CN105701835A - Image edge detection method and system facing electric power facilities - Google Patents
Image edge detection method and system facing electric power facilities Download PDFInfo
- Publication number
- CN105701835A CN105701835A CN201610108566.5A CN201610108566A CN105701835A CN 105701835 A CN105701835 A CN 105701835A CN 201610108566 A CN201610108566 A CN 201610108566A CN 105701835 A CN105701835 A CN 105701835A
- Authority
- CN
- China
- Prior art keywords
- image
- edge
- edge detection
- node
- electric power
- 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 66
- 238000003708 edge detection Methods 0.000 title claims abstract description 55
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 55
- 230000011218 segmentation Effects 0.000 claims abstract description 29
- 230000005540 biological transmission Effects 0.000 claims abstract description 27
- 238000001914 filtration Methods 0.000 claims abstract description 21
- 238000003709 image segmentation Methods 0.000 claims abstract description 9
- 230000005611 electricity Effects 0.000 claims description 25
- 238000001514 detection method Methods 0.000 claims description 22
- 238000007637 random forest analysis Methods 0.000 claims description 14
- 239000011159 matrix material Substances 0.000 claims description 7
- 238000012216 screening Methods 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 5
- 238000003066 decision tree Methods 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 abstract 1
- 239000012212 insulator Substances 0.000 description 3
- 230000002547 anomalous effect Effects 0.000 description 2
- 230000002411 adverse Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
Classifications
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
-
- 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
-
- 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
- G06T2207/20081—Training; Learning
Abstract
The present invention belongs to the field of the computer image processing technology, and especially relates to an image edge detection method and system facing electric power facilities. The method comprises: obtaining image data of an electric power facility, obtaining an image to perform image preprocessing consisting of image graying, image enhancement, histogram equalization, median filtering and image segmentation through adoption a threshold segmentation improvement algorithm, and then performing edge detection through adoption of a dual-structured structure forest improvement algorithm; combining the threshold segmentation improvement algorithm and a traditional edge detection operator to rapidly and clearly detect the transmission line icing edge and neglect the texture condition in the transmission line; and employing structured learning input for picture information to ensure the reduced processing data volume and effectively distinguish whether the information in the picture is the edge information or not, so that the speed of image edge detection is enhanced, and more importantly, the accuracy of the edge detection algorithm is improved.
Description
Technical field
The invention belongs to computer image processing technology field, particularly relate to a kind of method for detecting image edge towards electric power facility and system。
Background technology
In numerous adverse effects to electric power facility, icing is exactly that electrical equipment is affected the most serious one。The electrical equipments such as transmission line of electricity can cause a lot of harm when icing is serious, for instance the toppling over of shaft tower, shaft tower deformation, wire fracture, insulator damage and flashover etc.。
Edge detection method in correlation technique is when detecting image border, it is necessary to consumes the substantial amounts of time, can not meet the requirement of real-time, and the accuracy extracting edge is relatively low。
Summary of the invention
In order to fill up the blank of existing electric power facility accident detection system such as icing rim detection, and solving Image Edge-Detection process longer, problem that accuracy is not high consuming time, the present invention proposes a kind of method for detecting image edge towards electric power facility and system。
Method includes:
Step 1, obtain electric power facility view data I (x, y)；
Step 2, above-mentioned image is carried out Image semantic classification, including image gray processing, image enhaucament, histogram equalization, medium filtering and adopt Threshold segmentation innovatory algorithm that image is split, obtain pretreated greyscale image data J (x, y)；
Step 3, by the structure forest innovatory algorithm of dual-structured, image border is detected。
Image is split by described employing Threshold segmentation innovatory algorithm, specifically includes:
By Threshold Segmentation Algorithm and traditional edge detection operator in combination with together with, the edge of powerline ice-covering can quickly and be clearly detected by traditional edge detection operator, and pass through Threshold Segmentation Algorithm, then the texture situation within transmission line of electricity can be neglected, the convenient thickness measure for powerline ice-covering。
Image border is detected by the described structure forest innovatory algorithm by dual-structured, specifically includes:
In linear discrete space, input and output for various algorithms all carry out structurized definition；
For the output valve Y of random forest, adopt the method for structuring study one pixel and point distributed about thereof to be combined, form a region Z；
Using powerline ice-covering picture as input, extract the structure in input picture and texture information, obtaining line edges output information by model training and random forest rim detection, the method then passing through structuring study obtains last powerline ice-covering edge-detected image。
Further, in such a way a pixel and point distributed about thereof are combined: ∏: Y → Z。Wherein, Y is a structuring output valve comprising 16 × 16 pixels, by the process of above formula, the information of each pixel in 16 × 16 these regions is transformed into Z；Z is the binary variable of, and the inside contains the binary-coded information of each pixel in Y, and Z is the vector of a multidimensional；Below equation need to be met when Z is carried out dimensionality reduction:Wherein p*Extensive error for structure forest；S is the classical strength of tree, and p is the average correlation coefficient between tree。
Further as, each decision tree in structure forest innovatory algorithm is all regarded a Weak Classifier, adopt the method for recurrence to be independently trained；For arriving the training sample set Ω of the transmission line of electricity image of node, calculate and obtain a sorting criterion h (x, Θ), calculate sorting criterion according to the following formula: h (x, Θ)=δ (φT(x) ψ (x) > 0), x ∈ RMRepresenting training sample, the parameter that Θ=[φ, ψ] is Weak Classifier, wherein Θ () is screening function, and ψ is Argument List vector or parameter matrix, and Θ determines the Classification Hyperplane form of Weak Classifier, and δ is an indicator function；
When training sample meets h (x, Θ) less than when setting threshold value, it falls into left child node；Otherwise, then right child node is fallen into；Recurrence said process is until falling into the number of samples of node less than threshold value, or sample purity is more than threshold value；Now, this node becomes leaf node, otherwise, to find the coefficient Θ of optimum at each node place*The sorted information gain of training sample is made to maximize；Leaf node all of after training is combined, just constitutes grader and model that transmission line of electricity training obtains。
Acquisition module that system includes being sequentially connected, pretreatment module, edge detection module；
Acquisition module: be used for obtaining electric power facility view data I (x, y)；
Pretreatment module: for above-mentioned electric power facility view data I (x, y) proceed gray processing, histogram equalization, image filtering and image segmentation, obtain pretreated view data J (x, y)；
Edge detection module: for (x y) carries out the extraction at edge to above-mentioned pretreated view data J。
Gray processing unit that described pretreatment module includes being sequentially connected, histogram equalization unit, image filtering unit, image segmentation unit:
Gray processing unit: utilize the electric power facility coloured image that video camera is obtained by weighted mean method to be converted to gray level image, highlights texture edge and the overall permanence of image to greatest extent；
Histogram equalization unit: utilize histogram equalization method to change pixel tonal range and make the edge of image highlight reflecting of better off such that it is able to better obtain the details of image, and then strengthen the accuracy of electrical equipment rim detection；
Image filtering unit: adopt the method for medium filtering to be filtered image processing；
Image segmentation unit: adopt Threshold segmentation innovatory algorithm, by Threshold Segmentation Algorithm and traditional edge detection operator in combination with together with, the edge of powerline ice-covering can quickly and be clearly detected by traditional edge detection operator, and pass through Threshold Segmentation Algorithm, then the texture situation within transmission line of electricity can be neglected, the convenient thickness measure for powerline ice-covering。
Described edge detection module, the view data for obtaining after pretreatment carries out the extraction at edge, including:
First determines unit: for data sample is trained, and uses for later stage classification；
Second determines unit: by random forests algorithm, image is classified, thus extracting image border。
Described first determines that unit includes:
Module training unit: for arriving the training sample set Ω of the transmission line of electricity image of node, calculate and obtain a sorting criterion h (x, Θ), calculate sorting criterion according to the following formula: h (x, Θ)=δ (φT(x) ψ (x) > 0), x ∈ RMRepresenting training sample, the parameter that Θ=[φ, ψ] is Weak Classifier, wherein Θ () is screening function, and ψ is Argument List vector or parameter matrix, and Θ determines the Classification Hyperplane form of Weak Classifier, and δ is an indicator function；
When training sample meets h (x, Θ) less than when setting threshold value, it falls into left child node；Otherwise, then right child node is fallen into；Recurrence said process is until falling into the number of samples of node less than threshold value, or sample purity is more than threshold value；Now, this node becomes leaf node, otherwise, to find the coefficient Θ of optimum at each node place*The sorted information gain of training sample is made to maximize；Leaf node all of after training is combined, just constitutes grader and model that transmission line of electricity training obtains。
Described second determines that unit includes:
Edge detection unit: for by the model that obtains after training image border being detected and extracting。
The beneficial effects of the present invention is: before electric power facility image enters random forests algorithm process, first the information of picture is carried out the input of structuring study, so both can guarantee that the minimizing processing data volume, and by the mode that structuring learns, can effectively distinguish whether the information in picture is marginal information, so not only increasing the speed of Image Edge-Detection, what is more important improves the accuracy of edge detection algorithm。
Accompanying drawing explanation
Fig. 1 is the flow chart of electric power facility Image Edge-Detection according to embodiments of the present invention；
Fig. 2 is the schematic flow sheet of the electric power facility Image Edge-Detection of preferred embodiment according to embodiments of the present invention；
Fig. 3 is the structured flowchart of electric power facility Image Edge-Detection device according to embodiments of the present invention。
Detailed description of the invention
Below in conjunction with accompanying drawing, embodiment is elaborated。
Fig. 1 is the flow chart of the method for electric power facility Image Edge-Detection according to embodiments of the present invention, as it is shown in figure 1, the method mainly includes step 1 to step 3。
Step 1, and acquisition electric power facility image I (x, y)。
Step 2, carries out Image semantic classification by above-mentioned image, including image gray processing, image enhaucament, histogram equalization, medium filtering and adopt Threshold segmentation innovatory algorithm carry out image segmentation, obtain pretreated greyscale image data J (x, y)。
Step 3, is detected image border by the structure forest innovatory algorithm of dual-structured。
The technical scheme of the application embodiment of the present invention, during detection electric power facility image border, improves the accuracy of Image Edge-Detection, shortens the time of rim detection。
Adopt Threshold segmentation innovatory algorithm that image is split, by Threshold Segmentation Algorithm and traditional edge detection operator in combination with together with, the edge of powerline ice-covering can quickly and be more clearly detected by traditional edge detection operator, and pass through Threshold Segmentation Algorithm, then the texture situation within transmission line of electricity can be neglected, for the thickness measure of powerline ice-covering after convenient
The method of the embodiment of the present invention may apply to several scenes, and especially various electric power facilities, such as shaft tower, wire, insulator, the rim detection of the images such as transformator。
Below the preferred implementation of the embodiment of the present invention each step above-mentioned is described。
(1) step 2
Utilizing the electric power facility coloured image that video camera is obtained by weighted mean method to be converted to gray level image in an embodiment of the embodiment of the present invention, weighted mean method can highlight texture edge and the overall permanence of image to greatest extent。
Further, in an embodiment of the embodiment of the present invention, utilizing histogram equalization method to change pixel tonal range makes the edge of image highlight reflecting of better off such that it is able to better obtains the details of image, and then strengthens the accuracy of electrical equipment rim detection。
Further, in an embodiment of the embodiment of the present invention, the non-linear filtering method of medium filtering is adopted to be filtered image processing。This method can remove the impact that noise brings, and can retain again the detail section information of image as much as possible, contribute to rim detection below。
Further, in an embodiment of the embodiment of the present invention, adopt Threshold segmentation innovatory algorithm, by Threshold Segmentation Algorithm and traditional edge detection operator in combination with together with。Can quickly and more clearly be detected the edge of powerline ice-covering by traditional edge detection operator, and by Threshold Segmentation Algorithm, then the texture situation within transmission line of electricity can be neglected, for the thickness measure of powerline ice-covering after convenient。
Certainly, in actual applications, it is also possible to apply other criterions and carry out image gray processing, image enhaucament, histogram equalization and image filtering, to obtain being prone to the gray level image that image border detected by the later stage, this is not construed as limiting by the embodiment of the present invention。
(2) step 3
In embodiments of the present invention, electric power facility image border is detected by the structure forest innovatory algorithm utilizing dual-structured。
Further, using powerline ice-covering picture as input, extract the structure in input picture and texture information, obtaining line edges output information by the automated border detection of training and the random forest automatically of model, the method then passing through structuring study obtains last powerline ice-covering edge-detected image。
Further, in linear discrete space, input and output for various algorithms carry out structurized definition。Including: for the output valve Y of random forest, adopt the method for structuring study one pixel and point distributed about thereof to be combined, form a region Z。
Further, in such a way a pixel and point distributed about thereof are combined: ∏: Y → Z。Y is a structuring output valve comprising 16 × 16 pixels, and by the process of above-mentioned formula, the information of each pixel in 16 × 16 these regions is transformed into Z by it。Z is a very long binary variable, and the inside contains the binary-coded information of each pixel in Y, and Z is the vector of a multidimensional。
Further, below equation need to be met when Z being carried out dimensionality reduction:Wherein p*Extensive error for structure forest；S is the classical strength of tree, and p is the average correlation coefficient between tree。In the process that Z is carried out reduction dimension, it is necessary to assure the extensive error of structure forest is less than maximum。
Further, above-mentioned transmission line of electricity edge output information is obtained by the automated border detection of training and the random forest automatically of model。Each decision tree in structure forest algorithm can regard a weak grader as, and they are all independently be trained by the method for recurrence。For arriving the training sample set Ω of the transmission line of electricity image of node, calculate and obtain a sorting criterion h (x, Θ), calculate sorting criterion: h (x, Θ)=δ (φ according to below equationT(x) ψ (x) > 0)。X ∈ RMRepresenting training sample, the parameter that Θ=[φ, ψ] is this Weak Classifier, wherein Θ () is screening function, and ψ is Argument List vector or parameter matrix。Θ determines the Classification Hyperplane form of Weak Classifier。δ is an indicator function。When sample meets h (x, Θ) less than when setting threshold value, it falls into left child node；Otherwise, then right child node is fallen into。The above-mentioned process of recurrence is until falling into the number of samples of node less than threshold value, or sample purity is more than threshold value。Now, this node becomes leaf node。Otherwise, the coefficient Θ of optimum is found at each node place*The sorted information gain of training sample is made to maximize。By all of leaf node after training according to any principle combinations together, just constitute the transmission line of electricity grader that obtains of training and model。
Before image enters random forests algorithm process, first the information of picture is carried out the input of structuring study, both can guarantee that the minimizing processing data volume, and by the mode that structuring learns, can effectively distinguish whether the information in picture is marginal information, not only increase the speed of Image Edge-Detection, and improve the accuracy of edge detection algorithm。
With an instantiation, the said method of the embodiment of the present invention is described in detail below。
Fig. 2 is the schematic flow sheet of the electric power facility Image Edge-Detection of preferred embodiment according to embodiments of the present invention, as in figure 2 it is shown, the method mainly includes the following aspects:
1, utilize the electric power facility coloured image that video camera is obtained by weighted mean method to be converted to gray level image, highlight texture edge and the overall permanence of image to greatest extent。
2, utilize histogram equalization method to change pixel tonal range and make the edge of image highlight reflecting of better off such that it is able to better obtain the details of image, and then strengthen the accuracy of electrical equipment rim detection。
3, adopt the non-linear filtering method of medium filtering to be filtered image processing。This method can remove the impact that noise brings, and can retain again the detail section information of image as much as possible, contribute to rim detection below。
4, adopt Threshold segmentation innovatory algorithm, by Threshold Segmentation Algorithm and traditional edge detection operator in combination with together with。Can quickly and more clearly be detected the edge of powerline ice-covering by traditional edge detection operator, and by Threshold Segmentation Algorithm, then the texture situation within transmission line of electricity can be neglected, for the thickness measure of powerline ice-covering after convenient。
5, in linear discrete space, input and output for various algorithms carry out structurized definition。For the output valve Y of random forest, adopt the method for structuring study one pixel and point distributed about thereof to be combined, form a region Z。Wherein, Y is a structuring output valve comprising 16 × 16 pixels, and Z is a very long binary variable, and the inside contains the binary-coded information of each pixel in Y, and Z is the vector of a multidimensional。
6, obtain above-mentioned transmission line of electricity edge output information by the automated border detection of training and the random forest automatically of model。Each decision tree in structure forest algorithm can regard a weak grader as, and they are all independently be trained by the method for recurrence。For arriving the training sample set Ω of the transmission line of electricity image of node, calculate and obtain a sorting criterion h (x, Θ), calculate sorting criterion: h (x, Θ)=δ (φ according to below equationT(x) ψ (x) > 0)。X ∈ RMRepresenting training sample, the parameter that Θ=[φ, ψ] is this Weak Classifier, wherein Θ () is screening function, and ψ is Argument List vector or parameter matrix。Θ determines the Classification Hyperplane form of Weak Classifier。δ is an indicator function。When sample meets h (x, Θ) less than when setting threshold value, it falls into left child node；Otherwise, then right child node is fallen into。The above-mentioned process of recurrence is until falling into the number of samples of node less than threshold value, or sample purity is more than threshold value。Now, this node becomes leaf node。Otherwise, the coefficient Θ of optimum is found at each node place*The sorted information gain of training sample is made to maximize。By all of leaf node after training according to any principle combinations together, just constitute the transmission line of electricity grader that obtains of training and model。
According to embodiments of the present invention, additionally provide the device of a kind of Image Edge-Detection, in order to realize the said method that the embodiment of the present invention provides。
Fig. 3 is the structured flowchart of Image Edge-Detection device according to embodiments of the present invention, as it is shown on figure 3, this device specifically includes that acquisition module 1, pretreatment module 2 and edge detection module 3。Wherein, acquisition module 1, be connected with photographic head, for obtain electric power facility view data I (x, y)；Pretreatment module 2, is connected with acquisition module 1, for upper electric power facility image data I (x, y) proceeds gray processing, histogram equalization and image filtering, obtain pretreated greyscale image data J (x, y)；Edge detection module 3, is connected with pretreatment module 2, for the view data J obtained after above-mentioned pretreatment, (x y) carries out the extraction at edge。
The technical scheme of the application embodiment of the present invention, improves the speed of Image Edge-Detection and the accuracy of edge detection algorithm。The method of the embodiment of the present invention may apply to several scenes, and especially various electric power facilities, such as shaft tower, wire, insulator, the rim detection of the images such as transformator。
(1) pretreatment module 2
In an embodiment of the embodiment of the present invention, photographic head the electric power facility image obtained can be processed by pretreatment module 2, uses for edge detection module 3。
Further, in an embodiment of the embodiment of the present invention, pretreatment module 2 may include that gray processing unit, utilizes the electric power facility coloured image that video camera is obtained by weighted mean method to be converted to gray level image, highlights texture edge and the overall permanence of image to greatest extent；Histogram equalization unit: utilize histogram equalization method to change pixel tonal range and make the edge of image highlight reflecting of better off such that it is able to better obtain the details of image, and then strengthen the accuracy of electrical equipment rim detection；Filter unit: adopt the non-linear filtering method of medium filtering to be filtered image processing；Image segmentation unit: adopt Threshold segmentation innovatory algorithm, by Threshold Segmentation Algorithm and traditional edge detection operator in combination with together with, the edge of powerline ice-covering can quickly and be more clearly detected by traditional edge detection operator, and pass through Threshold Segmentation Algorithm, then the texture situation within transmission line of electricity can be neglected, for the thickness measure of powerline ice-covering after convenient。
Certainly, in actual applications, it is also possible to apply other criterions and carry out image gray processing, image enhaucament, histogram equalization and image filtering, to obtain being prone to the gray level image that image border detected by the later stage, this is not construed as limiting by the embodiment of the present invention。
(2) edge detection module 3
In an embodiment of the embodiment of the present invention, edge detection module 3 can utilize the structure forest innovatory algorithm of dual-structured that above-mentioned pretreated view data is carried out edge extracting detection。
Further, above-mentioned edge detection module 3 may include that first determines unit, for data sample is trained, uses for later stage classification；Second determines unit, by random forests algorithm, image is classified, thus extracting image border。
In an embodiment of the embodiment of the present invention, above-mentioned first determines that unit may include that module training unit。For arriving the training sample set Ω of the transmission line of electricity image of node, calculate and obtain a sorting criterion h (x, Θ), calculate sorting criterion: h (x, Θ)=δ (φ according to below equationT(x) ψ (x) > 0)。X ∈ RMRepresenting training sample, the parameter that Θ=[φ, ψ] is this Weak Classifier, wherein Θ () is screening function, and ψ is Argument List vector or parameter matrix。Θ determines the Classification Hyperplane form of Weak Classifier。δ is an indicator function。When sample meets h (x, Θ) less than when setting threshold value, it falls into left child node；Otherwise, then right child node is fallen into。The above-mentioned process of recurrence is until falling into the number of samples of node less than threshold value, or sample purity is more than threshold value。Now, this node becomes leaf node。Otherwise, the coefficient Θ of optimum is found at each node place*The sorted information gain of training sample is made to maximize。By all of leaf node after training according to any principle combinations together, just constitute the transmission line of electricity grader that obtains of training and model。
Further, in an embodiment of the embodiment of the present invention, above-mentioned second determines that unit includes: edge detection unit, for by the model that obtains after above-mentioned training image border being detected and extracting。
For electric power facility anomalous event diagnostic system, the technical scheme of application the invention process example, after obtaining electric power facility image, image is carried out pretreatment, and carry out edge extracting by the structure forest innovatory algorithm of dual-structured, the speed of rim detection can be accelerated, and the accuracy of rim detection can be improved。This Image Edge-Detection system is carry out the rim detection of electric power facility merely with video camera to provide possibility, diagnoses for the anomalous event of electric power facility and reduces cost。
Those skilled in the art is it can be understood that can add the mode of required general hardware platform by software to the present invention and realize, naturally it is also possible to by hardware, but in a lot of situation, the former is embodiment more preferably。Based on such understanding, the part that prior art is contributed by technical scheme substantially in other words can embody with the form of software product, this computer software product can be stored in storage medium, such as ROM/RAM, magnetic disc, CD etc., including some instructions with so that a computer equipment (can be personal computer, server, or the network equipment etc.) perform the method described in some part of each embodiment of the present invention or embodiment。
This embodiment is only the present invention preferably detailed description of the invention; but protection scope of the present invention is not limited thereto; any those familiar with the art in the technical scope that the invention discloses, the change that can readily occur in or replacement, all should be encompassed within protection scope of the present invention。Therefore, protection scope of the present invention should be as the criterion with scope of the claims。
Claims (10)
1. the method for detecting image edge towards electric power facility, it is characterised in that including:
Step 1, obtain electric power facility view data I (x, y)；
Step 2, above-mentioned image is carried out Image semantic classification, including image gray processing, image enhaucament, histogram equalization, medium filtering and adopt Threshold segmentation innovatory algorithm that image is split, obtain pretreated greyscale image data J (x, y)；
Step 3, by the structure forest innovatory algorithm of dual-structured, image border is detected。
2. method according to claim 1, it is characterised in that image is split by described employing Threshold segmentation innovatory algorithm, specifically includes:
By Threshold Segmentation Algorithm and traditional edge detection operator in combination with together with, the edge of powerline ice-covering can quickly and be clearly detected by traditional edge detection operator, and pass through Threshold Segmentation Algorithm, then the texture situation within transmission line of electricity can be neglected, the convenient thickness measure for powerline ice-covering。
3. method according to claim 1, it is characterised in that image border is detected by the described structure forest innovatory algorithm by dual-structured, specifically includes:
In linear discrete space, input and output for various algorithms all carry out structurized definition；
For the output valve Y of random forest, adopt the method for structuring study one pixel and point distributed about thereof to be combined, form a region Z；
Using powerline ice-covering picture as input, extract the structure in input picture and texture information, obtaining line edges output information by model training and random forest rim detection, the method then passing through structuring study obtains last powerline ice-covering edge-detected image。
4. method according to claim 3, it is characterized in that, described a pixel and point distributed about thereof are combined and specifically included: ∏: Y → Z, wherein, Y is a structuring output valve comprising 16 × 16 pixels, by the process of above formula, the information of each pixel in 16 × 16 these regions is transformed into Z；Z is the binary variable of, and the inside contains the binary-coded information of each pixel in Y, and Z is the vector of a multidimensional；Below equation need to be met when Z is carried out dimensionality reduction:Wherein p*Extensive error for structure forest；S is the classical strength of tree, and p is the average correlation coefficient between tree。
5. method according to claim 1, it is characterised in that each decision tree in the structure forest innovatory algorithm in described step 3 is all regarded as a Weak Classifier, adopts the method for recurrence to be independently trained；For arriving the training sample set Ω of the transmission line of electricity image of node, calculate and obtain a sorting criterion h (x, Θ), calculate sorting criterion according to the following formula: h (x, Θ)=δ (φT(x) ψ (x) > 0), x ∈ RMRepresenting training sample, the parameter that Θ=[φ, ψ] is Weak Classifier, wherein Θ () is screening function, and ψ is Argument List vector or parameter matrix, and Θ determines the Classification Hyperplane form of Weak Classifier, and δ is an indicator function；
When training sample meets h (x, Θ) less than when setting threshold value, it falls into left child node；Otherwise, then right child node is fallen into；Recurrence said process is until falling into the number of samples of node less than threshold value, or sample purity is more than threshold value；Now, this node becomes leaf node, otherwise, to find the coefficient Θ of optimum at each node place*The sorted information gain of training sample is made to maximize；Leaf node all of after training is combined, just constitutes grader and model that transmission line of electricity training obtains。
6. the Image Edge-Detection system towards electric power facility, it is characterised in that the acquisition module that includes being sequentially connected, pretreatment module, edge detection module；
Acquisition module: be used for obtaining electric power facility view data I (x, y)；
Pretreatment module: for above-mentioned electric power facility view data I (x, y) proceed gray processing, histogram equalization, image filtering and image segmentation, obtain pretreated view data J (x, y)；
Edge detection module: for (x y) carries out the extraction at edge to above-mentioned pretreated view data J。
7. system according to claim 6, it is characterised in that gray processing unit that described pretreatment module includes being sequentially connected, histogram equalization unit, image filtering unit, image segmentation unit:
Gray processing unit: utilize the electric power facility coloured image that video camera is obtained by weighted mean method to be converted to gray level image, highlights texture edge and the overall permanence of image to greatest extent；
Histogram equalization unit: utilize histogram equalization method to change pixel tonal range and make the edge of image highlight reflecting of better off such that it is able to better obtain the details of image, and then strengthen the accuracy of electrical equipment rim detection；
Image filtering unit: adopt the method for medium filtering to be filtered image processing；
Image segmentation unit: adopt Threshold segmentation innovatory algorithm, by Threshold Segmentation Algorithm and traditional edge detection operator in combination with together with, the edge of powerline ice-covering can quickly and be clearly detected by traditional edge detection operator, and pass through Threshold Segmentation Algorithm, then the texture situation within transmission line of electricity can be neglected, the convenient thickness measure for powerline ice-covering。
8. system according to claim 6, it is characterised in that described edge detection module, the view data for obtaining after pretreatment carries out the extraction at edge, including:
First determines unit: for data sample is trained, and uses for later stage classification；
Second determines unit: by random forests algorithm, image is classified, thus extracting image border。
9. system according to claim 8, it is characterised in that described first determines that unit includes:
Module training unit: for arriving the training sample set Ω of the transmission line of electricity image of node, calculate and obtain a sorting criterion h (x, Θ), calculate sorting criterion according to the following formula: h (x, Θ)=δ (φT(x) ψ (x) > 0), x ∈ RMRepresenting training sample, the parameter that Θ=[φ, ψ] is Weak Classifier, wherein Θ () is screening function, and ψ is Argument List vector or parameter matrix, and Θ determines the Classification Hyperplane form of Weak Classifier, and δ is an indicator function；
When training sample meets h (x, Θ) less than when setting threshold value, it falls into left child node；Otherwise, then right child node is fallen into；Recurrence said process is until falling into the number of samples of node less than threshold value, or sample purity is more than threshold value；Now, this node becomes leaf node, otherwise, to find the coefficient Θ of optimum at each node place*The sorted information gain of training sample is made to maximize；Leaf node all of after training is combined, just constitutes grader and model that transmission line of electricity training obtains。
10. system according to claim 8, it is characterised in that described second determines that unit includes:
Edge detection unit: for by the model that obtains after training image border being detected and extracting。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610108566.5A CN105701835A (en) | 2016-02-26 | 2016-02-26 | Image edge detection method and system facing electric power facilities |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610108566.5A CN105701835A (en) | 2016-02-26 | 2016-02-26 | Image edge detection method and system facing electric power facilities |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105701835A true CN105701835A (en) | 2016-06-22 |
Family
ID=56223494
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610108566.5A Pending CN105701835A (en) | 2016-02-26 | 2016-02-26 | Image edge detection method and system facing electric power facilities |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105701835A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107491730A (en) * | 2017-07-14 | 2017-12-19 | 浙江大学 | A kind of laboratory test report recognition methods based on image procossing |
| CN108197225A (en) * | 2017-12-28 | 2018-06-22 | 广东欧珀移动通信有限公司 | Sorting technique, device, storage medium and the electronic equipment of image |
| CN111047535A (en) * | 2019-12-17 | 2020-04-21 | 江苏深渡消防装备科技有限公司 | Fire control system capable of intelligently identifying fire source and automatically adjusting fire extinguishing flow |
| CN111047293A (en) * | 2019-12-12 | 2020-04-21 | 云南云电同方科技有限公司 | Method and system for managing graphic data resources |
| CN112381745A (en) * | 2020-11-06 | 2021-02-19 | 山东大学 | Side-scan sonar image data enhancement method and system based on multi-source data synthesis |
| CN112528971A (en) * | 2021-02-07 | 2021-03-19 | 北京智芯微电子科技有限公司 | Power transmission line abnormal target detection method and system based on deep learning |
| CN114792299A (en) * | 2021-01-25 | 2022-07-26 | 山东信通电子股份有限公司 | Method and device for detecting abnormal state of power transmission line |
| CN116468958A (en) * | 2023-05-18 | 2023-07-21 | 中维建技术有限公司 | Communication tower safety detection method and system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101793501A (en) * | 2010-04-14 | 2010-08-04 | 华中科技大学 | Transmission line ice coating status detection method based on image |
| CN102346015A (en) * | 2011-06-24 | 2012-02-08 | 西安工程大学 | Method for measuring icing thickness of insulator of power transmission line based on video difference analysis |
| CN103389043A (en) * | 2012-05-10 | 2013-11-13 | 河南科信电缆有限公司 | Ice coating detecting system of electric power circuit |
| CN103528534A (en) * | 2013-10-25 | 2014-01-22 | 江西省电力公司检修分公司 | Image monitoring based method for detecting thickness of icing on power transmission line |
| US20150206319A1 (en) * | 2014-01-17 | 2015-07-23 | Microsoft Corporation | Digital image edge detection |
-
2016
- 2016-02-26 CN CN201610108566.5A patent/CN105701835A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101793501A (en) * | 2010-04-14 | 2010-08-04 | 华中科技大学 | Transmission line ice coating status detection method based on image |
| CN102346015A (en) * | 2011-06-24 | 2012-02-08 | 西安工程大学 | Method for measuring icing thickness of insulator of power transmission line based on video difference analysis |
| CN103389043A (en) * | 2012-05-10 | 2013-11-13 | 河南科信电缆有限公司 | Ice coating detecting system of electric power circuit |
| CN103528534A (en) * | 2013-10-25 | 2014-01-22 | 江西省电力公司检修分公司 | Image monitoring based method for detecting thickness of icing on power transmission line |
| US20150206319A1 (en) * | 2014-01-17 | 2015-07-23 | Microsoft Corporation | Digital image edge detection |
Non-Patent Citations (2)
| Title |
|---|
| ZHENYU WANG 等: "the ice edge detection of transmission line based on structured forest algorithm", 《PROCEEDINGS OF THE SECOND INTERNATIONAL CONFERENCE ON MECHATRONICS AND AUTOMATIC CONTROL》 * |
| 张爱华 等: "一种基于边缘检测的局部闻值分割算法", 《小型微型计算机系统》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107491730A (en) * | 2017-07-14 | 2017-12-19 | 浙江大学 | A kind of laboratory test report recognition methods based on image procossing |
| CN108197225A (en) * | 2017-12-28 | 2018-06-22 | 广东欧珀移动通信有限公司 | Sorting technique, device, storage medium and the electronic equipment of image |
| CN108197225B (en) * | 2017-12-28 | 2020-12-01 | Oppo广东移动通信有限公司 | Image classification method and device, storage medium and electronic equipment |
| CN111047293A (en) * | 2019-12-12 | 2020-04-21 | 云南云电同方科技有限公司 | Method and system for managing graphic data resources |
| CN111047293B (en) * | 2019-12-12 | 2023-11-03 | 云南云电同方科技有限公司 | Method and system for managing graphic data resources |
| CN111047535A (en) * | 2019-12-17 | 2020-04-21 | 江苏深渡消防装备科技有限公司 | Fire control system capable of intelligently identifying fire source and automatically adjusting fire extinguishing flow |
| CN112381745A (en) * | 2020-11-06 | 2021-02-19 | 山东大学 | Side-scan sonar image data enhancement method and system based on multi-source data synthesis |
| CN112381745B (en) * | 2020-11-06 | 2023-09-22 | 山东大学 | Side-scan sonar image data enhancement method and system based on multi-source data synthesis |
| CN114792299A (en) * | 2021-01-25 | 2022-07-26 | 山东信通电子股份有限公司 | Method and device for detecting abnormal state of power transmission line |
| CN112528971A (en) * | 2021-02-07 | 2021-03-19 | 北京智芯微电子科技有限公司 | Power transmission line abnormal target detection method and system based on deep learning |
| CN116468958A (en) * | 2023-05-18 | 2023-07-21 | 中维建技术有限公司 | Communication tower safety detection method and system |
| CN116468958B (en) * | 2023-05-18 | 2024-04-05 | 中维建技术有限公司 | Communication tower safety detection method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105701835A (en) | Image edge detection method and system facing electric power facilities | |
| CN109035149B (en) | License plate image motion blur removing method based on deep learning | |
| CN109543760B (en) | Confrontation sample detection method based on image filter algorithm | |
| CN107563433B (en) | Infrared small target detection method based on convolutional neural network | |
| US20200117975A1 (en) | Methods and apparatus for preserving information between layers within a neural network | |
| CN110415260B (en) | Smoke image segmentation and identification method based on dictionary and BP neural network | |
| WO2021118463A1 (en) | Defect detection in image space | |
| CN104537359A (en) | Vehicle object detection method and device | |
| Kaur et al. | Performance evaluation of fuzzy and histogram based color image enhancement | |
| CN112733929A (en) | Improved method for detecting small target and shielded target of Yolo underwater image | |
| KR101268520B1 (en) | The apparatus and method for recognizing image | |
| CN113888462A (en) | Crack identification method, system, readable medium and storage medium | |
| Rhee | Detection of spliced image forensics using texture analysis of median filter residual | |
| CN115131590A (en) | Training method of target detection model, target detection method and related equipment | |
| CN105354547A (en) | Pedestrian detection method in combination of texture and color features | |
| Chen et al. | Visual depth guided image rain streaks removal via sparse coding | |
| CN107392127B (en) | Transmission line of electricity extracting method based on Chebyshev polynomials description | |
| CN112990357A (en) | Black box video countermeasure sample generation method based on sparse disturbance | |
| CN112418241A (en) | Power equipment identification method based on infrared imaging technology | |
| CN115564709A (en) | Evaluation method and system for robustness of power algorithm model in confrontation scene | |
| Yorulmaz et al. | Cepstrum based feature extraction method for fungus detection | |
| Ciobanu et al. | Color feature vectors based on optimal LAB histogram bins | |
| CN110717490A (en) | Bird nest identification method and system for power transmission line tower and storage medium | |
| CN106469267A (en) | A kind of identifying code sample collection method and system | |
| JP6060638B2 (en) | Object identification device, learning sample creation device, and program |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160622 |
|
| RJ01 | Rejection of invention patent application after publication |