the brightness distribution condition and the texture feature of the gray level image;
(3) and solving the established model by adopting an algorithm based on density clustering.
Further, the device image segmentation model based on improved watershed in step (1) is specifically: firstly, preprocessing an image by using a threshold segmentation method, and then eliminating tiny holes and fine lines in the image by using a morphological processing method.
Further, the preprocessing the image by using the threshold segmentation method specifically includes:
calculating the optimal threshold value in the statistical sense by utilizing the gray level histogram of the image to obtain a binary image:
in the formula, μ 1 and μ 2 are respectively the gray average value of the two divided pixel sets, t is an initial threshold, ni is the total number of pixels with gray value i, and L is the gray level of the whole image;
order to
The between-class variance σ2(t) is expressed as:
the maximum inter-class variance is a segmentation threshold, each minimum value point in the original image is marked by using a watershed algorithm, and when the distance from a certain minimum value to a watershed is smaller than a set threshold, a mark corresponding to the minimum value is eliminated.
Further, the removing of the fine holes and fine lines in the image by using the morphological processing method specifically comprises:
the corrosion operation is adopted to eliminate or weaken the boundary points of the object, and is expressed as follows:
wherein E is an image generated by erosion operation, A is a target area to be processed on a plane, and S (x, y) is an area represented by a structural element on a plane coordinate;
the boundary points of the object are expanded or enhanced by adopting an expansion operation, and the expression is as follows:
in the formula, D is an image generated by expansion operation;
meanwhile, boundary points of the object are processed by corrosion operation and expansion operation, the original image is processed by opening operation and closing operation, the opening operation and the closing operation are combination of corrosion and expansion, the opening operation is used for carrying out corrosion operation before expansion operation on the original image, the boundary of the object can be smoothly processed when the area is divided, small isolated points are eliminated, the closing operation is used for carrying out corrosion operation after expansion operation on the original image, cavities appearing in the object can be filled, and adjacent areas are connected.
Further, in the device image feature extraction model based on the gray level co-occurrence matrix in step (2), the calculation process of the gray level co-occurrence matrix is as follows: the point A (x, y) is any point in the N-level gray scale image, and the point B (x + delta x, y + delta y) is obtained after the deviation of the position deviation operator (delta x, delta y), wherein the brightness values of the point A and the point B are g respectively1And g2A, B the combined brightness of the two points (g)1,g2) At all probabilities P (g) of each possible value1,g2) The formed N multiplied by N matrix is the gray level co-occurrence matrix of the gray level image.
Further, the following statistics based on the gray level co-occurrence matrix are used to describe the texture features of the image: angular second moment, moment of inertia, correlation, entropy, and inverse difference.
Further, the solving method of the density cluster-based algorithm in the step (3) further includes the steps of:
1) calculating the density of all points in the data set according to a density estimation function;
2) judging density attraction points according to a set threshold value;
3) selecting a point from the data set, judging whether the point is a density attraction point, if so, performing the step 5); if not, performing step 4);
4) searching the density attraction point clustered by the point;
5) judging whether all the points are processed, and if not, performing the step 3); if yes, stopping calculation and outputting the result.
In summary, according to the transformer substation equipment image identification method based on density cluster analysis provided by the invention, the separation independence of each equipment in a field image containing complex scenes is realized by constructing an equipment image segmentation model based on an improved watershed, then an equipment image feature extraction model based on a gray level co-occurrence matrix is constructed, the brightness distribution condition and the texture feature of the gray level image are extracted, and finally the established model is solved by adopting a density cluster-based algorithm, so that the accuracy of transformer substation equipment image identification is greatly improved. The problems that images with background noise cannot be processed, the requirement on template selection is high, and segmentation is excessive and the like possibly caused in the identification method in the prior art are solved.
The technical scheme of the invention has the following beneficial technical effects: the method provided by the invention can realize accurate segmentation and identification of the transformer substation equipment image, solves the problems that the image transmitted back by the transformer substation inspection robot is complex in scene and difficult to identify, and enables a power company to monitor the running state of equipment in the transformer substation in real time, so that transformer substation faults are prevented, the intelligent development of the transformer substation is facilitated, the operation and maintenance efficiency of the transformer substation is improved, the running safety of a power system is improved, and the power supply reliability of a power distribution network is improved.
Drawings
FIGS. 1 (a) - (c) are gray scale images of images comprising the insulating sleeve and the densitometer taken at different times (t1-t 3); (d) a gray scale image of an image shot when a shielding object is arranged on the sleeve at the time t 4; (e) a grayscale map for a single device (densitometer) image at time t 5; (f) the gray scale image of the single device (bushing) at time t 6.
In fig. 2, (a) - (c) are respectively (t1-t3) time scene images, and target images of the densitometer and the insulating sleeve at the time are segmented by the improved watershed transform, wherein the left image is a densitometer image, and the right image is an insulating sleeve image; (d) the segmentation result at the time t4 is shown, wherein the left image is a densitometer image, the middle image is an image with a shielding insulating sleeve, and the right image is a shielding object image; (e) the segmentation result of the densitometer at the time t 5; (f) which is the result of the segmentation of the bushing at time t 6.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings in conjunction with the following detailed description. It should be understood that the description is intended to be exemplary only, and is not intended to limit the scope of the present invention. Moreover, in the following description, descriptions of well-known structures and techniques are omitted so as to not unnecessarily obscure the concepts of the present invention.
The method aims to solve the problems that in the image recognition method of the substation equipment in the prior art, images with noisy backgrounds cannot be processed, the requirement on template selection is high, and excessive segmentation is possibly caused. The invention provides a transformer substation equipment image recognition method based on density clustering analysis, which is characterized in that the independent segmentation of each equipment in a field image containing complex scenes is realized by constructing an equipment image segmentation model based on an improved watershed, then an equipment image feature extraction model based on a gray level co-occurrence matrix is constructed, the brightness distribution condition and the texture feature of the gray level image are extracted, and finally the established model is solved by adopting a density clustering-based algorithm, so that the accuracy of transformer substation equipment image recognition is greatly improved.
The technical scheme of the invention is explained in detail below, and the method for evaluating the residual magnetism of the transformer based on the particle swarm optimization comprises the following steps:
(1) constructing an improved watershed-based device image segmentation model, and realizing independent segmentation of each device in a scene image containing a complex scene; the method comprises the following steps:
the traditional watershed algorithm takes the gradient of an image as an input, and the gradient image solving formula is as follows:
g(x,y)＝{[f(x,y)-f(x-1,y)]2+[f(x,y)-f(x,y-1)]2}0.5
in the formula, f (x, y) is an input image, g (x, y) is an image output after gradient operation, and x and y are respectively the abscissa and the ordinate of the image.
The traditional watershed algorithm is easy to excessively segment because each target comprises a plurality of minimum value points due to the influence of texture, light and noise factors in an original image, so that local extreme values are excessive, and excessive watersheds are formed during image segmentation. Aiming at the problem of excessive segmentation of the traditional watershed algorithm, the method firstly utilizes a threshold segmentation method to preprocess the image and then utilizes a morphological processing method to eliminate tiny holes and fine lines in the image.
Calculating the optimal threshold value in the statistical sense by utilizing the gray level histogram of the image to obtain a binary image:
in the formula, μ 1 and μ 2 are respectively the gray average value of the two divided pixel sets, t is an initial threshold, ni is the total number of pixels with gray value i, and L is the gray level of the whole image;
order to
The between-class variance σ2(t) is expressed as:
the maximum between-class variance is the segmentation threshold. And marking each minimum value point in the original image by using a watershed algorithm, and eliminating the mark corresponding to a certain minimum value when the distance from the certain minimum value to the watershed is less than a set threshold value.
The morphological processing method comprises corrosion operation, expansion operation, opening operation and closing operation. The erosion algorithm can be used to eliminate or weaken the boundary points of an object, expressed as:
wherein E is an image generated by erosion operation, A is a target area to be processed on a plane, and S (x, y) is an area represented by a structural element on a plane coordinate;
the dilation operation is the opposite of the erosion operation, and its effect is to expand or enhance the boundary points of the object, expressed as:
in the formula, D is an image generated by expansion operation;
the meaning of the open operation is that firstly the image is corroded and then the corroded image is expanded; the closed operation means that the image is firstly subjected to expansion processing, and then the image subjected to expansion processing is subjected to corrosion processing.
(2) Constructing an equipment image feature extraction model based on the gray level co-occurrence matrix, and extracting the brightness distribution condition and the texture feature of the gray level image; the method comprises the following steps:
the gray level co-occurrence matrix is adopted to reflect the brightness distribution condition of a gray level image, the position distribution condition between two pixels with the same or similar brightness is described, and the joint probability density of the two pixels is utilized to express the second-order statistic of the brightness change. The calculation process of the gray level co-occurrence matrix is as follows: the point A (x, y) is any point in the N-level gray scale image, and the point B (x + delta x, y + delta y) is obtained after the deviation of the position deviation operator (delta x, delta y), wherein the brightness values of the point A and the point B are g respectively1And g2A, B the combined brightness of the two points (g)1,g2) At all probabilities P (g) of each possible value1,g2) The formed N × N matrix is the grayAnd (5) gray level co-occurrence matrix of the degree image.
The gray level co-occurrence matrix only represents the joint brightness distribution condition between every two elements in the image and cannot directly represent the characteristics of the image texture. Therefore, the following statistics based on the gray level co-occurrence matrix are used to describe the texture features of the image:
1) second moment of angle
The angle second moment represents the energy of the gray level co-occurrence matrix, and the larger the energy value is, the more concentrated the distribution of the elements with larger numerical values in the matrix is, and the texture of the image is changed regularly; with a small energy value, the texture in the image is generally fine. The angular second moment ASM is expressed as:
wherein N is the gray scale order of the image, and P (i, j) is the element in the gray scale co-occurrence matrix;
2) moment of inertia
The moment of inertia represents the contrast in the image, and a larger value indicates a larger contrast in the image, and the deeper the texture, the sharper the image. The larger the distance between the larger value element in the gray level co-occurrence matrix and the diagonal line, the larger the inertia moment value. The moment of inertia CON is expressed as:
3) correlation
The correlation represents the degree of similarity of local gray levels in an image, and the correlation in the row or column direction of elements is reflected in the gray level co-occurrence matrix. The larger the correlation value is, the more uniform the value distribution of the elements in the matrix is, and if the horizontal distribution texture in the image is more, the correlation value of the horizontal direction gray level co-occurrence matrix is larger than that of other directions. The correlation COR is expressed as:
in the formula, mux、μy、
For intermediate variables, the following is stated:
4) entropy of the entropy
Entropy reflects the amount of information contained in the image. When the element values of the gray level co-occurrence matrix are uniformly distributed, the entropy value is large, and the texture distribution randomness in the image is large, and the complexity is high. The entropy H is expressed as:
5) inverse difference
The inverse difference reflects the change of texture characteristics in the image, and the larger the inverse difference value is, the stronger the texture homogeneity and the uniform distribution in the image are. The inverse gap IDM is expressed as:
(3) solving the established model by adopting density clustering-based algorithm
Implementing density of data objects using gaussian kernel functionsAnd estimating, determining a clustering center at the local maximum value point of the density, and judging the final category according to the similarity of the object and the clustering center. Let x1,x2,…,xnFor an object in the data set D, a density estimation function for any point in the data set
The expression is as follows:
wherein, | x-xiL is the distance between two objects, c is the width parameter of the density estimation function;
calculating the density of all points of the data set according to the density estimation function if the density estimation function satisfies the requirements
And xi is a set threshold value used for discriminating whether the low-density maximum value is noise or not, and the maximum value point is used as the center of a cluster and is also called a density attraction point. For sample objects x except the density attraction points, with the gradient of the density estimation function of the sample objects x as guidance, performing hill climbing search to find the density attraction point closest to the sample objects x, wherein the expression is as follows:
where k is the number of searches, a convergence parameter,
estimating a gradient of the function for the density;
if it satisfies
The hill climbing process is terminated and point x is clustered to a density attraction point xk. If it is
Object x is an outlier.
The algorithm calculation flow based on density clustering is as follows:
1) calculating the density of all points in the data set according to a density estimation function;
2) judging density attraction points according to a set threshold value;
3) selecting a point from the data set, judging whether the point is a density attraction point, if so, performing the step 5); if not, performing step 4);
4) searching the density attraction point clustered by the point;
5) judging whether all the points are processed, and if not, performing the step 3); if yes, stopping calculation and outputting the result.
The following further describes the result of image recognition on the substation equipment by using the image recognition method with reference to the accompanying drawings.
The experimental data adopted by the invention come from images shot by a certain 500kV substation inspection robot on the field of a densimeter and an insulating sleeve of a 500kV high-voltage sulfur hexafluoride circuit breaker at different moments, the gray level images of the images collected on the field at t1-t6 are shown in figure 1, and the phenomena of affine transformation, mutual shielding and foreign matter coverage exist in the target objects of the images on the field at different moments. FIGS. 1 (a) - (c) are gray scale images of images comprising the insulating sleeve and the densitometer taken at different times (t1-t 3); (d) a gray scale image of an image shot when a shielding object is arranged on the sleeve at the time t 4; (e) a grayscale map for a single device (densitometer) image at time t 5; (f) the gray scale image of the single device (bushing) at time t 6.
According to fig. 1, the image is segmented by using the improved watershed-based device image segmentation model of the invention, and the segmentation result is shown in fig. 2. FIGS. 2(a) - (c) are respectively a field image at time t1-t3 and a target image of a densitometer and an insulating sleeve at the time after the field image is divided by an improved watershed transform, wherein the left image is a densitometer image, and the right image is an insulating sleeve image; (d) the segmentation result at the time t4 is shown, wherein the left image is a densitometer image, the middle image is an image with a shielding insulating sleeve, and the right image is a shielding object image; (e) the segmentation result of the densitometer at the time t 5; (f) which is the result of the segmentation of the bushing at time t 6.
The equipment image feature extraction model based on the gray level co-occurrence matrix processes the image in the graph 2, and the statistical feature quantity of the obtained gray level co-occurrence matrix is shown in the table 1.
TABLE 1 Gray level co-occurrence matrix statistical feature quantity
The algorithm based on density clustering processes the data in table 1, wherein the convergence parameter is 0.7, the density attraction point threshold xi is 0.5, and c is 0.5. The density attraction points are calculated to be at a densitometer at the time t5 and at a t2 insulating sleeve, returned clusters are C1 ═ {1,3,5,7,10}, C2 ═ 2,4,6,8,11}, and C3 ═ 9}, wherein the numerical elements in the set are the device target orders in Table 1, and the elements in the C3 set do not satisfy the requirements of the device target orders in Table 1
Therefore, the method is judged to be the outlier, is consistent with the actual situation, and verifies that the method can effectively identify the equipment target when the inspection robot shoots the image and is shielded by the abnormal object due to affine transformation, can realize outlier detection of the abnormal object, and can position the abnormal state of equipment operation.
In conclusion, the invention provides a transformer substation equipment image identification method based on density cluster analysis, which is characterized in that the separation independence of each equipment in a field image containing complex scenes is realized by constructing an equipment image segmentation model based on an improved watershed, then an equipment image feature extraction model based on a gray level co-occurrence matrix is constructed, the brightness distribution condition and the texture feature of the gray level image are extracted, and finally the established model is solved by adopting an algorithm based on density cluster; in the step of adopting the equipment image feature extraction model based on the gray level co-occurrence matrix, the texture features of the image are described by adopting the statistics of the angle second moment, the moment of inertia, the correlation, the entropy, the inverse difference distance and the like based on the gray level co-occurrence matrix, so that the accuracy of the image recognition of the substation equipment is greatly improved. The problems that images with background noise cannot be processed, the requirement on template selection is high, and segmentation is excessive and the like possibly caused in the identification method in the prior art are solved.
It is to be understood that the above-described embodiments of the present invention are merely illustrative of or explaining the principles of the invention and are not to be construed as limiting the invention. Therefore, any modification, equivalent replacement, improvement and the like made without departing from the spirit and scope of the present invention should be included in the protection scope of the present invention. Further, it is intended that the appended claims cover all such variations and modifications as fall within the scope and boundaries of the appended claims or the equivalents of such scope and boundaries.
Claims (7)
1. A transformer substation equipment image identification method based on density cluster analysis is characterized by comprising the following steps:
(1) constructing an improved watershed-based device image segmentation model, and realizing independent segmentation of each device in a scene image containing a complex scene;
(2) constructing an equipment image feature extraction model based on the gray level co-occurrence matrix, and extracting the brightness distribution condition and the texture feature of the gray level image;
(3) and solving the established model by adopting an algorithm based on density clustering.
2. The transformer substation equipment image recognition method based on density cluster analysis according to claim 1, wherein the device image segmentation model based on improved watershed in the step (1) is specifically: firstly, preprocessing an image by using a threshold segmentation method, and then eliminating holes and fine lines in the image by using a morphological processing method.
3. The transformer substation equipment image recognition method based on density cluster analysis according to claim 2, wherein the preprocessing of the image by using the threshold segmentation method specifically comprises:
calculating the optimal threshold value in the statistical sense by utilizing the gray level histogram of the image to obtain a binary image:
in the formula, μ 1 and μ 2 are respectively the gray average value of the two divided pixel sets, t is an initial threshold, ni is the total number of pixels with gray value i, and L is the gray level of the whole image;
order to
The between-class variance σ2(t) is expressed as:
the maximum inter-class variance is a segmentation threshold, each minimum value point in the original image is marked by using a watershed algorithm, and when the distance from a certain minimum value to a watershed is smaller than a set threshold, a mark corresponding to the minimum value is eliminated.
4. The transformer substation equipment image recognition method based on density cluster analysis according to claim 3, wherein the removing of the fine holes and fine lines in the image by using the morphological processing method specifically comprises:
the boundary points of the object are eliminated or weakened by corrosion operation, and are expressed as:
wherein E is an image generated by erosion operation, A is a target area to be processed on a plane, and S (x, y) is an area represented by a structural element on a plane coordinate;
the boundary points of the object are expanded or enhanced by adopting an expansion operation, and the expression is as follows:
in the formula, D is an image generated by expansion operation;
meanwhile, boundary points of the object are processed by corrosion operation and expansion operation, the original image is processed by opening operation and closing operation, the opening operation and the closing operation are combination of corrosion and expansion, the opening operation is used for carrying out corrosion operation before expansion operation on the original image, when the area is divided, the boundary of the object is smoothly processed, small isolated points are eliminated, the closing operation is used for carrying out corrosion operation after expansion operation on the original image, cavities appearing in the object are filled, and adjacent areas are connected.
5. The substation equipment image recognition method based on density cluster analysis according to claim 1, wherein in the equipment image feature extraction model based on the gray level co-occurrence matrix in step (2), the calculation process of the gray level co-occurrence matrix is as follows: the point A (x, y) is any point in the N-level gray scale image, and the point B (x + delta x, y + delta y) is obtained after the deviation of the position deviation operator (delta x, delta y), wherein the brightness values of the point A and the point B are g respectively1And g2A, B the combined brightness of the two points (g)1,g2) At all probabilities P (g) of each possible value1,g2) The formed N multiplied by N matrix is the gray level co-occurrence matrix of the gray level image.
6. The substation equipment image identification method based on density cluster analysis according to claim 5, characterized in that the following statistics based on gray level co-occurrence matrix are used to describe the texture features of the image: angular second moment, moment of inertia, correlation, entropy, and inverse difference.
7. The substation equipment image identification method based on density cluster analysis according to claim 1, wherein the solving method of the density cluster based algorithm in the step (3) further comprises the steps of:
1) calculating the density of all points in the data set according to a density estimation function;
2) judging density attraction points according to a set threshold value;
3) selecting a point from the data set, judging whether the point is a density attraction point, if so, performing the step 5); if not, performing step 4);
4) searching the density attraction point clustered by the point;
5) judging whether all the points are processed, and if not, performing the step 3); if yes, stopping calculation and outputting the result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010664013.4A CN112131924A (en) | 2020-07-10 | 2020-07-10 | Transformer substation equipment image identification method based on density cluster analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010664013.4A CN112131924A (en) | 2020-07-10 | 2020-07-10 | Transformer substation equipment image identification method based on density cluster analysis |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112131924A true CN112131924A (en) | 2020-12-25 |
Family
ID=73850227
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010664013.4A Pending CN112131924A (en) | 2020-07-10 | 2020-07-10 | Transformer substation equipment image identification method based on density cluster analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112131924A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112861985A (en) * | 2021-02-24 | 2021-05-28 | 郑州轻工业大学 | Automatic book classification method based on artificial intelligence |
| CN112989940A (en) * | 2021-02-08 | 2021-06-18 | 国家海洋环境监测中心 | Raft culture area extraction method based on high-resolution three-satellite SAR image |
| CN113177614A (en) * | 2021-05-27 | 2021-07-27 | 广州白云电器设备股份有限公司 | Image recognition system and method for power supply switch cabinet of urban rail transit |
| CN113343917A (en) * | 2021-06-30 | 2021-09-03 | 上海申瑞继保电气有限公司 | Histogram-based substation equipment identification method |
| CN115861320A (en) * | 2023-02-28 | 2023-03-28 | 天津中德应用技术大学 | Intelligent detection method for automobile part machining information |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107292882A (en) * | 2017-08-02 | 2017-10-24 | 国网电力科学研究院武汉南瑞有限责任公司 | One kind is based on the adaptive electrical equipment malfunction detection methods of Meanshift |
| CN110648307A (en) * | 2019-07-24 | 2020-01-03 | 国网浙江省电力有限公司湖州供电公司 | Method for checking state of transformer substation pressure plate by using image comparison technology |
-
2020
- 2020-07-10 CN CN202010664013.4A patent/CN112131924A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107292882A (en) * | 2017-08-02 | 2017-10-24 | 国网电力科学研究院武汉南瑞有限责任公司 | One kind is based on the adaptive electrical equipment malfunction detection methods of Meanshift |
| CN110648307A (en) * | 2019-07-24 | 2020-01-03 | 国网浙江省电力有限公司湖州供电公司 | Method for checking state of transformer substation pressure plate by using image comparison technology |
Non-Patent Citations (1)
| Title |
|---|
| 崔巨勇: "智能变电站信息一体化平台数据挖掘技术研究", 《工程科技Ⅱ辑》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112989940A (en) * | 2021-02-08 | 2021-06-18 | 国家海洋环境监测中心 | Raft culture area extraction method based on high-resolution three-satellite SAR image |
| CN112989940B (en) * | 2021-02-08 | 2023-08-01 | 国家海洋环境监测中心 | Raft culture area extraction method based on high-resolution third satellite SAR image |
| CN112861985A (en) * | 2021-02-24 | 2021-05-28 | 郑州轻工业大学 | Automatic book classification method based on artificial intelligence |
| CN112861985B (en) * | 2021-02-24 | 2023-01-31 | 郑州轻工业大学 | Automatic book classification method based on artificial intelligence |
| CN113177614A (en) * | 2021-05-27 | 2021-07-27 | 广州白云电器设备股份有限公司 | Image recognition system and method for power supply switch cabinet of urban rail transit |
| CN113343917A (en) * | 2021-06-30 | 2021-09-03 | 上海申瑞继保电气有限公司 | Histogram-based substation equipment identification method |
| CN113343917B (en) * | 2021-06-30 | 2024-05-31 | 上海申瑞继保电气有限公司 | Substation equipment identification method based on histogram |
| CN115861320A (en) * | 2023-02-28 | 2023-03-28 | 天津中德应用技术大学 | Intelligent detection method for automobile part machining information |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112131924A (en) | Transformer substation equipment image identification method based on density cluster analysis | |
| CN109063723B (en) | Weak supervision image semantic segmentation method based on common features of iteratively mined objects | |
| CN112199993B (en) | Method for identifying transformer substation insulator infrared image detection model in any direction based on artificial intelligence | |
| CN107944396B (en) | Knife switch state identification method based on improved deep learning | |
| CN109118479B (en) | Capsule network-based insulator defect identification and positioning device and method | |
| CN106407928B (en) | Transformer composite insulator casing monitoring method and system based on raindrop identification | |
| CN109389180A (en) | A power equipment image-recognizing method and inspection robot based on deep learning | |
| Wang et al. | Recognition and drop-off detection of insulator based on aerial image | |
| CN108334824B (en) | High-voltage isolating switch state identification method based on background difference and iterative search | |
| CN112417937B (en) | Substation video target detection method based on time sequence | |
| CN113888462A (en) | Crack identification method, system, readable medium and storage medium | |
| Zhang et al. | Zero-DD: Zero-sample defect detection for industrial products | |
| Zhang et al. | Defect detection of anti-vibration hammer based on improved faster R-CNN | |
| Niu et al. | Electrical equipment identification method with synthetic data using edge-oriented generative adversarial network | |
| CN117593499A (en) | Hydropower station electromechanical equipment fault identification method based on distributed routing inspection strategy | |
| CN116630989A (en) | Visual fault detection method and system for intelligent ammeter, electronic equipment and storage medium | |
| Jianfeng et al. | IMPLEMENTATION OF THE DUAL-BODY INTELLIGENT INSPECTION ROBOT IN SUBSTATION BASED ON DATA MINING ALGORITHM. | |
| Lu et al. | Anomaly detection method for substation equipment based on feature matching and multi-Semantic classification | |
| CN115311625A (en) | Monitoring method for judging whether target contacts power transmission line | |
| Hao et al. | Based on Surf feature extraction and insulator damage identification for capsule networks | |
| Zhu et al. | Moving object real-time detection and tracking method based on improved Gaussian mixture model | |
| CN113496159B (en) | Multi-scale convolution and dynamic weight cost function smoke target segmentation method | |
| CN111079750B (en) | Power equipment fault region extraction method based on local region clustering | |
| CN112785613A (en) | Intelligent hearth flame image identification method | |
| Zhang et al. | Improving class activation map for weakly supervised object localization |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201225 |
|
| RJ01 | Rejection of invention patent application after publication |