CN110796206A - Data enhancement method and device for partial discharge map - Google Patents
Data enhancement method and device for partial discharge map Download PDFInfo
- Publication number
- CN110796206A CN110796206A CN201911074459.5A CN201911074459A CN110796206A CN 110796206 A CN110796206 A CN 110796206A CN 201911074459 A CN201911074459 A CN 201911074459A CN 110796206 A CN110796206 A CN 110796206A
- Authority
- CN
- China
- Prior art keywords
- data
- module
- weight
- noise
- picture
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 44
- 230000008878 coupling Effects 0.000 claims abstract description 48
- 238000010168 coupling process Methods 0.000 claims abstract description 48
- 238000005859 coupling reaction Methods 0.000 claims abstract description 48
- 238000012549 training Methods 0.000 claims abstract description 20
- NINIDFKCEFEMDL-UHFFFAOYSA-N Sulfur Chemical compound [S] NINIDFKCEFEMDL-UHFFFAOYSA-N 0.000 claims abstract description 8
- 229910052717 sulfur Inorganic materials 0.000 claims abstract description 8
- 239000011593 sulfur Substances 0.000 claims abstract description 8
- 239000011159 matrix material Substances 0.000 claims description 51
- 238000006243 chemical reaction Methods 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 11
- 238000005315 distribution function Methods 0.000 claims description 10
- 238000000605 extraction Methods 0.000 claims description 9
- 238000013135 deep learning Methods 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 4
- 239000000284 extract Substances 0.000 claims description 2
- 238000004364 calculation method Methods 0.000 abstract description 3
- 238000001228 spectrum Methods 0.000 description 6
- 238000010801 machine learning Methods 0.000 description 2
- 230000003321 amplification Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Images
Classifications
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Processing (AREA)
Abstract
The invention provides a data enhancement method and a data enhancement device for a partial discharge map, which apply a Gaussian fuzzy method on the basis of adding on-site common noises such as mobile phone interference, radar interference and microwave sulfur lamp interference to randomly disturb RGB pixels so as to train a network with stronger generalization capability. The partial discharge atlas is preprocessed by methods of background noise coupling, Gaussian blur and the like, so that a plurality of atlases can be generated by one atlas, a large number of labeled samples are generated with less calculation amount, and the problems of high acquisition cost and insufficient training data of the labeled samples are solved; common interference on site is considered on the basis of an original data sample, real site data are simulated, and the anti-interference capability of a training model is improved; the diversity of the training data is expanded, the model is trained by the expanded training data, and overfitting of the model is avoided.
Description
Technical Field
The invention relates to the field of partial discharge detection, in particular to a data enhancement method aiming at a partial discharge map.
Background
With the development of smart grids, the demand for data diagnosis is increasing. Machine learning diagnostics is a solution to replace manual diagnostics. When a machine learning model is set up to train and verify data, how to quickly, efficiently and conveniently obtain a large number of labeled training samples is an urgent problem to be solved.
Partial discharge map recognition requires a large number of labeled training samples. The collection process of the labeled samples is costly, and it is often difficult to quickly build up sufficient numbers of samples to support training for the low frequency partial discharge phenomena. In a conventional partial discharge data enhancement method, simple processing modes such as translation, rotation, scaling and the like are mainly used. Atlas data processed in these ways: 1) the interference that the data may be subjected to in real traffic situations is not taken into account; 2) the data enhancement method based on shape processing does not expand data pixels, and the trained model is only suitable for images with specified colors and specifications and has poor generalization on data from different sources.
Disclosure of Invention
In order to overcome the defects in the prior art, the partial discharge spectrum is preprocessed by methods of background noise coupling, Gaussian blur and the like, so that one spectrum can generate a plurality of spectra. The method can solve the problem of insufficient data in the identification of the partial discharge atlas, and the number of samples is increased and overfitting is avoided by using a mode of generating a plurality of atlases by one atlas.
The invention designs a data enhancement method and a data enhancement device based on noise disturbance and Gaussian blur, which apply the Gaussian blur method on the basis of adding on-site common noise such as mobile phone interference, radar interference and microwave sulfur lamp interference to randomly disturb RGB pixels so as to train a network with stronger generalization capability.
Specifically, the data enhancement method for the partial discharge map, provided by the invention, utilizes a background noise coupling and Gaussian blur method to preprocess the partial discharge map, so that one map can generate a plurality of maps, and comprises the following steps:
s1: carrying out noise coupling on the partial discharge original data to generate a picture file;
s2: further processing the picture file after noise coupling by adopting a Gaussian blur method;
s3: and combining the samples subjected to noise coupling and Gaussian blur and the original samples for training the deep learning network model.
The first step specifically comprises the following steps:
s11: taking the partial discharge original data acquired by the acquisition front end as input data;
s12: converting input data into three-dimensional data;
s13: respectively generating three-dimensional data corresponding to each noise according to the data characteristics of the noise interference;
s14: accumulating the input data and the noise data to obtain data after noise coupling;
s15: and converting the data after the noise coupling into a picture file.
Further, the air conditioner is provided with a fan,
the three-dimensional data takes the phase as an x-axis, the period as a y-axis and the amplitude as a z-axis.
Further, the air conditioner is provided with a fan,
the noise interference comprises radar noise, mobile phone noise and microwave sulfur lamp interference.
Further, the second step specifically comprises:
s21: taking the picture file after noise coupling in the step one as input data;
s22: extracting RGB (Red, Green, Blue) values of an input picture file;
s23: respectively taking the central point of the picture in the picture file as a zero point, and drawing horizontal and vertical coordinate axes;
s24: calculating the weight of each pixel point in the picture by using a two-dimensional Gaussian distribution function to form a weight matrix;
s25: carrying out normalization processing on the weight matrix;
s26: updating the RGB value of the picture file by the normalized weight;
s27: and storing the picture file with the updated RGB values as a new sample.
Further, the air conditioner is provided with a fan,
the two-dimensional gaussian distribution function described in step S24 is as follows:
in the formula, x and y are horizontal and vertical coordinates of each pixel point in the graph from a zero point, G (x and y) is a weight value from the point to the zero point, pi is a circumferential rate, e is a natural constant, sigma is a standard deviation of normal distribution, sigma is generally between 1 and 3, and the image is smoother when the value is larger.
Further, the air conditioner is provided with a fan,
in step S25, the normalization processing of the weight matrix includes:
and calculating the sum m of all weight values in the weight matrix, and multiplying each weight value in the weight matrix by 1/m to obtain the normalized weight matrix.
Further, in step S26,
and for the weight matrix of each zero point, multiplying the weight value of each normalized weight matrix by the pixel value at the position of the weight value, summing the pixel values to serve as a new pixel value of the zero point, and respectively updating the RGB values by the weight updating method to obtain an updated three-color pixel value.
The invention also provides a data enhancement device for the partial discharge map, which comprises:
the device comprises a noise coupling module, a Gaussian fuzzy module and a sample generation module, wherein the output of the noise coupling module is connected with the input of the Gaussian fuzzy module, the output of the Gaussian module is connected with the sample generation module, and the partial discharge spectrum is preprocessed by using a background noise coupling and Gaussian fuzzy method, so that a plurality of spectra can be generated by one spectrum.
Further, the air conditioner is provided with a fan,
the noise coupling module comprises a data input module, a three-dimensional data generation module, a data coupling module and a picture conversion module, wherein the output of the data input module is connected with the input of the three-dimensional data generation module, the output of the three-dimensional data generation module is connected with the input of the data coupling module, and the output of the data coupling module is connected with the input of the picture conversion module.
Further, the air conditioner is provided with a fan,
and the data input module is used for inputting the partial discharge original data acquired by the acquisition front end.
Further, the air conditioner is provided with a fan,
the three-dimensional data generation module converts input data into three-dimensional data with the phase as an x axis, the period as a y axis and the amplitude as a z axis; and respectively generating three-dimensional data corresponding to each noise according to the data characteristics of radar noise, mobile phone noise and microwave sulfur lamp interference.
Further, the air conditioner is provided with a fan,
and the data coupling module accumulates the three-dimensional data of the input data and the noise data to obtain data after noise coupling.
Further, the air conditioner is provided with a fan,
and the picture conversion module is used for converting the data after noise coupling into a picture file.
Further, the air conditioner is provided with a fan,
the Gaussian blur module comprises a pixel extraction module, a weight matrix generation module, a weight update module and a file generation module, wherein the output of the picture conversion module is connected with the input of the pixel extraction module, the output of the pixel extraction module is connected with the input of the weight matrix generation module, the output of the weight matrix generation module is connected with the input of the weight update module, and the output of the weight update module is connected with the input of the file generation module and is used for carrying out Gaussian blur processing on the picture after noise coupling.
Further, the pixel extraction module inputs the picture file converted by the picture conversion module, and extracts an RGB (Red, Green, Blue) value of the picture file.
Further, the air conditioner is provided with a fan,
the weight matrix generation module is used for respectively taking the central point of the picture in the picture file as a zero point and drawing a horizontal axis and a vertical axis;
and calculating the weight of each pixel point in the picture by using a two-dimensional Gaussian distribution function to form a weight matrix, and further obtaining a normalized weight matrix.
Further, the air conditioner is provided with a fan,
the two-dimensional gaussian distribution function is as follows:
in the formula, x and y are horizontal and vertical coordinates of each pixel point in the graph from a zero point, G (x and y) is a weight value from the point to the zero point, pi is a circumferential rate, e is a natural constant, sigma is a standard deviation of normal distribution, sigma is generally between 1 and 3, and the image is smoother when the value is larger.
Further, the air conditioner is provided with a fan,
the obtaining of the normalized weight matrix specifically includes:
and calculating the sum m of all weight values in the weight matrix, and multiplying each weight value in the weight matrix by 1/m to obtain the normalized weight matrix.
Further, the air conditioner is provided with a fan,
and the weight value updating module updates the RGB value of the picture file by weight.
Further, the air conditioner is provided with a fan,
updating the RGB values of the picture with the weights specifically includes:
and for the weight matrix of each zero point, multiplying the weight value of each normalized weight matrix by the pixel value at the position of the weight value, summing the pixel values to serve as a new pixel value of the zero point, and respectively updating the RGB values by the weight updating method to obtain an updated three-color pixel value.
Further, the air conditioner is provided with a fan,
and the file generation module is used for storing the picture file with the updated RGB value as a new sample.
Further, the air conditioner is provided with a fan,
and the sample generation module is used for combining the noise-coupled and Gaussian-fuzzy sample and the original sample and is used for training the deep learning network model.
Compared with the method of directly using the sample as the training data, the method of the invention has the following advantages:
(1) a large number of labeled samples are generated with less calculation amount, so that the problems of high acquisition cost and insufficient training data of the labeled samples are solved;
(2) common interference on site is considered on the basis of an original data sample, real site data are simulated, and the anti-interference capability of a training model is improved;
(3) the diversity of the training data is expanded, and the method is expanded from a basic sample to a multi-pixel and multi-shape sample. And the extended training data is used for training the model, so that overfitting of the model can be avoided.
Drawings
FIG. 1 is a flow chart of a data enhancement method based on noise disturbance and Gaussian blur according to the present invention.
Fig. 2 is a schematic diagram of a data enhancement device based on noise disturbance and gaussian blur according to the present invention.
Fig. 3 is a three-dimensional image of raw data without processing.
Fig. 4 is an image file after noise superimposition.
Fig. 5 is an image file after superimposing gaussian blur.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the drawings of the embodiments of the present invention. It is to be understood that the embodiments described are only a few embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the described embodiments of the invention, are within the scope of the invention.
Referring to a flow chart shown in fig. 1, the technical scheme adopted by the invention is that a data enhancement method aiming at a partial discharge map based on noise disturbance and gaussian blur comprises the following steps:
(1) noise coupling
① the input data is the original data of partial discharge collected by the front end;
② converting the input data into three-dimensional data with phase as x-axis, period as y-axis and amplitude as z-axis;
③ respectively generating three-dimensional data corresponding to each noise according to the data characteristics of radar noise, mobile phone noise and microwave sulfur lamp interference;
④ accumulating the three-dimensional data of the input data and the noise data to obtain data after noise coupling;
⑤ converts the noise-coupled data into a picture file.
(2) Gaussian blur
① inputting data as picture files after noise coupling;
② extracting RGB (Red, Green, Blue) value of the input picture file;
③, respectively taking the center point of the picture in the picture file as the zero point, drawing the horizontal and vertical axes, calculating the weight matrix of each pixel point in the picture by a two-dimensional Gaussian distribution function, wherein the two-dimensional Gaussian distribution is as follows:
in the formula, x and y are respectively horizontal and vertical coordinates of each pixel point in the graph from a zero point, G (x and y) is a weighted value from the point to the zero point, pi is a circumferential rate, e is a natural constant, sigma is a standard deviation of normal distribution, sigma is generally between 1 and 3, and the image is smoother when the value is larger;
taking an image of 3 × 3 pixels as an example, when the midpoint of the image is taken as a zero point, the values of (x, y) of each point are shown in the following table:
| (-1，1) | (0，1) | (1，1) |
| (-1，0) | (0，0) | (1，0) |
| (-1，-1) | (0，-1) | (1，-1) |
the weights were calculated for each group (x, y) taking σ as 1.5, and the results are shown in the following table:
| 0.04535 | 0.05664 | 0.04535 |
| 0.05664 | 0.07074 | 0.05664 |
| 0.04535 | 0.05664 | 0.04535 |
④, normalizing the weight matrix, namely calculating the sum m of all weight values in the weight matrix, and multiplying each weight value in the weight matrix by 1/m to obtain the normalized weight matrix;
⑤ update the RGB values of the picture file with weights:
multiplying the weight value of each normalized weight matrix by the pixel value at the position of the weight value, summing the pixel values to obtain a new pixel value of the zero point, and respectively updating the RGB values by the weight updating method to obtain updated three-color pixel values;
⑥ store the picture file after updating the RGB values as a new sample.
(3) And combining the samples subjected to noise coupling and Gaussian blur and the original samples into a new sample set for training the deep learning network model.
Referring to fig. 2, the present invention further provides a data enhancement apparatus based on noise disturbance and gaussian blur, including: the device comprises a noise coupling module, a Gaussian blur module and a sample generation module.
The noise coupling module comprises a data input module, a three-dimensional data generation module, a data coupling module and an image conversion module.
The data input module is used for inputting the partial discharge original data acquired by the acquisition front end;
the three-dimensional data generation module is used for converting input data into three-dimensional data with the phase as an x axis, the period as a y axis and the amplitude as a z axis; respectively generating three-dimensional data corresponding to each noise according to data characteristics of radar noise, mobile phone noise and microwave sulfur lamp interference;
the data coupling module accumulates the input data and the noise data to obtain data after noise coupling;
and the image conversion module is used for converting the data after the noise coupling into an image file.
The Gaussian blur module comprises:
the pixel extraction module is used for inputting the picture file after noise coupling and extracting the RGB (Red, Green and Blue) value of the input picture file;
the weight matrix generation module is used for respectively taking the central point of the picture in the picture file as a zero point and drawing a horizontal axis and a vertical axis; and calculating the weight of each pixel point in the picture by using a two-dimensional Gaussian distribution function to form a weight matrix, and further obtaining a normalized weight matrix.
The weight value updating module updates the RGB value of the picture file by weight;
and for the weight matrix of each zero point, multiplying the weight value of each normalized weight matrix by the pixel value at the position of the weight value, summing the pixel values to serve as a new pixel value of the zero point, and respectively updating the RGB values by the weight updating method to obtain an updated three-color pixel value.
And the file generation module is used for storing the picture file with the updated RGB value as a new sample.
And the sample generation module is used for combining the noise-coupled Gaussian-fuzzy sample and the original sample and is used for training the deep learning network model.
Examples
The method of the present invention is further illustrated below with reference to the application data presented in figures 3-5.
The invention relates to a data enhancement method aiming at a local amplification map and based on noise disturbance and Gaussian blur, which comprises the following steps:
a strip of partial discharge raw data is acquired, and an unprocessed three-dimensional image thereof is shown in fig. 3.
And (3) respectively superposing the light, radar and mobile phone noise interference on the original sample data, and performing three-dimensional processing on the data subjected to superposition interference to obtain an image file subjected to noise superposition, as shown in fig. 4.
Taking sigma as 1.5, respectively calculating Gaussian weight matrixes for RGB (red, green and blue) three-color pixels of the image file after noise superposition, and superposing Gaussian blur to obtain the image file as shown in figure 5.
And (4) bringing the image file after the Gaussian blur and the original file into a sample library together to complete data enhancement based on noise superposition and Gaussian blur.
Therefore, compared with the method of directly using the sample as the training data, the method can generate a large number of labeled samples with less calculation amount, considers the common interference on site on the basis of the original data sample, can solve the problem of insufficient data in the identification of the partial discharge atlas, increases the number of samples and avoids overfitting by using a mode of generating a plurality of atlases by one atlas.
While the best mode for carrying out the invention has been described in detail and illustrated in the accompanying drawings, it is to be understood that the same is by way of illustration and example only and is not to be taken by way of limitation, the scope of the invention should be determined by the appended claims and any changes or modifications which fall within the true spirit and scope of the invention should be construed as broadly described herein.
Claims (23)
1. A data enhancement method for a partial discharge map is characterized in that the partial discharge map is preprocessed by using a background noise coupling and Gaussian blur method, so that a plurality of maps can be generated by one map, and the method comprises the following steps:
s1: carrying out noise coupling on the partial discharge original data to generate a picture file;
s2: further processing the picture file after noise coupling by adopting a Gaussian blur method;
s3: and combining the samples subjected to noise coupling and Gaussian blur and the original samples for training the deep learning network model.
2. The method according to claim 1, wherein step one comprises in particular:
s11: taking the partial discharge original data acquired by the acquisition front end as input data;
s12: converting input data into three-dimensional data;
s13: respectively generating three-dimensional data corresponding to each noise according to the data characteristics of the noise interference;
s14: accumulating the input data and the noise data to obtain data after noise coupling;
s15: and converting the data after the noise coupling into a picture file.
3. The method of claim 2,
the three-dimensional data takes the phase as an x-axis, the period as a y-axis and the amplitude as a z-axis.
4. The method of claim 2,
the noise interference comprises radar noise, mobile phone noise and microwave sulfur lamp interference.
5. The method according to one of claims 1 to 4, characterized in that step two comprises in particular:
s21: taking the picture file after noise coupling in the step one as input data;
s22: extracting RGB (Red, Green, Blue) values of an input picture file;
s23: respectively taking the central point of the picture in the picture file as a zero point, and drawing horizontal and vertical coordinate axes;
s24: calculating the weight of each pixel point in the picture by using a two-dimensional Gaussian distribution function to form a weight matrix;
s25: carrying out normalization processing on the weight matrix;
s26: updating the RGB value of the picture file by the normalized weight;
s27: and storing the picture file with the updated RGB values as a new sample.
6. The method according to claim 5, wherein the two-dimensional Gaussian distribution function in step S24 is as follows:
in the formula, x and y are horizontal and vertical coordinates of each pixel point in the graph from a zero point, G (x and y) is a weight value from the point to the zero point, pi is a circumferential rate, e is a natural constant, sigma is a standard deviation of normal distribution, sigma is generally between 1 and 3, and the image is smoother when the value is larger.
7. The method according to claim 5 or 6, wherein the step S25, the normalization of the weight matrix comprises:
and calculating the sum m of all weight values in the weight matrix, and multiplying each weight value in the weight matrix by 1/m to obtain the normalized weight matrix.
8. The method according to claim 7, wherein, in step S26,
and for the weight matrix of each zero point, multiplying the weight value of each normalized weight matrix by the pixel value at the position of the weight value, summing the pixel values to serve as a new pixel value of the zero point, and respectively updating the RGB values by the weight updating method to obtain an updated three-color pixel value.
9. A data enhancement apparatus for a partial discharge map, comprising:
the device comprises a noise coupling module, a Gaussian blur module and a sample generation module, wherein the output of the noise coupling module is connected with the input of the Gaussian blur module, the output of the Gaussian blur module is connected with the sample generation module, and a partial discharge map is preprocessed by using a background noise coupling and Gaussian blur method, so that a plurality of maps can be generated by one map.
10. The apparatus of claim 9,
the noise coupling module comprises a data input module, a three-dimensional data generation module, a data coupling module and a picture conversion module, wherein the output of the data input module is connected with the input of the three-dimensional data generation module, the output of the three-dimensional data generation module is connected with the input of the data coupling module, and the output of the data coupling module is connected with the input of the picture conversion module.
11. The apparatus of claim 10,
and the data input module is used for inputting the partial discharge original data acquired by the acquisition front end.
12. The apparatus of claim 11,
the three-dimensional data generation module converts input data into three-dimensional data with the phase as an x axis, the period as a y axis and the amplitude as a z axis; and respectively generating three-dimensional data corresponding to each noise according to the data characteristics of radar noise, mobile phone noise and microwave sulfur lamp interference.
13. The apparatus of claim 12,
and the data coupling module accumulates the three-dimensional data of the input data and the noise data to obtain data after noise coupling.
14. The apparatus of claim 13,
and the picture conversion module is used for converting the data after the noise coupling into a picture file.
15. The apparatus according to claim 14, wherein the gaussian blur module comprises a pixel extraction module, a weight matrix generation module, a weight update module and a file generation module, wherein an output of the picture conversion module is connected to an input of the pixel extraction module, an output of the pixel extraction module is connected to an input of the weight matrix generation module, an output of the weight matrix generation module is connected to an input of the weight update module, and an output of the weight update module is connected to an input of the file generation module, and is configured to perform gaussian blur processing on the noise-coupled picture.
16. The apparatus of claim 15,
the pixel extraction module inputs the picture file converted by the picture conversion module and extracts the RGB (Red, Green, Blue) value of the picture file.
17. The apparatus of claim 16,
the weight matrix generation module is used for respectively taking the central point of the picture in the picture file as a zero point and drawing a horizontal axis and a vertical axis; and calculating the weight of each pixel point in the picture by using a two-dimensional Gaussian distribution function to form a weight matrix, and further obtaining a normalized weight matrix.
18. The apparatus of claim 17,
the two-dimensional gaussian distribution function is as follows:
in the formula, x and y are horizontal and vertical coordinates of each pixel point in the graph from a zero point, G (x and y) is a weight value from the point to the zero point, pi is a circumferential rate, e is a natural constant, sigma is a standard deviation of normal distribution, sigma is generally between 1 and 3, and the image is smoother when the value is larger.
19. The apparatus of claim 18,
the obtaining of the normalized weight matrix specifically includes:
and calculating the sum m of all weight values in the weight matrix, and multiplying each weight value in the weight matrix by 1/m to obtain the normalized weight matrix.
20. The apparatus of claim 19,
and the weight value updating module updates the RGB value of the picture file by weight.
21. The apparatus of claim 20,
updating the RGB values of the picture with the weights specifically includes:
and for the weight matrix of each zero point, multiplying the weight value of each normalized weight matrix by the pixel value at the position of the weight value, summing the pixel values to serve as a new pixel value of the zero point, and respectively updating the RGB values by the weight updating method to obtain an updated three-color pixel value.
22. The apparatus of claim 21,
and the file generation module is used for storing the picture file with the updated RGB value as a new sample.
23. The apparatus of claim 22,
and the sample generation module is used for combining the noise-coupled and Gaussian-fuzzy sample and the original sample and is used for training the deep learning network model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911074459.5A CN110796206B (en) | 2019-11-06 | 2019-11-06 | Data enhancement method and device for partial discharge map |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911074459.5A CN110796206B (en) | 2019-11-06 | 2019-11-06 | Data enhancement method and device for partial discharge map |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110796206A true CN110796206A (en) | 2020-02-14 |
| CN110796206B CN110796206B (en) | 2022-08-30 |
Family
ID=69442926
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911074459.5A Active CN110796206B (en) | 2019-11-06 | 2019-11-06 | Data enhancement method and device for partial discharge map |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110796206B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112101294A (en) * | 2020-09-29 | 2020-12-18 | 支付宝(杭州)信息技术有限公司 | Enhanced training method and device for image recognition model |
| CN117975201A (en) * | 2024-03-29 | 2024-05-03 | 苏州元脑智能科技有限公司 | Training data generation method, device, computer equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130179100A1 (en) * | 2012-01-11 | 2013-07-11 | Utilx Corporation | System for analyzing and locating partial discharges |
| CN108470139A (en) * | 2018-01-25 | 2018-08-31 | 天津大学 | A kind of small sample radar image human action sorting technique based on data enhancing |
| CN110208660A (en) * | 2019-06-05 | 2019-09-06 | 国网江苏省电力有限公司电力科学研究院 | A kind of training method and device for power equipment shelf depreciation defect diagonsis |
| CN110703057A (en) * | 2019-11-04 | 2020-01-17 | 国网山东省电力公司电力科学研究院 | Power equipment partial discharge diagnosis method based on data enhancement and neural network |
-
2019
- 2019-11-06 CN CN201911074459.5A patent/CN110796206B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130179100A1 (en) * | 2012-01-11 | 2013-07-11 | Utilx Corporation | System for analyzing and locating partial discharges |
| CN108470139A (en) * | 2018-01-25 | 2018-08-31 | 天津大学 | A kind of small sample radar image human action sorting technique based on data enhancing |
| CN110208660A (en) * | 2019-06-05 | 2019-09-06 | 国网江苏省电力有限公司电力科学研究院 | A kind of training method and device for power equipment shelf depreciation defect diagonsis |
| CN110703057A (en) * | 2019-11-04 | 2020-01-17 | 国网山东省电力公司电力科学研究院 | Power equipment partial discharge diagnosis method based on data enhancement and neural network |
Non-Patent Citations (2)
| Title |
|---|
| 在路上DI蜗牛: "图像处理：高斯模糊，https://blog.csdn.net/qinghuaci666/article/details/81870277", 《网页文档》 * |
| 李沅箐: "基于深度学习的声目标识别方法研究", 《万方数据》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112101294A (en) * | 2020-09-29 | 2020-12-18 | 支付宝(杭州)信息技术有限公司 | Enhanced training method and device for image recognition model |
| US11403487B2 (en) | 2020-09-29 | 2022-08-02 | Alipay (Hangzhou) Information Technology Co., Ltd. | Enhanced training method and apparatus for image recognition model |
| CN117975201A (en) * | 2024-03-29 | 2024-05-03 | 苏州元脑智能科技有限公司 | Training data generation method, device, computer equipment and storage medium |
| CN117975201B (en) * | 2024-03-29 | 2024-06-25 | 苏州元脑智能科技有限公司 | Training data generation method, device, computer equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110796206B (en) | 2022-08-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110503108B (en) | Method and device for identifying illegal buildings, storage medium and computer equipment | |
| CN110619282B (en) | Automatic extraction method for unmanned aerial vehicle orthoscopic image building | |
| CN109872305B (en) | No-reference stereo image quality evaluation method based on quality map generation network | |
| CN110796206B (en) | Data enhancement method and device for partial discharge map | |
| CN107680077A (en) | A kind of non-reference picture quality appraisement method based on multistage Gradient Features | |
| CN109801218B (en) | Multispectral remote sensing image Pan-sharpening method based on multilayer coupling convolutional neural network | |
| CN111444973A (en) | Method for detecting commodities on unmanned retail shopping table | |
| CN113808180B (en) | Heterologous image registration method, system and device | |
| CN108615228A (en) | Facial image complementing method based on hybrid neural networks | |
| CN104616026A (en) | Monitor scene type identification method for intelligent video monitor | |
| CN103955907A (en) | Method for telemetering pointer type SF6 gas density meter | |
| CN113327255A (en) | Power transmission line inspection image processing method based on YOLOv3 detection, positioning and cutting and fine-tune | |
| CN113269725A (en) | Coal gangue rapid detection method based on imaging technology and deep learning | |
| CN113421222A (en) | Lightweight coal gangue target detection method | |
| CN116665092A (en) | Method and system for identifying sewage suspended matters based on IA-YOLOV7 | |
| CN104809735B (en) | The system and method for image haze evaluation is realized based on Fourier transformation | |
| CN118297755A (en) | Building construction analysis management and control method and system based on Internet of things and big data technology | |
| CN105389825A (en) | Image processing method and system | |
| CN112800857A (en) | Bare land rapid extraction method based on high-resolution satellite data | |
| CN113537306A (en) | Image classification method based on progressive growth element learning | |
| CN105205485A (en) | Large scale image segmentation algorithm based on multi-inter-class maximum variance algorithm | |
| CN110135274B (en) | Face recognition-based people flow statistics method | |
| CN116957974A (en) | Image rain removing method and device based on hybrid network | |
| CN111986079A (en) | Pavement crack image super-resolution reconstruction method and device based on generation countermeasure network | |
| CN115100451B (en) | Data expansion method for monitoring oil leakage of hydraulic pump |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |