CN108665468A - A kind of device and method extracting tangent tower insulator chain based on dimensionality reduction gray scale texture-form matrix - Google Patents
A kind of device and method extracting tangent tower insulator chain based on dimensionality reduction gray scale texture-form matrix Download PDFInfo
- Publication number
- CN108665468A CN108665468A CN201711105128.4A CN201711105128A CN108665468A CN 108665468 A CN108665468 A CN 108665468A CN 201711105128 A CN201711105128 A CN 201711105128A CN 108665468 A CN108665468 A CN 108665468A
- Authority
- CN
- China
- Prior art keywords
- image
- value
- insulator chain
- tangent tower
- gray
- 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
- 239000012212 insulator Substances 0.000 title claims abstract description 92
- 239000011159 matrix material Substances 0.000 title claims abstract description 45
- 230000009467 reduction Effects 0.000 title claims abstract description 33
- 238000000034 method Methods 0.000 title claims abstract description 22
- 238000012545 processing Methods 0.000 claims abstract description 25
- 238000003708 edge detection Methods 0.000 claims abstract description 22
- 230000006870 function Effects 0.000 claims description 9
- 238000001914 filtration Methods 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 7
- 238000001514 detection method Methods 0.000 claims description 6
- 238000004458 analytical method Methods 0.000 claims description 5
- 230000005540 biological transmission Effects 0.000 claims description 5
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 claims description 4
- 238000004891 communication Methods 0.000 claims description 4
- 238000013461 design Methods 0.000 claims description 4
- 238000009413 insulation Methods 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 4
- 239000002131 composite material Substances 0.000 claims description 3
- 239000011521 glass Substances 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 claims description 3
- 238000009825 accumulation Methods 0.000 claims description 2
- 230000003044 adaptive effect Effects 0.000 claims description 2
- 230000004323 axial length Effects 0.000 claims description 2
- 238000000205 computational method Methods 0.000 claims description 2
- 230000004069 differentiation Effects 0.000 claims description 2
- 230000009977 dual effect Effects 0.000 claims description 2
- 230000000717 retained effect Effects 0.000 claims description 2
- 230000001629 suppression Effects 0.000 claims description 2
- 241000238097 Callinectes sapidus Species 0.000 claims 1
- 230000007547 defect Effects 0.000 abstract description 4
- 239000000284 extract Substances 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 9
- 238000000605 extraction Methods 0.000 description 2
- 241001212149 Cathetus Species 0.000 description 1
- 230000009172 bursting Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration by the use of histogram techniques
-
- G06T5/70—
-
- 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
- G06T7/001—Industrial image inspection using an image reference approach
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- 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
- G06T2207/20024—Filtering details
- G06T2207/20032—Median filtering
Abstract
A kind of device and method being extracted tangent tower insulator chain based on dimensionality reduction gray scale textural shape matrix, is comprised the following modules：Sample memory module, the image for storing various model sub-pieces；Unmanned plane module, for obtaining the tangent tower insulator chain coloured image on extra high voltage network, edge detection module generates first processing picture for carrying out edge detection to pretreated image；Aerial Images segment module, calculate every small picture in energy, contrast, correlation, the value of the uniformity；Contrast module is communicated, the actual value of the reference value and unmanned plane in the sample database of ground is compared under respective angles；Insulator chain locating module, the tangent tower insulator chain image finally extracted.The present invention can accurately extract the tangent tower insulator chain on extra high voltage network in unmanned plane image, to judge whether tangent tower insulator chain is complete and identification insulator disk lays technical foundation with the presence or absence of defect.
Description
Technical field
The present invention relates to the technical fields of unmanned plane, and in particular to a kind of one kind applied in power industry is based on dimensionality reduction
The device and method of gray scale texture-form matrix extraction tangent tower insulator chain.
Background technology
In the extra high voltage network in China is safeguarded, bad insulator chain performance is the major reason for occurring accident, because
Insulator chain fracture causes accident to become the maximum disaster of electric system at present.Insulator chain on grid equipment is due to long-term
Phenomena such as exposure in a natural environment, is susceptible to crack after service phase length, pollution flashover is with bursting, this can seriously affect electric system
Normal operation.
At present both at home and abroad the defects detection of the insulator chain on high voltage power transmission and transforming overhead line structures mainly by acquiring image after,
Using traditional artificial macroscopic method.There is artificial detection very strong subjectivity, verification and measurement ratio can also vary with each individual, shadow
The efficiency of defect recognition is rung.If grasping the operating status of insulator chain in being on active service in time, it will greatly reduce or avoid electricity
Force system breaks down.The method of traditional inspection insulator chain operating status be periodical power failure or electrification artificial detection, these
Detection task not only needs manpower working at height, and high-altitude present in working at height, high pressure, high temperature current conditions be to need
It requires efforts consideration.
Therefore, Maintenance of Electric Transmission Line replaces manual patrol using unmanned plane, using Digital Image Processing and pattern-recognition
Device extracts insulator chain image, is modernization, an urgent demand of intelligent power network.
Invention content
Technical problem to be solved by the invention is to provide one kind extracting tangent tower based on dimensionality reduction gray scale texture-form matrix
The device and method of insulator chain can make unmanned plane in patrolling power transmission lines, be taken photo by plane to one by its included camera
Then the insulator chain image of series of straight lines tower is positioned by dimensionality reduction gray scale texture-form matrix, extracts image cathetus tower
Insulator chain image, in next step judge that the quality of the insulator chain performance of tangent tower lays technical foundation.
The technical solution that the present invention takes is：
A kind of device being extracted tangent tower insulator chain based on dimensionality reduction gray scale texture-form matrix, is comprised the following modules：
Sample memory module, the image for storing various model sub-pieces, every image have big in first quartile
Small elliptical edge point line 0 °, 30 °, 45 °, 60 °, 90 ° of lower energy, contrast, correlation, the value of the uniformity, and distribute four
The certain weights of a characteristic value generate a reference value；
Unmanned plane module for obtaining the tangent tower insulator chain coloured image on extra high voltage network, and has
Standby preprocessing function；
Edge detection module, for carrying out edge detection to pretreated image, if the edge line that detected is even
It is connected into the then reservation of a rectangle, the position mark of edge line, and is shown in pretreated image, removal is unlabelled
Pattern generates first processing picture；
Aerial Images segment module, equally that every small picture is same first processing picture segmentation at one piece of small picture of block
In sample database equally with two it is oval handle, calculate every small picture energy under 0 °, 30 °, 45 °, 60 °, 90 ° of first quartile
Amount, contrast, correlation, the value of the uniformity, and distribute four characteristic values certain weights, obtain 0 ° of first quartile, 30 °,
Each actual value under 45 °, 60 °, 90 °；
Contrast module is communicated, the actual value of the reference value and unmanned plane in the sample database of ground is compared under respective angles,
If the two gap is in certain threshold range, then it is assumed that the physical location for having found insulator chain in Aerial Images, the position
Label；
The position of label is generated a width bianry image and is obtained with originally secondary processing image dot product by insulator chain locating module
To the tangent tower insulator chain image finally extracted.
A kind of device and method extracting tangent tower insulator chain based on dimensionality reduction gray scale texture-form matrix of the present invention, energy
Tangent tower insulator chain in accurate extraction unmanned plane image on extra high voltage network, to judge tangent tower insulator chain
Whether complete and identification insulator disk lays technical foundation with the presence or absence of defect.
Description of the drawings
Fig. 1 is the module relation diagram of the present invention.
Wherein：Sample memory module 1 in Fig. 1, unmanned plane module 2, edge detection module 3, Aerial Images segment mould
Block 4 communicates contrast module 5, insulator chain locating module 6, unmanned airborne device 7, ground end device 8.
Fig. 2 is the target line tower insulator chain structural schematic diagram of the present invention.
Wherein：Tangent line L in Fig. 2, insulator chain ontology 9, shaft tower component 10.
Fig. 3 is the sample sub-pieces image procossing schematic diagram of the present invention.
Fig. 4 is the insulator chain image procossing schematic diagram of taking photo by plane of the present invention.
Wherein：O is coordinate origin in Fig. 3,4, and m, n are the length of side of Aerial Images, and w is grown at the interval between size ellipse.
Sample sub-pieces edge 11, sample graph 12 is small by oval 13, big ellipse 14, actual insulation substring 15, m*n net
Lattice 16, Aerial Images 17.
Fig. 5 is the dimensionality reduction gray scale texture-schematic diagram of form matrix at 0.
Fig. 6 is dimensionality reduction gray scale texture-schematic diagram of the form matrix under 30 °.
Fig. 7 is dimensionality reduction gray scale texture-schematic diagram of the form matrix under 45 °.
Fig. 8 is the dimensionality reduction gray scale texture-schematic diagram of form matrix at 60 deg..
Fig. 9 is dimensionality reduction gray scale texture-schematic diagram of the form matrix under 90 °.
Specific implementation mode
A kind of device being extracted tangent tower insulator chain based on dimensionality reduction gray scale texture-form matrix, is comprised the following modules, such as
Shown in Fig. 1.
Sample memory module, the image for storing various model sub-pieces, every image have big in first quartile
Small elliptical edge point line 0 °, 30 °, 45 °, 60 °, 90 ° of lower energy, contrast, correlation, the value of the uniformity, and distribute four
The certain weights of a characteristic value generate a reference value.
Unmanned plane module for obtaining the tangent tower insulator chain coloured image on extra high voltage network, and has
Standby preprocessing function.
Edge detection module, for carrying out edge detection to pretreated image, if the edge line that detected is even
It is connected into the then reservation of a rectangle, the position mark of edge line, and is shown in pretreated image, removal is unlabelled
Pattern generates first processing picture.
Aerial Images segment module, equally that every small picture is same first processing picture segmentation at one piece of small picture of block
In sample database equally with two it is oval handle, calculate every small picture energy under 0 °, 30 °, 45 °, 60 °, 90 ° of first quartile
Amount, contrast, correlation, the value of the uniformity, and distribute four characteristic values certain weights, obtain 0 ° of first quartile, 30 °,
Each actual value under 45 °, 60 °, 90 °.
Contrast module is communicated, the actual value of the reference value and unmanned plane in the sample database of ground is compared under respective angles,
If the two gap is in certain threshold range, then it is assumed that the physical location for having found insulator chain in Aerial Images, the position
Label.
The position of label is generated a width bianry image and is obtained with originally secondary processing image dot product by insulator chain locating module
To the tangent tower insulator chain image finally extracted.
The sample memory module, finds the model of insulator chain, to different model in the design data of manufacturer
Sub-pieces establish the sample database of standard, and one kind is generated about sub-pieces dimensionality reduction gray scale texture-form matrix to sample image.
As shown in figure 3, it is the coloured image for shooting tangent tower insulator chain monolithic insulator under certain distance angle, image
Color analysis carries out different conversions to the coloured image of composite insulator and glass insulator, becomes gray level image.Then it uses
One small elliptic overlay lives a part for sub-pieces ontology, another big ellipse is more than the size of entire sub-pieces, but
Two elliptical central points and the central point of sub-pieces are completely superposed, and then retain the gray value between two elliptical sides, Fig. 3
Shown in shadow region, and the gray value of image other parts is set to 0, generates the sub-pieces gray-scale map of a width specific sample
Picture, it is dimensionality reduction gray scale texture-shape then to consider that this shape obtains the gray level co-occurrence matrixes under this kind of specific gray level image
Matrix, using this matrix (as shown in figures 5-9) calculate 0 ° of the size elliptical edge point line in first quartile, 30 °,
45 °, 60 °, 90 ° of lower energy, contrast, correlation, the value of the uniformity, and distribute four characteristic values certain weights, obtain
Each reference value under 0 °, 30 °, 45 °, 60 °, 90 °.
Specific steps are as follows shown：
Step 1.1, the model that insulator chain is found in the design data of manufacturer builds the sub-pieces of different model
The sample database of day-mark standard, then establishes the image library of extra high voltage network tangent tower insulator chain, these image libraries have
Image under different distance and shooting angle, and these tangent tower insulator chain images can be consulted and be backed up at any time；
Step 1.2, the monolithic insulator Color judgment on images of sample, two different sizes of setting are grown by long semi-minor axis
Ellipse, the sub-pieces of sample are plummet states, its disk is just an elliptical shape, therefore its long semi-minor axis is being schemed
It can be calculated as in, the small elliptic overlay for being less than half axial length of disk with one in this way lives a part for sub-pieces ontology,
Another big ellipse is more than the size of entire sub-pieces, but two elliptical central points and the central point of sub-pieces are complete
It overlapping, the interval of size elliptical side is set as w, then artificially retains the gray value between two elliptical sides, and the other portions of image
The gray value divided is set to 0, and is generated the sub-pieces gray level image of a width specific sample, is considerably reduced the data of image in this way
Amount；
Step 1.3, it is dimensionality reduction then to consider that this shape obtains the gray level co-occurrence matrixes under this kind of specific gray level image
Gray scale texture-form matrix, using this matrix calculate 0 °, 30 °, 45 °, 60 °, 90 ° of lower energy, contrast, correlation,
The value of the uniformity, and distribute four characteristic values certain weights, obtain each reference value under 0 °, 30 °, 45 °, 60 °, 90 °；
Wherein, energy, contrast, correlation, the uniformity are defined as follows：
1., energy：The quadratic sum of gray level co-occurrence matrixes element value.Energy value shows greatly a kind of more uniform and regular variation
Texture pattern.
What wherein i and j was represented is the coordinate position of gray level co-occurrence matrixes P, and L is the gray level of image.
2., contrast：Reflect the degree of the clarity and the texture rill depth of image.
Wherein i, j, L meaning are the same as (1) formula.
3., correlation：Its metric space co-occurrence matrix element be expert at or column direction on similarity degree, therefore, correlation
Size reflects local gray level correlation in image.
Wherein i, j, L meaning are the same as (1) formula.
4., the uniformity：Reflect the roughness of image texture, coarse grained uniformity is bigger, and the uniformity of close grain is smaller.
Wherein i, j, L meaning are the same as (1) formula.
Then, to 0 ° of size elliptical edge point line in first quartile, 30 °, 45 °, 60 °, 90 ° of lower energy, contrast,
Correlation, the value of the uniformity carry out weight distribution, are q1, q2, q3, q4, are multiplied by are calculated by formula (1), (2), (3), (8) respectively
Come energy, contrast, correlation, the value of the uniformity, then add up, obtain reference value.Specific formula is as follows：
RV (θ)=q1*ASM+q2*CON+q3*COR+q4*IDM
(9)
Wherein, q1+q2+q3+q4=1, θ are 0 °, 30 °, 45 °, 60 °, 90 ° desirable.
The unmanned plane module carries ultrasonic ranging during electric system unmanned plane line walking with it
Holder focusing video camera collects the coloured image of tangent tower insulator chain on the extra high voltage network under certain distance, ensures
The position of insulator chain is plummet state.But since the background place in the coloured image of acquisition is ever-changing, therefore to figure of taking photo by plane
Picture is pre-processed, including color analysis, filtering, histogram equalization, and this reduces the data volumes of image, have filtered bat
The noise generated in the process is taken the photograph, the contrast of tangent tower insulator chain and other background patterns in picture is also enhanced.
Include the following steps：
Step 2.1, to carry the super-pressure that the holder focusing video camera of ultrasonic ranging collects under certain distance with it defeated
The distance shot in the coloured image of tangent tower insulator chain in electric line, distance and sample database should it is consistent, unmanned plane with directly
The distance of transmission tower insulator chain is determined by ultrasonic wave；
Step 2.2, it is establish tri- color components of gray scale G (i), i=1,2 and R, G, B corresponding picture color analysis,
So-called gray-scale map is exactly that the value of each component of each pixel color is equal in fact.
If for composite insulator, then converted using formula (10).
G1=R/ (R+G+B) (10)
If for glass insulator, then converted using formula (11).
In formula (10), (11), R is the red component of the pixel color indicated, and similarly G, B indicate green and blue respectively
Component, G indicate the transformed grey level of point.Finally, only the pixel RGB component value need to be all set as G.With this
The gray value of value expression image, to be converted to the gray scale picture of range 0 to 255.
Step 2.3, using medium filtering, can inhibit not make edge blurry while random noise, its mathematical description
It is as follows：If S is pixel (x0, y0) field set, (x, y) indicate S in element, f (x, y) indicate (x, y) point gray scale
Value, | S | indicate the number of element in set S, Sort indicates sequence, then to (x0, y0) be smoothly represented by：
Medium filtering is not simply to be averaged, therefore generated Fuzzy comparisons are few, matlab image processing tools
The two dimension median filter processing that medfilt2 functions are used for realizing digital picture is provided in case.
Step 2.4, adaptive histogram equalization is carried out to the image after medium filtering.Reinforced insulation substring and gray-scale map
The contrast of other patterns as in.Its algorithm realizes that process is as follows：
1. providing all gray level i of original image, each gray-scale pixel number n of statistics original imagei；
2. calculating the histogram and accumulation histogram of original image；
3. k values are calculated with formula (12),
Wherein k' is proportionality coefficient,For the noise variance of entire image,For the gray variance in window W.
With formula (13)
Carry out calculating local gray-value；Wherein xi,jAnd x'i,jIndicate the front and back gray value of image of transformation, mi,jBe expressed as with
xi,jCentered on window neighboring mean value, T indicate to xi,jTransforming function transformation function.
4. calculating local contrast with formula (14), equalization is realized；
x'i,j=mi,j+k(xi,j-mi,j) (15)
Wherein xi,j、x'i,jIt is distributed as converting forward and backward center pixel,It is averaged for pixel in window W
Ash value.
5. with p (ti)=ni/ n calculates new histogram.
The edge detection module, includes the following steps：
Edge detection is carried out to the pretreated image of unmanned plane module, edge detection detects gray level image
Edge lines, since tangent tower insulator chain is plummet state, the tangent line of itself pattern or so lower edges point just should
A rectangular frame is formed, as shown in figure 3, finding the position for having such rectangular frame and label, and is shown to pretreated
In image, only retain the pattern inside label rectangle frame, remove the pattern of other parts, generates first processing image.
The step of edge detection ' canny ' operator, realization detection image edge, is as follows with method：
A. Gaussian filter smoothed image is used.
B. amplitude and the direction of filtered image gradient are calculated.
C. to gradient magnitude application non-maxima suppression, process is the Local modulus maxima found out in image gradient,
Other non local maximum zero setting, with the edge refined.
D. dual threashold value-based algorithm is used to detect and connect edge again.
Use dual-threshold voltage to carry out edge differentiation, what every edge strength was more than high threshold must be marginal point, small here
What it is in Low threshold is not centainly marginal point, if edge strength is more than Low threshold but be less than high threshold, sees this pixel
Either with or without the marginal point more than high threshold in adjacent pixels, if so, it is exactly marginal point；If not provided, it is not just edge
Point.
Then the tangent line of marginal point is sought, if in the presence of tangent line vertically and horizontally, is retained, and horizontal and vertical tangent line
Internal all position marks, and be shown in pretreated image, only retain the pattern inside label rectangle frame, removes it
The pattern of its part generates first processing image.
The Aerial Images segment module, and the first processing image generated to edge detection module is divided into the small of m × n
Picture, as shown in figure 4, m*n small picture is equally calculated every small picture with equally being handled with two ellipses in sample database
Dimensionality reduction gray scale texture-form matrix under 0 °, 30 °, 45 °, 60 °, 90 ° under size elliptical edge line, such as Fig. 5-9 institutes
Show, and obtain the energy of m*n small picture, contrast, correlation, the value of the uniformity, and distributes four characteristic values certain power
Value, obtain 0 ° of size elliptical edge point line in first quartile, 30 °, 45 °, 60 °, 90 ° lower four characteristic quantities combinations reality
Value, and the actual value under 0 °, 30 °, 45 °, 60 °, 90 ° of angles is stored in A, B, C, D, E five according to the sequence of picture number
In matrix.
Specific steps are as follows：
Step 4.1, gray level image is divided into using matlab softwares the small picture of m × n, and size elliptical edge point is connected
0 ° of line, 30 °, 45 °, 60 °, 90 ° of goniometer calculate the m*n small respective dimensionality reduction gray scale texture-form matrix of picture, the matrix
Computational methods it is identical as the generator matrix method of step 1 sample insulator, result in the dimensionality reduction under four different angles
Gray scale texture-form matrix；
Step 4.2, the energy that calculates according to formula (1), (2), (3), (8), contrast, correlation, the value of the uniformity；
Step 4.3, actual value of the every small picture under five angles is calculated also according to formula (9) method, and value is pressed
The sequence of photograph and picture number is stored in the actual value under 0 °, 30 °, 45 °, 60 °, 90 ° of angles in five matrixes of A, B, C, D, E.
The communication contrast module, includes the following steps：
Reference value and actual value are compared under 0 °, 30 °, 45 °, 60 °, 90 ° of five respective angles, as formula (16) make it is poor
Absolute value is sought, if the two gap is in certain threshold value Ti, i=1, in 2,3,4,5 ranges, then it is assumed that be that there may be tangent tower insulation
The region of substring, and mark regional location；
|RV(0°)-A|≤T1
|RV(30°)-B|≤T2
|RV(45°)-C|≤T3
|RV(60°)-D|≤T4
|RV(90°)-E|≤T5
(16)
The insulator chain locating module, includes the following steps：
The mark position obtained to communication contrast module is used, and generates a width and marks bianry image, with edge detection
The first processing image dot product of module, that removes the interference of other background areas, the tangent tower finally extracted is exhausted
Edge substring image.
Claims (8)
1. a kind of device extracting tangent tower insulator chain based on dimensionality reduction gray scale texture-form matrix, it is characterised in that including with
Lower module：
Sample memory module, the image for storing various model sub-pieces, every image have size in first quartile ellipse
The edge of the circle point line 0 °, 30 °, 45 °, 60 °, 90 ° of lower energy, contrast, correlation, the value of the uniformity, and distribute four spies
The certain weights of value indicative generate a reference value；
Unmanned plane module for obtaining the tangent tower insulator chain coloured image on extra high voltage network, and has pre-
Processing function；
Edge detection module, for carrying out edge detection to pretreated image, if the edge line that detected connects into
The then reservation of one rectangle the position mark of edge line, and is shown in pretreated image, removes unlabelled figure
Case generates first processing picture；
Aerial Images segment module, first processing picture segmentation at one piece of small picture of block, equally the every small same sample of picture
In library equally with two it is oval handle, calculate every small picture 0 ° of first quartile, 30 °, 45 °, 60 °, it is 90 ° of lower energy, right
Than degree, correlation, the value of the uniformity, and distribute four characteristic values certain weights, obtain 0 ° of first quartile, 30 °, 45 °, 60 °,
Each actual value under 90 °；
Contrast module is communicated, the actual value of the reference value and unmanned plane in the sample database of ground, the two are compared under respective angles
If gap is in certain threshold range, then it is assumed that the physical location for having found insulator chain in Aerial Images, the position mark；
The position of label is generated a width bianry image and is obtained most with originally secondary processing image dot product by insulator chain locating module
The tangent tower insulator chain image extracted eventually.
2. a kind of device extracting tangent tower insulator chain based on dimensionality reduction gray scale texture-form matrix according to claim 1,
It is characterized in that：The sample memory module, includes the following steps：
Step 1.1, the model that insulator chain is found in the design data of manufacturer establishes mark to the sub-pieces of different model
Accurate sample database, then establishes the image library of extra high voltage network tangent tower insulator chain, these image libraries have in difference
Image under distance and shooting angle, and these tangent tower insulator chain images can be consulted and be backed up at any time；
Step 1.2, different size of ellipse by long semi-minor axis length setting two the monolithic insulator Color judgment on images of sample
The sub-pieces of circle, sample are plummet states, its disk is just an elliptical shape, therefore its long semi-minor axis is in the picture
It can calculate, the small elliptic overlay for being less than half axial length of disk with one in this way lives a part for sub-pieces ontology, another
A big ellipse is more than the size of entire sub-pieces, but two elliptical central points and the central point of sub-pieces are completely heavy
It closes, the interval of size elliptical side is set as w, then artificially retains the gray value between two elliptical sides, and image other parts
Gray value set to 0, generate a width specific sample sub-pieces gray level image, be considerably reduced the data volume of image in this way；
Step 1.3, it is dimensionality reduction gray scale then to consider that this shape obtains the gray level co-occurrence matrixes under this kind of specific gray level image
Texture-form matrix, using this matrix calculate 0 °, 30 °, 45 °, 60 °, 90 ° of lower energy, contrast, correlation, uniformly
The value of degree, and distribute four characteristic values certain weights, obtain each reference value under 0 °, 30 °, 45 °, 60 °, 90 °；
Wherein, energy, contrast, correlation, the uniformity are defined as follows：
1., energy：The quadratic sum of gray level co-occurrence matrixes element value；Energy value shows greatly the texture of a kind of more uniform and regular variation
Pattern；
What wherein i and j was represented is the coordinate position of gray level co-occurrence matrixes P, and L is the gray level of image；
2., contrast：Reflect the degree of the clarity and the texture rill depth of image；
Wherein i, j, L meaning are the same as (1) formula；
3., correlation：Its metric space co-occurrence matrix element be expert at or column direction on similarity degree, therefore, correlation size
Reflect local gray level correlation in image；
Wherein i, j, L meaning are the same as (1) formula；
4., the uniformity：Reflect the roughness of image texture, coarse grained uniformity is bigger, and the uniformity of close grain is smaller；
Wherein i, j, L meaning are the same as (1) formula；
Then, to 0 °, 30 °, 45 °, 60 °, 90 ° of lower energy, contrast, correlations of size elliptical edge point line in first quartile
Property, the uniformity value carry out weight distribution, be q1, q2, q3, q4, be multiplied by calculated by formula (1), (2), (3), (8) respectively
Energy, contrast, correlation, the value of the uniformity, then add up, obtain reference value；Specific formula is as follows：
RV (θ)=q1*ASM+q2*CON+q3*COR+q4*IDM (9)
Wherein, q1+q2+q3+q4=1, θ are 0 °, 30 °, 45 °, 60 °, 90 ° desirable.
3. a kind of device extracting tangent tower insulator chain based on dimensionality reduction gray scale texture-form matrix according to claim 1,
It is characterized in that：The unmanned plane module, includes the following steps：
Step 2.1, it carries the holder focusing video camera of ultrasonic ranging with it and collects ultra high-tension transmission line under certain distance
The distance shot in the coloured image of road tangent tower insulator chain, distance and sample database answers consistent, unmanned plane and tangent tower
The distance of insulator chain is determined by ultrasonic wave；
Step 2.2, it is establish tri- color components of gray scale G (i), i=1,2 and R, G, B corresponding picture color analysis, it is so-called
Gray-scale map be exactly in fact each pixel color each component value it is equal；
If for composite insulator, then converted using formula (10)；
G1=R/ (R+G+B) (10)
If for glass insulator, then converted using formula (11)；
In formula (10), (11), R is the red component of the pixel color indicated, similarly G, and B indicates green and blue point respectively
Amount, G indicate the transformed grey level of point；Finally, only the pixel RGB component value need to be all set as G；With this value
The gray value for expressing image, to be converted to the gray scale picture of range 0 to 255；
Step 2.3, using medium filtering, can inhibit not make edge blurry while random noise, its mathematical description is as follows：
If S is pixel (x0, y0) field set, (x, y) indicate S in element, f (x, y) indicate (x, y) point gray value, | S |
Indicate the number of element in set S, Sort indicates sequence, then to (x0, y0) be smoothly represented by：
Medium filtering is not simply to be averaged, therefore generated Fuzzy comparisons are few, in matlab image processing toolboxes
Provide the two dimension median filter processing that medfilt2 functions are used for realizing digital picture；
Step 2.4, adaptive histogram equalization is carried out to the image after medium filtering；In reinforced insulation substring and gray level image
The contrast of other patterns；Its algorithm realizes that process is as follows：
1. providing all gray level i of original image, each gray-scale pixel number n of statistics original imagei；
2. calculating the histogram and accumulation histogram of original image；
3. k values are calculated with formula (12),
Wherein k' is proportionality coefficient,For the noise variance of entire image,For the gray variance in window W；
With formula (13)
Carry out calculating local gray-value；Wherein xi,jWith x 'i,jIndicate the front and back gray value of image of transformation, mi,jIt is expressed as with xi,jFor
The window neighboring mean value at center, T are indicated to xi,jTransforming function transformation function；
4. calculating local contrast with formula (14), equalization is realized；
x′i,j=mi,j+k(xi,j-mi,j) (15)
Wherein xi,j、x′i,jIt is distributed as converting forward and backward center pixel,For the average ash of pixel in window W
Value；
5. with p (ti)=ni/ n calculates new histogram.
4. a kind of device extracting tangent tower insulator chain based on dimensionality reduction gray scale texture-form matrix according to claim 1,
It is characterized in that：The edge detection module, includes the following steps：
Edge detection is carried out to the pretreated image of unmanned plane module, edge detection is to detect the edge of gray level image
Lines, since tangent tower insulator chain is plummet state, the tangent line of itself pattern or so lower edges point should just form
One rectangular frame, finds the position for having such rectangular frame and label, and is shown in pretreated image, only retains mark
Remember the pattern inside rectangle frame, remove the pattern of other parts, generates first processing image；
The step of edge detection ' canny ' operator, realization detection image edge, is as follows with method：
A. Gaussian filter smoothed image is used；
B. amplitude and the direction of filtered image gradient are calculated；
C. to gradient magnitude application non-maxima suppression, process is the Local modulus maxima found out in image gradient, other
Non local maximum zero setting, with the edge refined；
D. dual threashold value-based algorithm is used to detect and connect edge again；
Use dual-threshold voltage to carry out edge differentiation, what every edge strength was more than high threshold must be marginal point, be less than low here
Threshold value is not centainly marginal point, if edge strength is more than Low threshold but be less than high threshold, sees the adjoining of this pixel
Either with or without the marginal point more than high threshold in pixel, if so, it is exactly marginal point；If not provided, it is not just marginal point；
Then the tangent line of marginal point is sought, if in the presence of tangent line vertically and horizontally, is retained, and inside horizontal and vertical tangent line
All position marks, and be shown in pretreated image, only retain the pattern inside label rectangle frame, remove other portions
The pattern divided generates first processing image.
5. a kind of device extracting tangent tower insulator chain based on dimensionality reduction gray scale texture-form matrix according to claim 1,
It is characterized in that：The Aerial Images segment module, include the following steps：
Step 4.1, gray level image is divided into using matlab softwares the small picture of m × n, and to size elliptical edge point line
0 °, 30 °, 45 °, 60 °, 90 ° of goniometer calculate the m*n small respective dimensionality reduction gray scale texture-form matrix of picture, the matrix
Computational methods are identical as the generator matrix method of step 1 sample insulator, result in the dimensionality reduction ash under four different angles
Spend texture-form matrix；
Step 4.2, the energy that calculates according to formula (1), (2), (3), (8), contrast, correlation, the value of the uniformity；
Step 4.3, actual value of the every small picture under five angles is calculated also according to formula (9) method, and will be worth according to figure
The sequence of piece number is stored in the actual value under 0 °, 30 °, 45 °, 60 °, 90 ° of angles in five matrixes of A, B, C, D, E.
6. a kind of device extracting tangent tower insulator chain based on dimensionality reduction gray scale texture-form matrix according to claim 1,
It is characterized in that：The communication contrast module, includes the following steps：
Reference value and actual value are compared under 0 °, 30 °, 45 °, 60 °, 90 ° of five respective angles, as formula (16) carries out asking exhausted as difference
To value, if the two gap is in certain threshold value Ti, i=1, in 2,3,4,5 ranges, then it is assumed that be that there may be tangent tower insulator chains
Region, and mark regional location；
7. a kind of device extracting tangent tower insulator chain based on dimensionality reduction gray scale texture-form matrix according to claim 1,
It is characterized in that：The insulator chain locating module, includes the following steps：
The mark position obtained to communication contrast module is used, and generates a width and marks bianry image, with edge detection module
First processing image dot product, that removes the interference of other background areas, the tangent tower insulator that is finally extracted
String image.
8. a kind of method for extracting tangent tower insulator chain based on dimensionality reduction gray scale texture-form matrix, it is characterised in that：
Step 1：The model that insulator chain is found in the design data of manufacturer establishes standard to the sub-pieces of different model
Sample database, the process that a kind of dimensionality reduction gray scale texture-form matrix and reference value are generated to sample image is as follows：
The coloured image that tangent tower insulator chain monolithic insulator is shot under certain distance angle, Color judgment on images,
A part for sub-pieces ontology is lived with a small elliptic overlay, another big ellipse is more than the size of entire sub-pieces,
But two elliptical central points and the central point of sub-pieces are completely superposed, and then retain the gray value between two elliptical sides,
And the gray value of image other parts is set to 0, the sub-pieces gray level image of a width specific sample is generated, then considers this shape
It is dimensionality reduction gray scale texture-form matrix that shape, which obtains the gray level co-occurrence matrixes under this kind of specific gray level image, utilizes this matrix
Calculate in first quartile size elliptical edge point line 0 °, 30 °, 45 °, 60 °, 90 ° of lower energy, contrast, correlation,
The value of the uniformity, and distribute four characteristic values certain weights, obtain each reference value under 0 °, 30 °, 45 °, 60 °, 90 °；
Step 2：During electric system unmanned plane line walking, carries the holder focusing video camera of ultrasonic ranging with it and acquire
The coloured image of tangent tower insulator chain on extra high voltage network under to certain distance ensures that the position of insulator chain is lead
Hammer state；But since the background place in the coloured image of acquisition is ever-changing, therefore Aerial Images are pre-processed, including color
Coloured silk analysis, filtering, histogram equalization have filtered the noise generated in shooting process this reduces the data volume of image,
Also enhance the contrast of tangent tower insulator chain and other background patterns in picture；
Step 3：Edge detection is carried out to the pretreated image of step 2, edge detection is to detect the edge line of gray level image
Item, since tangent tower insulator chain is plummet state, the tangent line of itself pattern or so lower edges point should just form one
A rectangular frame finds the position for having such rectangular frame and label, and is shown in pretreated image, only retains label
Pattern inside rectangle frame removes the pattern of other parts, generates first processing image；
Step 4：The small picture of m × n is divided into the first processing image that step 3 generates, it is equally that m*n small picture is same
Ben Kuli equally with two oval processing, calculates drop of the every small picture under 0 °, 30 °, 45 °, 60 °, 90 ° of first quartile
Gray scale texture-form matrix is tieed up, and obtains the energy of m*n small picture, contrast, correlation, the value of the uniformity, and distributes four
The certain weights of a characteristic value obtain each actual value under 0 °, 30 °, 45 °, 60 °, 90 °, and according to the sequence of picture number
Actual value under 0 °, 30 °, 45 °, 60 °, 90 ° of angles is stored in five matrixes of A, B, C, D, E；
Step 5：Reference value and actual value are compared under five respective angles, carry out seeking absolute value as difference, if the two gap is certain
In threshold range, then it is assumed that be that there may be the regions of tangent tower insulator chain, and mark regional location；
Step 6：It is used to the mark position that step 5 obtains, generates a width and mark bianry image, the first place with step 3
Manage image dot product, that removes the interference of other background areas, the tangent tower insulator chain image that is finally extracted.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711105128.4A CN108665468B (en) | 2017-11-10 | 2017-11-10 | Device and method for extracting tangent tower insulator string |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711105128.4A CN108665468B (en) | 2017-11-10 | 2017-11-10 | Device and method for extracting tangent tower insulator string |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108665468A true CN108665468A (en) | 2018-10-16 |
| CN108665468B CN108665468B (en) | 2021-05-14 |
Family
ID=63785013
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711105128.4A Active CN108665468B (en) | 2017-11-10 | 2017-11-10 | Device and method for extracting tangent tower insulator string |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108665468B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110378921A (en) * | 2019-07-22 | 2019-10-25 | 江苏海洋大学 | Navigation channel substrate stratum boundary intelligent identification Method based on mud scum rheological behavior and gray level co-occurrence matrixes |
| CN111539954A (en) * | 2020-05-25 | 2020-08-14 | 国网湖南省电力有限公司 | Method, system and medium for identifying cable buffer layer defect by adopting X-ray digital image characteristics |
| CN111984034A (en) * | 2020-08-24 | 2020-11-24 | 广东电网有限责任公司 | Unmanned aerial vehicle electric power patrols line system |
| CN112991432A (en) * | 2021-04-15 | 2021-06-18 | 重庆大学 | Icing shape identification method based on image processing |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102508110A (en) * | 2011-10-10 | 2012-06-20 | 上海大学 | Texture-based insulator fault diagnostic method |
| CN103529362A (en) * | 2013-10-28 | 2014-01-22 | 国家电网公司 | Perception based insulator recognition and defect diagnosis method |
| CN105427316A (en) * | 2015-11-25 | 2016-03-23 | 国网吉林省电力有限公司电力科学研究院 | Method for extracting insulator single disc surfaces in electric transmission line visible images |
| US9555264B1 (en) * | 2011-02-15 | 2017-01-31 | Velayudhan Sahadevan | MEMS based parallel microbeam radiosurgery without adaptive resistance to radiation |
| CN106570853A (en) * | 2015-10-08 | 2017-04-19 | 上海深邃智能科技有限公司 | Shape and color integration insulator identification and defect detection method |
-
2017
- 2017-11-10 CN CN201711105128.4A patent/CN108665468B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9555264B1 (en) * | 2011-02-15 | 2017-01-31 | Velayudhan Sahadevan | MEMS based parallel microbeam radiosurgery without adaptive resistance to radiation |
| CN102508110A (en) * | 2011-10-10 | 2012-06-20 | 上海大学 | Texture-based insulator fault diagnostic method |
| CN103529362A (en) * | 2013-10-28 | 2014-01-22 | 国家电网公司 | Perception based insulator recognition and defect diagnosis method |
| CN106570853A (en) * | 2015-10-08 | 2017-04-19 | 上海深邃智能科技有限公司 | Shape and color integration insulator identification and defect detection method |
| CN105427316A (en) * | 2015-11-25 | 2016-03-23 | 国网吉林省电力有限公司电力科学研究院 | Method for extracting insulator single disc surfaces in electric transmission line visible images |
Non-Patent Citations (2)
| Title |
|---|
| QINGGANG WU等: "《A Texture Segmentation Algorithm Based on PCA and Global Minimization Active Contour Model for Aerial Insulator Images》", 《 IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING》 * |
| 于兰英 等: "《一种基于多特征的绝缘子识别方法》", 《电瓷避雷器》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110378921A (en) * | 2019-07-22 | 2019-10-25 | 江苏海洋大学 | Navigation channel substrate stratum boundary intelligent identification Method based on mud scum rheological behavior and gray level co-occurrence matrixes |
| CN110378921B (en) * | 2019-07-22 | 2023-03-14 | 江苏海洋大学 | Intelligent identification method for substrate layer boundary of channel based on floating mud rheological property and gray level co-occurrence matrix |
| CN111539954A (en) * | 2020-05-25 | 2020-08-14 | 国网湖南省电力有限公司 | Method, system and medium for identifying cable buffer layer defect by adopting X-ray digital image characteristics |
| CN111539954B (en) * | 2020-05-25 | 2024-01-23 | 国网湖南省电力有限公司 | Method, system and medium for identifying cable buffer layer defect by adopting X-ray digital image characteristics |
| CN111984034A (en) * | 2020-08-24 | 2020-11-24 | 广东电网有限责任公司 | Unmanned aerial vehicle electric power patrols line system |
| CN112991432A (en) * | 2021-04-15 | 2021-06-18 | 重庆大学 | Icing shape identification method based on image processing |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108665468B (en) | 2021-05-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109859171A (en) | A kind of flooring defect automatic testing method based on computer vision and deep learning | |
| CN108665468A (en) | A kind of device and method extracting tangent tower insulator chain based on dimensionality reduction gray scale texture-form matrix | |
| CN103839065B (en) | Extraction method for dynamic crowd gathering characteristics | |
| CN102132323B (en) | System and method for automatic image straightening | |
| CN111292321B (en) | Transmission line insulator defect image identification method | |
| CN107679495B (en) | Detection method for movable engineering vehicles around power transmission line | |
| CN109410207A (en) | A kind of unmanned plane line walking image transmission line faultlocating method based on NCC feature | |
| CN109544501A (en) | A kind of transmission facility defect inspection method based on unmanned plane multi-source image characteristic matching | |
| CN105354847A (en) | Fruit surface defect detection method based on adaptive segmentation of sliding comparison window | |
| CN113887412B (en) | Detection method, detection terminal, monitoring system and storage medium for pollution emission | |
| CN110210608A (en) | The enhancement method of low-illumination image merged based on attention mechanism and multi-level features | |
| CN104867134A (en) | Identification method for transmission line tower inspection by unmanned aerial vehicle | |
| CN108537170A (en) | A kind of power equipment firmware unmanned plane inspection pin missing detection method | |
| CN108665464A (en) | A kind of foreign matter detecting method based on morphologic high tension electric tower and high-tension bus-bar | |
| CN107481237A (en) | A kind of infrared array image hot spot detection method based on multiframe temperature characterisitic | |
| CN114445331A (en) | Cable intermediate joint construction defect detection method, system and device based on image recognition | |
| Yang et al. | Vehicle color recognition using monocular camera | |
| CN112288682A (en) | Electric power equipment defect positioning method based on image registration | |
| CN101561316B (en) | On-line test visual data processing system based on region of interest (ROI) | |
| CN110245592A (en) | A method of for promoting pedestrian's weight discrimination of monitoring scene | |
| CN112686120B (en) | Power transmission line abnormity detection method based on unmanned aerial vehicle aerial image | |
| CN111310899B (en) | Power defect identification method based on symbiotic relation and small sample learning | |
| CN108830834B (en) | Automatic extraction method for video defect information of cable climbing robot | |
| CN109544608B (en) | Unmanned aerial vehicle image acquisition characteristic registration method | |
| CN114283157A (en) | Ellipse fitting-based ellipse object segmentation method |
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 | ||
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20181016 Assignee: Yichang Yizhixing Technology Co.,Ltd. Assignor: CHINA THREE GORGES University Contract record no.: X2023980034895 Denomination of invention: A device and method for extracting insulator strings from linear towers Granted publication date: 20210514 License type: Common License Record date: 20230426 |