CN109948629A - A kind of GIS equipment X-ray image failure detection method based on SIFT feature - Google Patents
A kind of GIS equipment X-ray image failure detection method based on SIFT feature Download PDFInfo
- Publication number
- CN109948629A CN109948629A CN201910204447.3A CN201910204447A CN109948629A CN 109948629 A CN109948629 A CN 109948629A CN 201910204447 A CN201910204447 A CN 201910204447A CN 109948629 A CN109948629 A CN 109948629A
- Authority
- CN
- China
- Prior art keywords
- image
- sift feature
- normal
- measured
- ray image
- 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
Landscapes
- Apparatus For Radiation Diagnosis (AREA)
- Image Analysis (AREA)
- Analysing Materials By The Use Of Radiation (AREA)
Abstract
The GIS equipment X-ray image failure detection method based on SIFT feature that the invention discloses a kind of, which comprises S1, obtain GIS device normal X ray image and radioscopic image to be measured；S2, the SIFT feature for extracting normal X ray image and radioscopic image to be measured respectively, count the position of identical SIFT feature；S3, image registration is carried out to radioscopic image to be measured and normal X ray image using identical SIFT feature；S4, to after registration radioscopic image to be measured and normal X ray image carry out image block；S5, the similarity that radioscopic image to be measured Yu normal X ray image identical image piecemeal are calculated using SIFT feature then determine that the image block is faulty when similarity is greater than the similar threshold value of setting.The present invention not only carries out GIS equipment X-ray image using SIFT feature to be registrated but also carry out intelligent fault diagnosis, and the automatization level of detection can be improved, provide reference for artificial detection, improves accuracy, the rapidity of fault diagnosis, realizes intelligent diagnostics.
Description
Technical field
The GIS equipment X-ray image failure detection method based on SIFT feature that the present invention relates to a kind of, belongs to GIS device X
Ray detection field.
Background technique
Cubicle Gas-Insulated Switchgear (GIS) is widely applied in China's power grid, once it is deposited inside it
It is in office why to hinder and defect, equipment overall performance may be all influenced, administrative some areas or even all the power failure thing in area are caused
Therefore because a little needs periodically detect GIS device, carrying out contactless non-destructive testing to GIS device using X-ray is one
The effective external diagnosis means of kind, to ensure the operation of GIS safety the case where not dismantling.It deposits at the scenes such as substation, power plant
It is largely interfering, obtained X-ray digital image is there are noise and not clear enough, and current research is more to be concentrated mainly on
Pretreatment is denoised, enhanced etc. to GIS equipment X-ray digital picture, after having handled, then is manually judged.And it is right
The method of the detection of GIS equipment X-ray image computer assist trouble and intelligent diagnostics is less.
Summary of the invention
For the problems in background technique, it is an object of the invention to: a kind of GIS device X based on SIFT feature is provided
Ray image fault detection method realizes the intelligent diagnostics of failure to improve accuracy, the rapidity of fault detection.
The technical scheme is that a kind of GIS equipment X-ray image failure detection method based on SIFT feature, institute
The method of stating includes:
S1, GIS device normal X ray image and radioscopic image to be measured are obtained；
S2, the SIFT feature for extracting normal X ray image and radioscopic image to be measured respectively, count identical SIFT feature
Position；
S3, image registration is carried out to radioscopic image to be measured and normal X ray image using identical SIFT feature；
S4, to after registration radioscopic image to be measured and normal X ray image carry out image block；
S5, the similarity that radioscopic image to be measured Yu normal X ray image identical image piecemeal are calculated using SIFT feature,
When similarity is greater than the similar threshold value of setting, then determine that the image block is faulty.
Optionally, the SIFT feature that error is greater than given threshold is rejected before registration.
Optionally, rejecting difference greater than the method for the SIFT feature of given threshold includes:
(1) the X mean difference and Y of radioscopic image and SIFT feature position identical in normal X ray image to be measured are calculated
Mean difference, formula are
Wherein, DxAnd DyRespectively X mean difference and Y mean difference, XoiAnd YoiFor the position of radioscopic image characteristic point to be measured
Set coordinate value, XciAnd YciFor the position coordinate value of normal radioscopic image characteristic point, N is the quantity of same characteristic features point；
(2) SIFT feature that difference is greater than given threshold is rejected, the point for meeting following condition is removed,
if(|Xoi-Xci|-Dx> μx) then rejecting
if(|Yoi-Yci|-Dy> μy) then rejecting
μx, μyIt is the given threshold of X-coordinate and Y-coordinate respectively,
(3) with remaining N after rejectingTBased on a identical SIFT feature, radioscopic image to be measured and normal is recalculated
The X mean difference of the identical point of SIFT feature and Y mean difference, formula are in radioscopic image
(4) with X0=Xc-Dx, Y0=Yc-DyOn the basis of, radioscopic image to be measured is registrated with normal X ray image.
Optionally, the method for calculating image block similarity includes:
(1) the number N of radioscopic image to be measured SIFT feature identical as normal X ray image in piecemeal is countedC, statistics
The number N of radioscopic image and the not identical SIFT feature of normal X ray image to be measured in piecemealm, count normal X ray image
The number N of SIFT featurez；
(2) mean value of radioscopic image to be measured Yu normal X ray image SIFT feature operator is counted respectively,
S is SIFT feature operator, is 0 without characteristic point pixel operator；
(3) similarity: i-th piece of similarity is calculated are as follows:
λ1, λ2, λ3For the coefficient less than 1；
(4) similarity is compared with similar threshold value
If (sim (i) > μ) then failure
μ is threshold value less than 1, and i-th piece is faulty, can in radioscopic image to be measured fault location.
Optionally, to normal X ray image and radioscopic image to be measured processing denoising before extracting SIFT feature.
Optionally, normal X ray image and radioscopic image to be measured are denoised using gaussian filtering.
The beneficial effects of the present invention are: the invention proposes the GIS equipment X-ray image failure intelligence based on SIFT feature
Diagnostic method not only carries out GIS equipment X-ray image using SIFT feature to be registrated but also carry out intelligent fault diagnosis, can be improved
The automatization level of detection provides reference for artificial detection, improves accuracy, the rapidity of fault diagnosis, realizes intelligent diagnostics.
Detailed description of the invention
Fig. 1 is the flow chart according to the method for the present invention；
Fig. 2 is X ray picture, and wherein a figure is normal X ray picture, and b is X ray picture to be checked；
Fig. 3 is that normal X-ray extracts SIFT feature figure；
Fig. 4 is that X-ray to be checked extracts SIFT spy's piece figure；
Fig. 5 block SIFT feature comparison diagram；
In figure, 1 is metal fall-out, and 2 be SIFT feature, and 3 be non-matching characteristic point.
Specific embodiment
With reference to the accompanying drawing and invention is described further in specific embodiment:
S1, GIS device normal X ray image and radioscopic image to be measured are obtained.
The normal X ray image and radioscopic image to be measured at the same position of GIS device are obtained, i.e., is run just in the GIS device
Normal X ray image is shot and stored in advance when often, is recorded the parameters such as its camera site distance, focal length, is detected to GIS device
When shoot radioscopic image to be measured, preferably its camera site and parameter etc. is identical as normal X ray image.
S2, image preprocessing.
To normal X ray image and radioscopic image to be measured processing denoising before extracting SIFT feature, it is preferable that using high
This filtering carries out identical denoising to normal X ray image and radioscopic image to be measured.For example, gaussian filtering is using two dimension
The discrete Gaussian function of zero-mean does smoothing filter, function expression are as follows:
S3, SIFT feature are extracted.
The SIFT feature of normal X ray image and radioscopic image to be measured is extracted respectively.For example, extracting whole picture radioscopic image
SIFT operator, the method used is that David G.Lowe proposed that a kind of part based on Scale-space theory is special in 1999
Sign extracts SIFT operator, which has good invariance to graphical rule scaling, rotation and affine transformation etc.
S4, the identical SIFT feature position of statistics.
Radioscopic image to be measured is compared with the SIFT operator of normal X ray image, detects the complete phase of SIFT operator
Same point, and record the position X in their place figures0iAnd Y0i., X0iIt indicates identical with radioscopic image SIFT operator to be checked
Normal X ray image to the X-coordinate value of above i-th point of the point；Y0iIt indicates identical with radioscopic image SIFT operator to be checked
Normal X ray image to the Y-coordinate value of above i-th point of the point.
S5, image registration.
Image registration is carried out to radioscopic image to be measured and normal X ray image using identical SIFT feature, before registration
Reject the SIFT feature that error is greater than given threshold.
It rejects difference and is greater than the method for SIFT feature of given threshold and include:
(1) the X mean difference and Y of radioscopic image and SIFT feature position identical in normal X ray image to be measured are calculated
Mean difference, formula are
Wherein, DxAnd DyRespectively X mean difference and Y mean difference, XoiAnd YoiFor the position of radioscopic image characteristic point to be measured
Set coordinate value, XciAnd YciFor the position coordinate value of normal radioscopic image characteristic point, N is the quantity of same characteristic features point；
(2) SIFT feature that difference is greater than given threshold is rejected, the point for meeting following condition is removed,
if(|Xoi-Xci|-Dx> μx) then rejecting
if(|Yoi-Yci|-Dy> μy) then rejecting
μx, μyIt is the given threshold of X-coordinate and Y-coordinate respectively,
(3) with remaining N after rejectingTBased on a identical SIFT feature, radioscopic image to be measured and normal is recalculated
The X mean difference of the identical point of SIFT feature and Y mean difference, formula are in radioscopic image
(4) with X0=Xc-Dx, Y0=Yc-DyOn the basis of, the matching of radioscopic image to be measured Yu normal X ray image is completed,
The registration mapping relations between image are established by the matching relationship of feature, and then complete registration.Image registration is this field
Common knowledge, details are not described herein again.
S6, image block.
To the radioscopic image to be measured and normal X ray image progress image block after registration.For example, radioscopic image to be checked
32*32 pixel, 64*16 pixel are pressed with normal X ray image, 16*64 pixel from left to right, or carries out identical from top to bottom
Piecemeal.
The similarity of SIFT feature in S7, statistics piecemeal.
(1) the number N of radioscopic image to be measured SIFT feature identical as normal X ray image in piecemeal is countedC, statistics
The number N of radioscopic image and the not identical SIFT feature of normal X ray image to be measured in piecemealm, count normal X ray image
The number N of SIFT featurez；
(2) mean value of radioscopic image to be measured Yu normal X ray image SIFT feature operator is counted respectively,
S is SIFT feature operator, is 0 without characteristic point pixel operator；
(3) similarity: i-th piece of similarity is calculated are as follows:
λ1, λ2, λ3For the coefficient less than 1.
S8, similarity and threshold value comparison export result.
When similarity is greater than the similar threshold value of setting, then determine that the image block is faulty.Using following formula
If (sim (i) > μ) then failure
μ is threshold value less than 1, and i-th piece is faulty, can in radioscopic image to be measured fault location.
For example, a figure is normal X ray picture in Fig. 2, b figure is X ray picture to be checked, and 1 is metal fall-out in figure.Using this
After invention detection method, Fig. 3 is that normal X-ray extracts SIFT feature figure, and 2 be SIFT feature in figure, and Fig. 4 is X-ray to be checked
SIFT feature figure is extracted, Fig. 5 is SIFT feature comparison diagram, and 3 be non-matching characteristic point, since non-matching characteristic point significantly increases,
Detect that the block is faulty.According to measuring and calculating, accuracy rate of the invention is up to 92% or more.
The above content is a further detailed description of the present invention in conjunction with specific preferred embodiments, and it cannot be said that
Specific implementation of the invention is only limited to these instructions.For those of ordinary skill in the art to which the present invention belongs, exist
Under the premise of not departing from present inventive concept, a number of simple deductions or replacements can also be made, all shall be regarded as belonging to of the invention
Protection scope.
Claims (6)
1. a kind of GIS equipment X-ray image failure detection method based on SIFT feature, which comprises
S1, GIS device normal X ray image and radioscopic image to be measured are obtained；
S2, the SIFT feature for extracting normal X ray image and radioscopic image to be measured respectively, count the position of identical SIFT feature
It sets；
S3, image registration is carried out to radioscopic image to be measured and normal X ray image using identical SIFT feature；
S4, to after registration radioscopic image to be measured and normal X ray image carry out image block；
S5, the similarity that radioscopic image to be measured Yu normal X ray image identical image piecemeal are calculated using SIFT feature, work as phase
When being greater than the similar threshold value of setting like degree, then determine that the image block is faulty.
2. the GIS equipment X-ray image failure detection method based on SIFT feature according to claim 1, it is characterised in that:
The SIFT feature that error is greater than given threshold is rejected before registration.
3. the GIS equipment X-ray image failure detection method based on SIFT feature according to claim 2, it is characterised in that:
It rejects difference and is greater than the method for SIFT feature of given threshold and include:
(1) it calculates radioscopic image to be measured and the X mean difference of SIFT feature position identical in normal X ray image and Y is average
Difference, formula are
Wherein, DxAnd DyRespectively X mean difference and Y mean difference, XoiAnd YoiIt is sat for the position of radioscopic image characteristic point to be measured
Scale value, XciAnd YciFor the position coordinate value of normal radioscopic image characteristic point, N is the quantity of same characteristic features point；
(2) SIFT feature that difference is greater than given threshold is rejected, the point for meeting following condition is removed,
if(|Xoi-Xci|-Dx> μx) then rejecting
if(|Yoi-Yci|-Dy> μy) then rejecting
μx, μyIt is the given threshold of X-coordinate and Y-coordinate respectively,
(3) with remaining N after rejectingTBased on a identical SIFT feature, radioscopic image to be measured and normal X ray are recalculated
The X mean difference of the identical point of SIFT feature and Y mean difference, formula are in image
(4) with X0=Xc-Dx, Y0=Yc-DyOn the basis of, radioscopic image to be measured is registrated with normal X ray image.
4. the GIS equipment X-ray image failure detection method based on SIFT feature according to claim 1, it is characterised in that:
Calculate image block similarity method include:
(1) the number N of radioscopic image to be measured SIFT feature identical as normal X ray image in piecemeal is countedCIt counts in piecemeal
The number N of radioscopic image to be measured and the not identical SIFT feature of normal X ray imagem, statistics normal X ray image SIFT spy
Levy the number N of pointz；
(2) mean value of radioscopic image to be measured Yu normal X ray image SIFT feature operator is counted respectively,
S is SIFT feature operator, is 0 without characteristic point pixel operator；
(3) similarity: i-th piece of similarity is calculated are as follows:
λ1, λ2, λ3For the coefficient less than 1；
(4) similarity is compared with similar threshold value
If (sim (i) > μ) then failure
μ is threshold value less than 1, and i-th piece is faulty, can in radioscopic image to be measured fault location.
5. the GIS equipment X-ray image failure detection method based on SIFT feature according to claim 1, it is characterised in that:
To normal X ray image and radioscopic image to be measured processing denoising before extracting SIFT feature.
6. the GIS equipment X-ray image failure detection method based on SIFT feature according to claim 5, it is characterised in that:
Normal X ray image and radioscopic image to be measured are denoised using gaussian filtering.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910204447.3A CN109948629B (en) | 2019-03-18 | 2019-03-18 | GIS equipment X-ray image fault detection method based on SIFT features |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910204447.3A CN109948629B (en) | 2019-03-18 | 2019-03-18 | GIS equipment X-ray image fault detection method based on SIFT features |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109948629A true CN109948629A (en) | 2019-06-28 |
| CN109948629B CN109948629B (en) | 2020-03-24 |
Family
ID=67010193
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910204447.3A Active CN109948629B (en) | 2019-03-18 | 2019-03-18 | GIS equipment X-ray image fault detection method based on SIFT features |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109948629B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110909669A (en) * | 2019-11-19 | 2020-03-24 | 湖南国奥电力设备有限公司 | Underground cable fault determination method and device based on image detection |
| CN111324145A (en) * | 2020-02-28 | 2020-06-23 | 厦门理工学院 | Unmanned aerial vehicle autonomous landing method, device, equipment and storage medium |
| CN113160194A (en) * | 2021-04-28 | 2021-07-23 | 贵州电网有限责任公司 | Distance self-adaptive registration method for ultraviolet and visible light images |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101957325A (en) * | 2010-10-14 | 2011-01-26 | 山东鲁能智能技术有限公司 | Substation equipment appearance abnormality recognition method based on substation inspection robot |
| CN103065311A (en) * | 2012-12-26 | 2013-04-24 | 中国土地勘测规划院 | Satellite image automatic registration method based on standard image database |
| CN104809732A (en) * | 2015-05-07 | 2015-07-29 | 山东鲁能智能技术有限公司 | Electrical equipment appearance abnormity detection method based on image comparison |
-
2019
- 2019-03-18 CN CN201910204447.3A patent/CN109948629B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101957325A (en) * | 2010-10-14 | 2011-01-26 | 山东鲁能智能技术有限公司 | Substation equipment appearance abnormality recognition method based on substation inspection robot |
| CN103065311A (en) * | 2012-12-26 | 2013-04-24 | 中国土地勘测规划院 | Satellite image automatic registration method based on standard image database |
| CN104809732A (en) * | 2015-05-07 | 2015-07-29 | 山东鲁能智能技术有限公司 | Electrical equipment appearance abnormity detection method based on image comparison |
Non-Patent Citations (4)
| Title |
|---|
| ANDREA VEDALDI: ""An open implementation of the SIFT detector and descriptor"", 《UCLA CSD TECHNICAL REPORT 070012》 * |
| YOU ZHAI，LUAN ZENG: ""A SIFT Matching Algorithm Based On Adaptive Contrast Threshold "", 《2011 INTERNATIONAL CONFERENCE ON CONSUMER ELECTRONICS，COMMUNICATION AND NETWORKS》 * |
| 王妍玮: ""工业X射线图像处理技术研究"", 《中国博士学位论文全文数据库信息科技辑》 * |
| 陈红岩: ""无人车立体视觉里程计研究"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110909669A (en) * | 2019-11-19 | 2020-03-24 | 湖南国奥电力设备有限公司 | Underground cable fault determination method and device based on image detection |
| CN111324145A (en) * | 2020-02-28 | 2020-06-23 | 厦门理工学院 | Unmanned aerial vehicle autonomous landing method, device, equipment and storage medium |
| CN113160194A (en) * | 2021-04-28 | 2021-07-23 | 贵州电网有限责任公司 | Distance self-adaptive registration method for ultraviolet and visible light images |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109948629B (en) | 2020-03-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108921176B (en) | Pointer instrument positioning and identifying method based on machine vision | |
| CN108416791A (en) | A kind of monitoring of parallel institution moving platform pose and tracking based on binocular vision | |
| CN106824806B (en) | The detection method of low module plastic gear based on machine vision | |
| CN109948629A (en) | A kind of GIS equipment X-ray image failure detection method based on SIFT feature | |
| CN110580480A (en) | Meter reading identification method based on image processing | |
| CN115841434B (en) | Infrared image enhancement method for gas concentration analysis | |
| CN110414308B (en) | Target identification method for dynamic foreign matters on power transmission line | |
| CN106780526A (en) | A kind of ferrite wafer alligatoring recognition methods | |
| CN108648184A (en) | A kind of detection method of remote sensing images high-altitude cirrus | |
| WO2023231262A1 (en) | Hoisting wire rope tension measurement method based on visual vibration frequency identification | |
| CN110222661A (en) | It is a kind of for motion estimate and the feature extracting method of tracking | |
| CN114663672B (en) | Method and system for detecting corrosion of steel member of power transmission line tower | |
| CN110276747B (en) | Insulator fault detection and fault rating method based on image analysis | |
| CN111145198A (en) | Non-cooperative target motion estimation method based on rapid corner detection | |
| CN104966283A (en) | Imaging layered registering method | |
| CN113780119A (en) | High-precision moving target detection method based on application scene of static camera | |
| CN117314986A (en) | Unmanned aerial vehicle cross-mode power distribution equipment inspection image registration method based on semantic segmentation | |
| CN112508947A (en) | Cable tunnel abnormity detection method | |
| CN109558881A (en) | A kind of crag avalanche monitoring method based on computer vision | |
| CN106504251B (en) | A kind of electric aluminum foil cubic texture detection method of content based on picture processing | |
| CN111798476B (en) | Extraction method for conductive arm axis of high-voltage isolating switch | |
| CN106980863A (en) | A kind of unit exception diagnostic model in transformer substation video monitoring | |
| Ge et al. | An optimized design of the pointer meter image enhancement and automatic reading system in low illumination environment | |
| Zeng et al. | An improved canny edge detector against impulsive noise based on CIELAB space | |
| Fraser et al. | A feature-based matching strategy for automated 3D model reconstruction in multi-image close-range photogrammetry |
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 |