CN105528595A - Method for identifying and positioning power transmission line insulators in unmanned aerial vehicle aerial images - Google Patents
Method for identifying and positioning power transmission line insulators in unmanned aerial vehicle aerial images Download PDFInfo
- Publication number
- CN105528595A CN105528595A CN201610074633.6A CN201610074633A CN105528595A CN 105528595 A CN105528595 A CN 105528595A CN 201610074633 A CN201610074633 A CN 201610074633A CN 105528595 A CN105528595 A CN 105528595A
- Authority
- CN
- China
- Prior art keywords
- sample
- aerial images
- training
- image
- insulator
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
- G06F18/2155—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the incorporation of unlabelled data, e.g. multiple instance learning [MIL], semi-supervised techniques using expectation-maximisation [EM] or naïve labelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/285—Selection of pattern recognition techniques, e.g. of classifiers in a multi-classifier system
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
- G06V10/443—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
- G06V10/449—Biologically inspired filters, e.g. difference of Gaussians [DoG] or Gabor filters
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Multimedia (AREA)
- Remote Sensing (AREA)
- Astronomy & Astrophysics (AREA)
- Health & Medical Sciences (AREA)
- Biodiversity & Conservation Biology (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Image Analysis (AREA)
Abstract
The invention belongs to the technical field of image processing, discloses a method for identifying and positioning power transmission line insulators in unmanned aerial vehicle aerial images, and solves the problems in the prior art that the detection precision of an identification algorithm of the insulators is not high, the robustness is low, and the identification algorithm is easy to be affected by sample number. A group of Gabor wavelet basis with different sizes and different directions and training sample images are taken as convolutions so as to form a group of characteristic vectors which accurately describe sample image texture characteristics. A random forest machine learning algorithm with a semi-supervised learning mode is used to train sample data sets of the known category and the unknown category so as to obtain an insulator identification model. Through the mode from left to right and from top to bottom, a detection window with the same size as the training sample traverses the input images with different sizes. The detection window combining the identification model detects and positions the positions of the insulators in the input images with different sizes. And finally the accurate positions of the insulators in the input image with the original size are determined by using a non-maximum inhibition method.
Description
Technical field
The present invention relates to a kind of semi-supervised learning random forests algorithm based on Gabor characteristic realize in unmanned plane Aerial Images to the recognition positioning method of electric transmission line isolator, belong to technical field of image processing.
Background technology
Transmission line of electricity is as the important component part in State Grid's system, and its safe condition concerns popular safety.Use maximum power equipments to be insulator in transmission line of electricity, its effect prevents electric current from going back to ground, and its performance quality directly determines the reliability of power transmission system.
The shortcomings such as at present, the detection method of insulator running status mainly contains ultrasonic Detection Method, IR thermometry, pulse current method etc., and these methods exist complicated operation mostly, cost is high, danger is large, climate impact is large.In recent years, along with the development of computer image processing technology and unmanned plane patrol and examine the increasingly mature of technology, isolator detecting method has had new breakthrough.
Because the insulator defective mode testing result based on unmanned plane transmission line of electricity Aerial Images is larger by background influence, problem is complicated, achievement is less, and current research is also in developing stage, and it depends on accurate identification and the location of insulator in transmission line of electricity image to a great extent.Therefore how effectively to distinguish background and insulator, from complex background image, accurately identify that insulator becomes the difficult point of testing.
Many scholars are devoted to research and utilization digital image processing techniques from Aerial Images, extract insulator image in recent years, mainly from edge detection method, based on threshold segmentation method and the Hough transform method location insulator utilizing insulator oblong nature.But because the usual background of Aerial Images is complicated, the method based on rim detection can introduce a large amount of Clutter edge, and accuracy of detection is not high; Utilize insulator color characteristics can detect roughly the target area of doubtful insulator based on threshold segmentation method, but when illumination or change of background, detection algorithm robustness is low; Method based on insulator oblong nature affects by insulator state in image, for blocking, continuously arranged isolator detecting weak effect.
In addition, the popularization of random machine learning techniques, some scholars propose to utilize the convolutional neural networks of supervised learning to carry out the identification of insulator image, and this algorithm relies on infinite many sample numbers, and calculation cost is high; When sample number is less, Detection accuracy is low.
Summary of the invention
Technical matters to be solved by this invention is: propose a kind of in unmanned plane Aerial Images the recognition positioning method to electric transmission line isolator, solve not high on the recognizer accuracy of detection of insulator in conventional art, robustness is low, the problem that easily affects by sample number.
The present invention solves the problems of the technologies described above adopted scheme:
To the recognition positioning method of electric transmission line isolator in unmanned plane Aerial Images, comprise the following steps:
A, structure sample training data set;
B, employing semi-supervised learning random forests algorithm are trained the sample training data set in steps A, obtain insulator model of cognition;
The insulator model of cognition of acquisition is trained to identify electric transmission line isolator in unmanned plane Aerial Images and locate in C, employing step B.
As further optimization, in steps A, the method for described structure sample training data set specifically comprises:
Unmanned plane Aerial Images is converted to gray level image, from gray level image, extracts the sample image for training, sample drawn image Gabor characteristic composition training dataset.
As further optimization, consistent for ensureing the Gabor characteristic dimension that different sample image extracts, same size to be converted into by unified for all sample images after having extracted training sample image, and the image of known class in the training sample image extracted (be insulator and be not insulator two class) be divided into the sample image of tape label and the sample image (sample class the unknown) of not tape label.
As further optimization, when extracting sample image Gabor characteristic, different directions and yardstick Gabor core is adopted to form one group of Gabor wavelet base, convolution is done respectively with sample image, obtain multiple eigenmatrix identical with sample size, again by these eigenmatrixes by being connected between the column and the column, form a multidimensional characteristic vector and be used for describing this sample image; Finally respectively the eigenvector correspondence of the eigenvector of tape label sample and not tape label sample is referred to tape label sample training data set and not tape label sample training data centralization.
As further optimization, in step B, described employing semi-supervised learning random forests algorithm is trained the sample training data set in steps A, and the concrete grammar obtaining insulator model of cognition comprises:
B1, training parameter initialization: characteristic attribute number, decision tree nodes split criterion needed for sample number, decision tree contained by the selection mode of contained decision tree number n, every decision tree training data subset used and data subset in the confidence threshold value θ of setting update mark sample set, random forest;
B2, training is carried out to tape label sample training data set obtain initial random forest classified device: to be concentrated in tape label sample data by certain sampling mode and extract n data subset, and carry out training by certain extraction algorithm selection characteristic attribute subset the random forest sorter obtained containing n decision tree: H={h
1, h
2..., h
n;
B3, to without exemplar data set prediction, upgrade each data subset: for each decision tree h in sorter set H
i, the sorter set that all the other n-1 decision tree forms, is called h
ireciprocity sorter set, be denoted as H
i; Use reciprocity sorter set H
ito without label data collection X
uin the sample mode that performs most ballot predict, use the consistance of label to represent the degree of confidence of sample, select degree of confidence be greater than default threshold θ without exemplar, copy to sorter h
idata subset in, realize each decision tree training set upgrade;
The data subset that B4, use upgrade trains corresponding decision tree, successively until model tends towards stability can obtain final random forest sorter model as insulator model of cognition.
As further optimization, in step C, train the insulator model of cognition of acquisition identify electric transmission line isolator in unmanned plane Aerial Images and locate in described employing step B, specifically comprise:
C1, input Aerial Images is converted to gray-scale map and obtains corresponding multiple dimensioned figure;
C2, with sample image detection window of the same size from left to right, travel through the Aerial Images under different scale from top to bottom, obtain image block;
C3, extract the Gabor characteristic of image block, send into insulator model of cognition and identify, mark and preserve the center point coordinate of the insulator candidate frame recognized;
C4, repetition step C2 and C3, until the Aerial Images under all yardsticks has all traveled through;
C5, the insulator candidate frame detected under different scale is transformed into original scale according to the change of scale factor under, adopt non-maxima suppression finally lock and mark the insulator in Aerial Images.
The invention has the beneficial effects as follows:
1) Two-Dimensional Gabor Wavelets is to neuronic good approximation in level vertebrate animals vision flat bed, feature based on Gabor filter is comparatively close to the feature that eye-observation arrives, some parameter can be extracted as its textural characteristics according to the different frequency of Gabor filter and direction to image filtering result, thus reach the object of texture analysis; The present invention extracts the feature of insulator on different scale, different directions by Gabor function, can describe the insulator of the different attitudes that unmanned plane obtains under difference shooting visual angle well, improve the discrimination of insulator further.
2) machine learning algorithm of semi-supervised learning random forest is adopted, randomness is introduced in the training process by obtaining data subset with putting back to and randomly drawing characteristic attribute ground mode, make to keep independence between every decision tree, and it can not carry out in the dimension-reduction treatment situation such as feature selecting or feature deletion, directly carry out the process of extensive high dimensional data, and there is good performance.In addition, the random forests algorithm of semi-supervised learning, tape label sample training is utilized to obtain sorter, then predict without label data, sample high for degree of confidence is joined in training set, then new training data is utilized to re-start the training of sorter, by having label and without label two class data set and comprehensive utilization, overcome the problem of the blindness of unsupervised learning and the data category mark difficulty of supervised learning, effectively can improve classification effectiveness by the parallel anticipation between decision tree.
Accompanying drawing explanation
Fig. 1 is Aerial Images isolator detecting identification process figure under single yardstick;
Fig. 2 is that data subset upgrades process flow diagram.
Embodiment
The present invention be intended to propose a kind of in unmanned plane Aerial Images to the recognition positioning method of electric transmission line isolator, solve not high on the recognizer accuracy of detection of insulator in conventional art, robustness is low, the problem that easily affects by sample number.
Because the angle of insulator in unmanned plane Aerial Images, size can take the difference of visual angle and height and different along with unmanned plane, traditional Feature Descriptor cannot adapt to the isolator detecting of different angles.The present invention by extracting the Gabor characteristic in insulator image different scale and direction, make the proper vector of extraction to insulator yardstick and angular transformation insensitive, be applicable to the isolator detecting of different attitude.In addition, the present invention adopts the random forests algorithm of semi-supervised learning, and compared to supervised learning, the present invention effectively can reduce the workload preparing tape label sample; Compared to unsupervised learning, energy effective prevention model training process of the present invention, do not need infinite multiple sample, calculation cost is relatively low.
The present invention utilizes one group of Gabor wavelet base of different scale and different directions and training sample image to do convolution, forms a stack features vector with accurate description sample image textural characteristics; Adopt the random forest machine learning algorithm training known class of semi-supervised learning mode and the sample data collection of unknown classification, obtain insulator model of cognition.By from left to right mode from top to bottom, with the input picture traveled through with training sample detection window of the same size under different scale, and in conjunction with the position of insulator in the input picture under model of cognition detection and positioning different scale; Recycling non-maxima suppression method finally determines the accurate location of insulator under input picture original scale.
In specific implementation, as follows to the recognition positioning method step of electric transmission line isolator in the present invention:
Step one: prepare training dataset: unmanned plane Aerial Images being converted to gray level image, extracting the sample image for training from gray level image, sample drawn image Gabor characteristic composition training dataset; Consistent for ensureing the Gabor characteristic dimension that different sample image extracts, same size to be converted into by unified for all sample images after having extracted training sample image, the positive negative sample of part known class (be insulator and be not insulator two class) in these samples, should be comprised; When extracting sample image Gabor characteristic, we form one group of Gabor wavelet base with different directions and yardstick Gabor core, convolution is done respectively with sample image, obtain eigenmatrix that is multiple and sample size, then by these eigenmatrixes by the composition multidimensional characteristic vector that is connected between the column and the column to describe sample image.The eigenvector of all sample images can form training dataset.
If I (x, y) is a width sample image, be exactly a convolution algorithm with Gabor base to the process that it carries out feature extraction, that is:
O
μ,v(z)＝I(z)*ψ
μ,v(z)
Wherein, z is image coordinate (x, y), ψ
u,vz () represents Gabor wavelet kernel function, its representation is as follows:
Wherein,
k
maxrepresent maximum sample frequency.μ and v represents direction and the scale factor of Gabor wavelet respectively, and μ, v are integer, and σ is the width of Gauss's window and the ratio of sinusoidal wave wavelength.
Now again by the multidimensional characteristic vector (M size and μ, v, sample image size relevant) of the image characteristic matrix obtained after these convolution by the composition 1*M that is connected between the column and the column:
χ＝(O
0.0 T,O
0,1 T,...,O
μ,v T)
Step 2: utilize the data of semi-supervised learning random forests algorithm to step one data centralization to train, obtain insulator model of cognition.
1) training parameter initialization: characteristic attribute number, decision tree nodes split criterion needed for sample number, decision tree contained by the selection mode of contained decision tree number n, every decision tree training data subset used and data subset in the confidence threshold value θ of setting update mark sample set, random forest.
2) training is carried out to tape label data set and obtain initial random forest classified device: extract n data subset by certain sampling mode in tape label data centralization, and select characteristic attribute subset to carry out training the random forest sorter obtained containing n decision tree with certain extraction algorithm: H={h
1, h
2..., h
n.
3) without the prediction of label data collection, each data subset is upgraded.For each decision tree h in sorter set H
i, the sorter set that all the other n-1 decision tree forms, is called h
ireciprocity sorter set, be denoted as H
i.Use reciprocity sorter set H
ito without label data collection X
uin the sample mode that performs most ballot predict, use the consistance of label to represent the degree of confidence of sample, select degree of confidence be greater than default threshold θ without exemplar, copy to sorter h
idata subset in, realize each decision tree training set upgrade.With sorter h
1for example, its data subset renewal process as shown in Figure 2.
4) data subset upgraded is used to train corresponding decision tree successively, until model tends towards stability can obtain final random forest sorter model.
Step 3: the identification of Aerial Images insulator and location: insulator in unmanned plane Aerial Images is detected and located with the insulator model of cognition trained.First Aerial Images is converted to gray-scale map.For avoiding unmanned plane distance of taking photo by plane, insulator recognition accuracy is impacted, we detect from multiple yardstick insulator unmanned plane Aerial Images, and under considering different scale isolator detecting result adopt non-maxima suppression algorithm finally to determine go forward side by side row labels in the position of insulator in original image.Detailed process is as follows:
1) input Aerial Images be converted to gray-scale map and obtain corresponding multiple dimensioned figure.
2) with same sample image detection window of the same size from left to right, travel through the Aerial Images under different scale from top to bottom, obtain image block.
3) extract the Gabor characteristic of image block, send into random forest sorter and identify, mark and preserve the center point coordinate of the insulator candidate frame recognized.
4) step 2 is repeated) and 3), until the Aerial Images under all yardsticks has all traveled through.
5), under the insulator candidate frame detected under different scale being transformed into original scale according to the change of scale factor, non-maxima suppression is adopted finally to lock and mark the insulator in Aerial Images.
Isolator detecting identification process is as shown in Figure 1 under single yardstick for the Aerial Images realized based on this recognition positioning method.
Do further to describe to the present invention below in conjunction with embodiment:
One, training dataset is prepared:
1) unmanned plane Aerial Images is converted into gray level image by brightness calculation formula.First coloured image is divided into R (red), G (green), B (indigo plant) three components, then the gray-scale value of pixel can be calculated by following formula: I=0.11R+0.59G+0.3B.
2) from a large amount of unmanned plane Aerial Images, insulator for training and background area is extracted as sample image (containing tape label and not tape label two class), and by unified for all samples to the same size, as 64 pixel * 128 pixels;
3) sample image and one group of Two-Dimensional Gabor Wavelets kernel function are carried out convolution algorithm, obtain the sample image feature of different scale different directions:
The implementation case is got
get 8 directions and 5 yardstick totally 40 Gabor wavelet bases, i.e. μ=0,1,2 ..., 7, v=0,1,2 ..., 4, get σ=2 π.
4) by the eigenvector of the image obtained after convolution by the composition 1*327680 size that is connected between the column and the column:
χ＝(O
0.0 T,O
0,1 T,...,O
4,7 T)
The eigenvector of all sample images can form data set X.
5) eigenvector of tape label sample in data set X added class label and be divided into tape label data set X
lin, by data set X not the eigenvector of tape label sample add without label data collection X
uin, final formation tape label data set X
lnot tape label data set X
utwo class data sets.
Two, utilize semi-supervised learning algorithm to train training dataset, obtain insulator model of cognition.Concrete steps are as follows:
1) training parameter initialization: defining classification model default threshold θ=0.6, the unmarked sample being greater than this threshold value just joins in tag set.The decision tree number of random forests algorithm is n=10, and the number of the characteristic attribute subset of decision tree is log
2m+1, wherein M represents the characteristic attribute number (M=327680) of data set, and it is have the bagging methods of sampling put back to and CART algorithm to carry out node split at random that data subset extracts mode.
2) training is carried out to tape label data set and obtain initial random forest classified device: by bagging method at tape label data set X
lmiddle extraction 10 data subsets, and in the characteristic attribute of data subset, Stochastic choice attribute set carries out training the random forest sorter obtained containing 10 decision trees: H={h
1, h
2..., h
10.
3) without the prediction of label data collection, each data subset is upgraded.Use reciprocity sorter set H successively
i(i=1,2 ..., 10) and to without label data collection X
lin the sample mode that performs most ballot predict, degree of confidence is greater than 0.6 without exemplar, copy to sorter h
idata subset in.
4) data subset upgraded is used to train corresponding decision tree successively, until model tends towards stability can obtain final random forest sorter model as insulator model of cognition.
Three, to insulator identification and location in Aerial Images:
1) input Aerial Images be converted to gray-scale map and obtain the image (changeable scale) be of a size of under 100%, 80%, 60% and 50% 4 yardstick of former figure.
2) with the detection window of 64 pixel * 128 pixel sizes from left to right, travel through the Aerial Images under different scale from top to bottom, obtain image block.
3) extract the Gabor characteristic of image block, send into random forest sorter and identify, mark and preserve the center point coordinate of the insulator candidate frame recognized.
4) step 2 is repeated) and 3), until the Aerial Images under four yardsticks has all traveled through.
5), under the insulator candidate frame detected under different scale being transformed into original scale according to the change of scale factor, non-maxima suppression is adopted finally to lock and mark the insulator in Aerial Images.
Claims (6)
1. in unmanned plane Aerial Images to the recognition positioning method of electric transmission line isolator, it is characterized in that, comprise the following steps:
A, structure sample training data set;
B, employing semi-supervised learning random forests algorithm are trained the sample training data set in steps A, obtain insulator model of cognition;
The insulator model of cognition of acquisition is trained to identify electric transmission line isolator in unmanned plane Aerial Images and locate in C, employing step B.
2. as claimed in claim 1 in unmanned plane Aerial Images to the recognition positioning method of electric transmission line isolator, it is characterized in that, in steps A, the method for described structure sample training data set specifically comprises:
Unmanned plane Aerial Images is converted to gray level image, from gray level image, extracts the sample image for training, sample drawn image Gabor characteristic composition training dataset.
3. as claimed in claim 2 in unmanned plane Aerial Images to the recognition positioning method of electric transmission line isolator, it is characterized in that, same size to be converted into by unified for all sample images after having extracted training sample image, and the image of known class in the training sample image of extraction be divided into tape label sample image and not tape label sample image.
4. as claimed in claim 3 in unmanned plane Aerial Images to the recognition positioning method of electric transmission line isolator, it is characterized in that, when extracting sample image Gabor characteristic, different directions and yardstick Gabor core is adopted to form one group of Gabor wavelet base, convolution is done respectively with sample image, obtain multiple eigenmatrix identical with sample size, then by these eigenmatrixes by being connected between the column and the column, forming a multidimensional characteristic vector and being used for describing this sample image; Finally respectively the eigenvector correspondence of the eigenvector of the sample of tape label and the not sample of tape label is referred to tape label sample training data set and not tape label sample training data centralization.
5. as claimed in claim 4 in unmanned plane Aerial Images to the recognition positioning method of electric transmission line isolator, it is characterized in that, in step B, described employing semi-supervised learning random forests algorithm is trained the sample training data set in steps A, and the concrete grammar obtaining insulator model of cognition comprises:
B1, training parameter initialization: characteristic attribute number, decision tree nodes split criterion needed for sample number, decision tree contained by the selection mode of contained decision tree number n, every decision tree training data subset used and data subset in the confidence threshold value θ of setting update mark sample set, random forest;
B2, training is carried out to tape label positive sample data collection obtain initial random forest classified device: to be concentrated in tape label sample data by certain sampling mode and extract n data subset, and carry out training by certain extraction algorithm selection characteristic attribute subset the random forest sorter obtained containing n decision tree: H={h
1, h
2..., h
n;
B3, to without exemplar data set prediction, upgrade each data subset: for each decision tree h in sorter set H
i, the sorter set that all the other n-1 decision tree forms, is called h
ireciprocity sorter set, be denoted as H
i; Use reciprocity sorter set H
ito without label data collection X
uin the sample mode that performs most ballot predict, use the consistance of label to represent the degree of confidence of sample, select degree of confidence be greater than default threshold θ without exemplar, copy to sorter h
idata subset in, realize each decision tree training set upgrade;
The data subset that B4, use upgrade trains corresponding decision tree, successively until model tends towards stability can obtain final random forest sorter model as insulator model of cognition.
6. as claimed in claim 5 in unmanned plane Aerial Images to the recognition positioning method of electric transmission line isolator, it is characterized in that, in step C, train the insulator model of cognition of acquisition identify electric transmission line isolator in unmanned plane Aerial Images and locate in described employing step B, specifically comprise:
C1, input Aerial Images is converted to gray-scale map and obtains corresponding multiple dimensioned figure;
C2, with sample image detection window of the same size from left to right, travel through the Aerial Images under different scale from top to bottom, obtain image block;
C3, extract the Gabor characteristic of image block, send into insulator model of cognition and identify, mark and preserve the center point coordinate of the insulator candidate frame recognized;
C4, repetition step C2 and C3, until the Aerial Images under all yardsticks has all traveled through;
C5, the insulator candidate frame detected under different scale is transformed into original scale according to the change of scale factor under, adopt non-maxima suppression finally lock and mark the insulator in Aerial Images.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610074633.6A CN105528595A (en) | 2016-02-01 | 2016-02-01 | Method for identifying and positioning power transmission line insulators in unmanned aerial vehicle aerial images |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610074633.6A CN105528595A (en) | 2016-02-01 | 2016-02-01 | Method for identifying and positioning power transmission line insulators in unmanned aerial vehicle aerial images |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105528595A true CN105528595A (en) | 2016-04-27 |
Family
ID=55770809
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610074633.6A Pending CN105528595A (en) | 2016-02-01 | 2016-02-01 | Method for identifying and positioning power transmission line insulators in unmanned aerial vehicle aerial images |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105528595A (en) |
Cited By (28)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105976368A (en) * | 2016-04-28 | 2016-09-28 | 华北电力大学（保定） | Insulator positioning method |
| CN106203396A (en) * | 2016-07-25 | 2016-12-07 | 南京信息工程大学 | Aerial Images object detection method based on degree of depth convolution and gradient rotational invariance |
| CN106250920A (en) * | 2016-07-26 | 2016-12-21 | 国网福建省电力有限公司 | The insulator state detection merged based on multicharacteristic information and diagnostic method |
| CN106290388A (en) * | 2016-08-03 | 2017-01-04 | 国网山东省电力公司电力科学研究院 | A kind of insulator breakdown automatic testing method |
| CN106326932A (en) * | 2016-08-25 | 2017-01-11 | 北京每刻风物科技有限公司 | Power line inspection image automatic identification method based on neural network and power line inspection image automatic identification device thereof |
| CN106375378A (en) * | 2016-08-25 | 2017-02-01 | 北京每刻风物科技有限公司 | Application deployment method and system based on local area network client server structure |
| CN106919932A (en) * | 2017-03-13 | 2017-07-04 | 华北电力大学（保定） | A kind of insulator of " simulation is true " parallel construction positions soft recognition methods |
| CN106960178A (en) * | 2017-02-23 | 2017-07-18 | 中国科学院自动化研究所 | The training method of insulator identification model and the identification of insulator and localization method |
| CN107145846A (en) * | 2017-04-26 | 2017-09-08 | 贵州电网有限责任公司输电运行检修分公司 | A kind of insulator recognition methods based on deep learning |
| CN107358142A (en) * | 2017-05-15 | 2017-11-17 | 西安电子科技大学 | Polarimetric SAR Image semisupervised classification method based on random forest composition |
| CN107403199A (en) * | 2017-08-07 | 2017-11-28 | 北京京东尚科信息技术有限公司 | Data processing method and device |
| CN107679469A (en) * | 2017-09-22 | 2018-02-09 | 东南大学—无锡集成电路技术研究所 | A kind of non-maxima suppression method based on deep learning |
| CN108205805A (en) * | 2016-12-20 | 2018-06-26 | 北京大学 | The dense corresponding auto-creating method of voxel between pyramidal CT image |
| CN108230296A (en) * | 2017-11-30 | 2018-06-29 | 腾讯科技（深圳）有限公司 | The recognition methods of characteristics of image and device, storage medium, electronic device |
| CN108805084A (en) * | 2018-06-14 | 2018-11-13 | 北京中飞艾维航空科技有限公司 | Image-recognizing method, device and server |
| CN109118479A (en) * | 2018-07-26 | 2019-01-01 | 中睿能源（北京）有限公司 | Defects of insulator identification positioning device and method based on capsule network |
| CN109299732A (en) * | 2018-09-12 | 2019-02-01 | 北京三快在线科技有限公司 | The method, apparatus and electronic equipment of unmanned behaviour decision making and model training |
| CN109785288A (en) * | 2018-12-17 | 2019-05-21 | 广东电网有限责任公司 | Transmission facility defect inspection method and system based on deep learning |
| CN109813276A (en) * | 2018-12-19 | 2019-05-28 | 五邑大学 | A kind of antenna for base station has a down dip angle measuring method and its system |
| CN110794861A (en) * | 2019-11-14 | 2020-02-14 | 国网山东省电力公司电力科学研究院 | Autonomous string falling method and system for flying on-line and off-line insulator string detection robot |
| CN110942805A (en) * | 2019-12-11 | 2020-03-31 | 云南大学 | Insulator element prediction system based on semi-supervised deep learning |
| CN111126381A (en) * | 2019-12-03 | 2020-05-08 | 浙江大学 | Insulator inclined positioning and identifying method based on R-DFPN algorithm |
| CN111464995A (en) * | 2019-01-18 | 2020-07-28 | 华为技术有限公司 | Label management method and device for terminal equipment |
| CN111523597A (en) * | 2020-04-23 | 2020-08-11 | 北京百度网讯科技有限公司 | Target recognition model training method, device, equipment and storage medium |
| CN111832328A (en) * | 2019-04-15 | 2020-10-27 | 北京京东尚科信息技术有限公司 | Bar code detection method, bar code detection device, electronic equipment and medium |
| CN112163998A (en) * | 2020-09-24 | 2021-01-01 | 肇庆市博士芯电子科技有限公司 | Single-image super-resolution analysis method matched with natural degradation conditions |
| CN112200178A (en) * | 2020-09-01 | 2021-01-08 | 广西大学 | Transformer substation insulator infrared image detection method based on artificial intelligence |
| CN113033489A (en) * | 2021-04-23 | 2021-06-25 | 华北电力大学 | Power transmission line insulator identification and positioning method based on lightweight deep learning algorithm |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080144941A1 (en) * | 2006-12-18 | 2008-06-19 | Sony Corporation | Face recognition apparatus, face recognition method, gabor filter application apparatus, and computer program |
| CN102937694A (en) * | 2012-11-27 | 2013-02-20 | 广州供电局有限公司 | Device for monitoring external insulation strength of dirty insulator |
| CN104978580A (en) * | 2015-06-15 | 2015-10-14 | 国网山东省电力公司电力科学研究院 | Insulator identification method for unmanned aerial vehicle polling electric transmission line |
| CN104992148A (en) * | 2015-06-18 | 2015-10-21 | 江南大学 | ATM terminal human face key points partially shielding detection method based on random forest |
-
2016
- 2016-02-01 CN CN201610074633.6A patent/CN105528595A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080144941A1 (en) * | 2006-12-18 | 2008-06-19 | Sony Corporation | Face recognition apparatus, face recognition method, gabor filter application apparatus, and computer program |
| CN102937694A (en) * | 2012-11-27 | 2013-02-20 | 广州供电局有限公司 | Device for monitoring external insulation strength of dirty insulator |
| CN104978580A (en) * | 2015-06-15 | 2015-10-14 | 国网山东省电力公司电力科学研究院 | Insulator identification method for unmanned aerial vehicle polling electric transmission line |
| CN104992148A (en) * | 2015-06-18 | 2015-10-21 | 江南大学 | ATM terminal human face key points partially shielding detection method based on random forest |
Non-Patent Citations (2)
| Title |
|---|
| 刘孝良: "基于半监督学习的随机森林算法研究与应用", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
| 李建平: "《非常规小波变换与军事生物信息安全》", 30 November 2008, 成都：电子科技大学出版社 * |
Cited By (44)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105976368A (en) * | 2016-04-28 | 2016-09-28 | 华北电力大学（保定） | Insulator positioning method |
| CN105976368B (en) * | 2016-04-28 | 2018-12-11 | 华北电力大学（保定） | A kind of insulator localization method |
| CN106203396A (en) * | 2016-07-25 | 2016-12-07 | 南京信息工程大学 | Aerial Images object detection method based on degree of depth convolution and gradient rotational invariance |
| CN106203396B (en) * | 2016-07-25 | 2019-05-10 | 南京信息工程大学 | Aerial Images object detection method based on depth convolution sum gradient rotational invariance |
| CN106250920A (en) * | 2016-07-26 | 2016-12-21 | 国网福建省电力有限公司 | The insulator state detection merged based on multicharacteristic information and diagnostic method |
| CN106290388A (en) * | 2016-08-03 | 2017-01-04 | 国网山东省电力公司电力科学研究院 | A kind of insulator breakdown automatic testing method |
| CN106290388B (en) * | 2016-08-03 | 2018-09-28 | 国网山东省电力公司电力科学研究院 | A kind of insulator breakdown automatic testing method |
| CN106326932A (en) * | 2016-08-25 | 2017-01-11 | 北京每刻风物科技有限公司 | Power line inspection image automatic identification method based on neural network and power line inspection image automatic identification device thereof |
| CN106375378A (en) * | 2016-08-25 | 2017-02-01 | 北京每刻风物科技有限公司 | Application deployment method and system based on local area network client server structure |
| CN108205805A (en) * | 2016-12-20 | 2018-06-26 | 北京大学 | The dense corresponding auto-creating method of voxel between pyramidal CT image |
| CN108205805B (en) * | 2016-12-20 | 2020-06-02 | 北京大学 | Automatic establishment method for dense correspondence of cone beam CT image intermediate pixels |
| CN106960178A (en) * | 2017-02-23 | 2017-07-18 | 中国科学院自动化研究所 | The training method of insulator identification model and the identification of insulator and localization method |
| CN106960178B (en) * | 2017-02-23 | 2020-02-07 | 中国科学院自动化研究所 | Training method of insulator recognition model and insulator recognition and positioning method |
| CN106919932A (en) * | 2017-03-13 | 2017-07-04 | 华北电力大学（保定） | A kind of insulator of " simulation is true " parallel construction positions soft recognition methods |
| CN107145846A (en) * | 2017-04-26 | 2017-09-08 | 贵州电网有限责任公司输电运行检修分公司 | A kind of insulator recognition methods based on deep learning |
| CN107145846B (en) * | 2017-04-26 | 2018-10-19 | 贵州电网有限责任公司输电运行检修分公司 | A kind of insulator recognition methods based on deep learning |
| CN107358142B (en) * | 2017-05-15 | 2020-12-08 | 西安电子科技大学 | Polarized SAR image semi-supervised classification method based on random forest composition |
| CN107358142A (en) * | 2017-05-15 | 2017-11-17 | 西安电子科技大学 | Polarimetric SAR Image semisupervised classification method based on random forest composition |
| CN107403199B (en) * | 2017-08-07 | 2021-02-26 | 北京京东尚科信息技术有限公司 | Data processing method and device |
| CN107403199A (en) * | 2017-08-07 | 2017-11-28 | 北京京东尚科信息技术有限公司 | Data processing method and device |
| CN107679469B (en) * | 2017-09-22 | 2021-03-30 | 东南大学—无锡集成电路技术研究所 | Non-maximum suppression method based on deep learning |
| CN107679469A (en) * | 2017-09-22 | 2018-02-09 | 东南大学—无锡集成电路技术研究所 | A kind of non-maxima suppression method based on deep learning |
| US11526983B2 (en) | 2017-11-30 | 2022-12-13 | Tencent Technology (Shenzhen) Company Limited | Image feature recognition method and apparatus, storage medium, and electronic apparatus |
| CN110349156A (en) * | 2017-11-30 | 2019-10-18 | 腾讯科技（深圳）有限公司 | The recognition methods of lesion characteristics and device, storage medium in the picture of eyeground |
| CN110349156B (en) * | 2017-11-30 | 2023-05-30 | 腾讯科技（深圳）有限公司 | Method and device for identifying lesion characteristics in fundus picture and storage medium |
| CN108230296B (en) * | 2017-11-30 | 2023-04-07 | 腾讯科技（深圳）有限公司 | Image feature recognition method and device, storage medium and electronic device |
| CN108230296A (en) * | 2017-11-30 | 2018-06-29 | 腾讯科技（深圳）有限公司 | The recognition methods of characteristics of image and device, storage medium, electronic device |
| CN108805084A (en) * | 2018-06-14 | 2018-11-13 | 北京中飞艾维航空科技有限公司 | Image-recognizing method, device and server |
| CN109118479A (en) * | 2018-07-26 | 2019-01-01 | 中睿能源（北京）有限公司 | Defects of insulator identification positioning device and method based on capsule network |
| CN109299732A (en) * | 2018-09-12 | 2019-02-01 | 北京三快在线科技有限公司 | The method, apparatus and electronic equipment of unmanned behaviour decision making and model training |
| CN109785288A (en) * | 2018-12-17 | 2019-05-21 | 广东电网有限责任公司 | Transmission facility defect inspection method and system based on deep learning |
| CN109813276A (en) * | 2018-12-19 | 2019-05-28 | 五邑大学 | A kind of antenna for base station has a down dip angle measuring method and its system |
| CN111464995A (en) * | 2019-01-18 | 2020-07-28 | 华为技术有限公司 | Label management method and device for terminal equipment |
| CN111832328A (en) * | 2019-04-15 | 2020-10-27 | 北京京东尚科信息技术有限公司 | Bar code detection method, bar code detection device, electronic equipment and medium |
| CN110794861A (en) * | 2019-11-14 | 2020-02-14 | 国网山东省电力公司电力科学研究院 | Autonomous string falling method and system for flying on-line and off-line insulator string detection robot |
| CN111126381A (en) * | 2019-12-03 | 2020-05-08 | 浙江大学 | Insulator inclined positioning and identifying method based on R-DFPN algorithm |
| CN110942805A (en) * | 2019-12-11 | 2020-03-31 | 云南大学 | Insulator element prediction system based on semi-supervised deep learning |
| CN111523597A (en) * | 2020-04-23 | 2020-08-11 | 北京百度网讯科技有限公司 | Target recognition model training method, device, equipment and storage medium |
| CN111523597B (en) * | 2020-04-23 | 2023-08-25 | 北京百度网讯科技有限公司 | Target recognition model training method, device, equipment and storage medium |
| CN112200178A (en) * | 2020-09-01 | 2021-01-08 | 广西大学 | Transformer substation insulator infrared image detection method based on artificial intelligence |
| CN112200178B (en) * | 2020-09-01 | 2022-10-11 | 广西大学 | Transformer substation insulator infrared image detection method based on artificial intelligence |
| CN112163998A (en) * | 2020-09-24 | 2021-01-01 | 肇庆市博士芯电子科技有限公司 | Single-image super-resolution analysis method matched with natural degradation conditions |
| CN113033489A (en) * | 2021-04-23 | 2021-06-25 | 华北电力大学 | Power transmission line insulator identification and positioning method based on lightweight deep learning algorithm |
| CN113033489B (en) * | 2021-04-23 | 2024-02-13 | 华北电力大学 | Power transmission line insulator identification positioning method based on lightweight deep learning algorithm |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105528595A (en) | Method for identifying and positioning power transmission line insulators in unmanned aerial vehicle aerial images | |
| CN106778835B (en) | Remote sensing image airport target identification method fusing scene information and depth features | |
| CN103400151B (en) | The optical remote sensing image of integration and GIS autoregistration and Clean water withdraw method | |
| CN106709486A (en) | Automatic license plate identification method based on deep convolutional neural network | |
| Zhang et al. | CDNet: A real-time and robust crosswalk detection network on Jetson nano based on YOLOv5 | |
| CN105469047A (en) | Chinese detection method based on unsupervised learning and deep learning network and system thereof | |
| CN105354568A (en) | Convolutional neural network based vehicle logo identification method | |
| Tao et al. | Scene context-driven vehicle detection in high-resolution aerial images | |
| CN105404886A (en) | Feature model generating method and feature model generating device | |
| CN111898432B (en) | Pedestrian detection system and method based on improved YOLOv3 algorithm | |
| Zhang et al. | Road recognition from remote sensing imagery using incremental learning | |
| CN103310195A (en) | LLC-feature-based weak-supervision recognition method for vehicle high-resolution remote sensing images | |
| CN113963222B (en) | High-resolution remote sensing image change detection method based on multi-strategy combination | |
| CN104598885A (en) | Method for detecting and locating text sign in street view image | |
| CN108647695A (en) | Soft image conspicuousness detection method based on covariance convolutional neural networks | |
| CN111507296A (en) | Intelligent illegal building extraction method based on unmanned aerial vehicle remote sensing and deep learning | |
| CN104657717A (en) | Pedestrian detection method based on layered kernel sparse representation | |
| CN104732244A (en) | Wavelet transform, multi-strategy PSO (particle swarm optimization) and SVM (support vector machine) integrated based remote sensing image classification method | |
| CN106503748A (en) | A kind of based on S SIFT features and the vehicle targets of SVM training aids | |
| CN102663401A (en) | Image characteristic extracting and describing method | |
| CN111460881A (en) | Traffic sign countermeasure sample detection method and classification device based on neighbor discrimination | |
| CN103279738A (en) | Automatic identification method and system for vehicle logo | |
| CN112733736A (en) | Class imbalance hyperspectral image classification method based on enhanced oversampling | |
| CN105404858A (en) | Vehicle type recognition method based on deep Fisher network | |
| Chen et al. | Research on fast recognition method of complex sorting images based on deep learning |
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: 20160427 |
|
| RJ01 | Rejection of invention patent application after publication |