CN104766297B - A kind of electric power video image striped fault detection method combined based on spatial domain and time-domain analysis - Google Patents
A kind of electric power video image striped fault detection method combined based on spatial domain and time-domain analysis Download PDFInfo
- Publication number
- CN104766297B CN104766297B CN201410525862.6A CN201410525862A CN104766297B CN 104766297 B CN104766297 B CN 104766297B CN 201410525862 A CN201410525862 A CN 201410525862A CN 104766297 B CN104766297 B CN 104766297B
- Authority
- CN
- China
- Prior art keywords
- picture
- edge
- image
- pixel
- electric power
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Abstract
The invention discloses a kind of electric power video image striped fault detection method combined based on spatial domain and time-domain analysis, its step is as follows: carry out spatial processing to original video picture and be converted to edge picture, spatial processing flow process mainly includes carries out image edge processing, Image Edge-Detection process to original video picture, edge picture after process is converted to frequency domain by spatial domain, edge picture is converted to frequency domain picture, at frequency domain part, by the treatment scheme that frequency domain picture processing, picture bright spot are extracted, finally draw the evaluation of estimate of video image striped.This method can be used for having feature fast, accurately to the fault detect of electric power video equipment.
Description
Technical field
The invention belongs to technical field of image detection, be specifically related to the technology such as image procossing and computing machine, particularly relate to image striped fault detection method and the image striped evaluation of estimate of electric power video.
Background technology
Electric power video monitoring system is widely used in each service application such as electrical production, security protection, emergent, marketing, capital construction, environment singularity by the interference of electricity substation high-intensity magnetic field affects, video equipment easily occurs that striped is abnormal, the availability of video monitoring system, practicality are had a strong impact on, therefore, how effectively to realize the detection to the striped fault of video equipment, for the maintenance of electric power video equipment provides foundation, the every application for electric power video is provided more effective technical support.
For video image striped fault, adopt the mode of artificial cognition can distinguish failure condition easily, but, huge due to electric power head end video number of devices, according to the method for artificial inspection, need to spend a large amount of manpowers, and work efficiency reduces, therefore, adopt computer technology, that image analysis technology realizes electric power video image striped fault detection method is significant for the fault detect of video equipment.
Detect according to the streak feature of single analytical approach to video image, due to the singularity of image striped, streak feature under the different scene of easy loss, varying environment, different striped type is lost, therefore, the accuracy of video striped fault detect effectively can be improved to greatest extent in conjunction with spatial method and time-domain analysis.
Summary of the invention
Goal of the invention: technical matters to be solved by this invention is for the deficiencies in the prior art, provides a kind of electric power video image striped fault detection method combined based on spatial method and time-domain analysis.
In order to solve the problems of the technologies described above, provided by the inventionly analyse the electric power video image striped fault detection method combined with time-domain analysis based on spatial domain, comprise the enter factor method of acquisition image edge extraction, image edge extracting method, the conversion of Fourier's frequency domain, video pictures striped evaluation of estimate acquisition methods Four processes, it is as follows that the method comprising the steps of:
S1: be converted to gray level image to original video picture, and the data such as the complexity of computed image, average gradient, as the enter factor of Edge extraction;
S2: adopt image edge enhancement method, sliding window edge extracting method, the enter factor of combining image edge extracting, extracts the edge in original image, final generation corresponding edge picture;
S3: the edge picture after spatial processing, adopts inverse discrete Fourier transform that edge picture is converted to frequency domain picture.
S4: the picture bright spot after conversion is all distributed in four angles of picture, after being changed by symmetry, bright spot is all concentrated on central point, the sliding window of 5*5 is adopted to detect picture bright spot, draw bright spot numerical value, and in conjunction with image edge average gradient, image edge complexity as multiple parameters such as factors of influence, draw final video pictures striped evaluation points, draw evaluation of estimate according to evaluation points characteristic distributions.
Wherein, the enter factor method concrete steps obtaining image edge extraction are as follows:
S11: original image is converted to gray scale picture;
S12: the part edge of excision gray scale picture, obtains new gray scale picture;
S13: the complexity new gray scale picture being asked for picture;
S14: the total average gradient value of the 3*3 pixel of picture is asked for new gray scale picture.
Wherein, image edge extracting method concrete steps are obtained as follows:
S21: the average gradient calculating the 3*3 pixel of each pixel for new gray scale picture;
S22: extract enter factor according to image complexity, overall average Grad edge calculation;
S23: according to the average gradient value of each pixel, obtains dynamic pixel and strengthens coefficient;
S24: according to the average gradient of edge extracting enter factor, dynamic pixel enhancing coefficient, pixel, obtain each pixel value of edge picture.
Wherein, video pictures striped evaluation of estimate acquisition methods concrete steps are as follows:
S41: the picture bright spot after conversion is all distributed in four angles of picture, after being changed, bright spot is all concentrated on central point by symmetry;
S42: adopt the sliding window of 5*5 to detect picture bright spot, draw bright spot numerical value;
S43: image edge average gradient, image edge complexity, as multiple parameters such as factors of influence, draw final video pictures striped evaluation points;
S44: draw evaluation of estimate according to evaluation points characteristic distributions.
The invention has the beneficial effects as follows: possess the detectability under the streak feature loss failure condition under the different scenes in striped fault, varying environment, different striped type, comprise: cord, stria, clear striped, fuzzy striped, horizontal stripe, nicking, slanted bar line, bright fringes, dark fringe etc., by enhancing the detection of streak feature.The present invention from other only adopt a kind of detection algorithm or feature bar detection algorithm different, by the mode that spatial method and time-domain analysis are combined, multiple characteristic parameters of combining image self, and image streak feature is strengthened, finally draw the evaluation of estimate to striped fault.
Method of the present invention is easy to realize and application, is mainly used in:
(1) electric power video equipment failure detection system, by the acquisition to mounted video equipment image, calculated by the online striped assessment of fault value to video image, thus show whether video equipment exists striped fault, thus provide direct basis for video equipment overhauls.
(2) the method is not only applicable to the video equipment fault detect in power industry, is equally applicable to the application of other industry, includes: traffic video, security protection video, bank's video etc.
Accompanying drawing explanation
To do the present invention below in conjunction with the drawings and specific embodiments and further illustrate, above-mentioned and/or otherwise advantage of the present invention will become apparent.
Fig. 1 is FB(flow block) of the present invention.
Embodiment
Each detailed problem involved in the technology of the present invention method is described in detail below in conjunction with accompanying drawing.Be to be noted that described embodiment is only intended to be convenient to the understanding of the present invention, and any restriction effect is not play to it.
The mode of the present invention by spatial method and time-domain analysis are combined, multiple characteristic parameters of combining image self, and image streak feature is strengthened, finally draw the evaluation of estimate to striped fault.Figure 1 shows that the FB(flow block) of the electric power video image striped fault detection method combined based on spatial method and time-domain analysis, the enter factor method that this method is divided into acquisition image edge to extract, image edge extracting method, the conversion of Fourier's frequency domain, video pictures striped evaluation of estimate acquisition methods Four processes.
The enter factor method that described acquisition image edge extracts comprises step: original image is converted to gray scale picture; The part edge of excision gray scale picture, obtains new gray scale picture; New gray scale picture is asked for the complexity of picture; The total average gradient value of the 3*3 pixel of picture is asked for new gray scale picture.
Described image edge extracting method comprises step: the average gradient calculating the 3*3 pixel of each pixel for new gray scale picture; Enter factor is extracted according to image complexity, overall average Grad edge calculation; According to the average gradient value of each pixel, obtain dynamic pixel and strengthen coefficient; According to the average gradient of edge extracting enter factor, dynamic pixel enhancing coefficient, pixel, obtain each pixel value of edge picture.
Described video pictures striped evaluation of estimate acquisition methods draws together step: the picture bright spot after conversion is all distributed in four angles of picture, after being changed, bright spot is all concentrated on central point by symmetry; Adopt the sliding window of 5*5 to detect picture bright spot, draw bright spot numerical value; Image edge average gradient, image edge complexity, as multiple parameters such as factors of influence, draw final video pictures striped evaluation points; Evaluation of estimate is drawn according to evaluation points characteristic distributions.
The hardware minimalist configuration that method of the present invention needs is: the PC of P4,3.0GCPU, 512M internal memory, on the hardware of this configuration level, adopts C/C++ Programming with Pascal Language to realize this method.Operating system can based on each type operating system of Windows or Linux.Describe in detail one by one the committed step of method design of the present invention below, the basic step in method of the present invention is identical, and concrete form is as described below:
First, be the enter factor obtaining image edge extraction:
(1) obtaining original image is placed in Mat_Origin_Pic [] as RGB picture matrix data;
(2) turn gray scale formula according to RGB: Gray=R*0.299+G*0.587+B*0.114, thus calculate Mat_Gray_Pic [];
(3) due to the upper left corner in actual picture or the upper right corner, the lower left corner or the lower right corner all comprise some Word messages usually, avoid Word message on the impact of monitoring, suppose that picture horizontal ordinate is 0<x<W1, picture ordinate is 0<y<H1, H1 wherein represents the height of picture, W1 represents the width of picture, picture horizontal ordinate is intercepted for W1/8<x<7*W1/8, picture ordinate is intercepted for H1/8<y<7*H1/8, obtain gray scale picture matrix M at_Gray2_Pic [], the width of new gray scale picture is W2, be highly H2,
(4) calculate the Grad of each pixel, adopt 3*3 window calculation, Grad (i, j) represents the Grad of each pixel, and Pixel (i, j) represents the gray-scale value of each pixel.Grad (i, j)=Pixel (i, j)-(position is at i, about the pixel of j, the gray-scale value of eight pixels is comprehensive)/8, i span is 1<i<W2-1, j span is 1<j<H2-1;
(5) the total Grad of picture is calculated, the each pixel Grad of Total_Grad=(i, j) Grad summation, i span is 1<i<W2-1, j span is 1<j<H2-1, calculate the average gradient value of picture, Avg_Grad=Total_Grad/ ((W2-2) * (H2-2));
(6) average gray value of picture is calculated, the each pixel Pixel (i of Avg_Gray=(, j) summation of gray-scale value)/((W2-2) * (H2-2)), wherein: 1<i<W2-1,1<j<H2-1;
(7) complexity factors of picture is calculated, Cpx=(normalized) (Avg_Grad*Avg_Gray);
(8) enter factor that the Grad that picture is total, the average gradient value of picture, the average gray value of picture, the complexity factors of picture are extracted as image edge.
Secondly, image edge is extracted:
(1) based on gray scale picture Mat_Gray2_Pic [], being newly defined by the edge picture after edge extracting is Mat_Edge_Pic [], and the size of edge picture is W2*H2;
(2) input value of Grad Grad (i, j) for edge picture of each pixel is calculated;
(3) calculate the Dynamic contrast enhance coefficient of picture edge extracting, Enh=Cpx*Param1, wherein Param1 adjusts according to test case;
(4) the disjunction mode of the Dynamic contrast enhance of picture edge extracting is calculated, adopt different Dynamic contrast enhance systems during different Grad Grad (i, j) namely to each pixel, adjust according to test case, intensity value ranges can be adopted to be 0-10,11-20, and 21-60 is as segmentation;
(5) value being defined by each pixel of the edge picture after edge extracting is Pixel_Edge (i, j), the account form of this value is: Pixel_Edge (i, j)=Enh*Grad (i, j) * Param2*Param3, the value of Param2 adjusts according to test case, Pixel (i, j) when 0-10 scope, Param2 value is 4-8, Pixel (i, j) when 1-20 scope, Param2 value is 2-3, Pixel (i, j) when 21-60 scope, Param2 value is 1-2, Pixel (i, j) when being greater than 60 scope, Param2 value is 1.
(6) by Pixel_Edge (i, j) assignment to edge picture Mat_Edge_Pic [].
Finally, video pictures striped evaluation of estimate is obtained:
(1) frequency domain picture Mat_FFT_Pic [] is obtained after edge picture Mat_Edge_Pic [] being adopted inverse discrete Fourier transform;
(2) picture the Mat_FFT_Pic [] bright spot after conversion is all distributed in four angles of picture, after being changed, bright spot is all concentrated on central point, obtain picture Mat_FFTShift_Pic [] by symmetry;
(3) the sliding window of 5*5 is adopted, calculate the Grad Grad_1 (i of picture Mat_FFTShift_Pic [] each point, j), Grad_2 (i, j), suppose that the value of each pixel of Mat_FFTShift_Pic [] is Pixel (i, j), calculate Grad_1 (i respectively, and the value of Grad_2 (i, j) j):
Grad_1(i,j)=2*Pixel(i,j)-(Pixel(i-1,j)+Pixel(i+1,j)+Pixel(i,j-1)+Pixel(i,j+1))；
Grad_2(i,j)=2*Pixel(i,j)-(Pixel(i-2,j)+Pixel(i+2,j)+Pixel(i,j-2)+Pixel(i,j+2))；
(4) set threshold value Thd, when Grad_1 (i, j) and Grad_2 (i, j) is greater than Thd simultaneously, represents that this point is bright spot, Bright_Grad (i, j) is designated as to bright spot;
(5) Bright_Grad (i is judged, j) distribution mode in Mat_FFTShift_Pic [] figure, when all bright spots present regularity distribution, calculate the quantity Bright_Point_Num of rule distributed points, by this value as the fringes noise failure condition judging video pictures.
Embodiment
Specific embodiment is as follows:
Based on the electric power video image striped fault detection method that spatial method and time-domain analysis combine, comprise the enter factor method of acquisition image edge extraction, image edge extracting method, the conversion of Fourier's frequency domain, video pictures striped evaluation of estimate acquisition methods, step is as follows:
Obtain the enter factor of image edge extraction as the initial treatment to Image Edge-Detection, the rim detection of image does not provide input parameter, gray level image is converted to original video picture, and the data such as the complexity of computed image, average gradient, as the enter factor of Edge extraction.
The enter factor step obtaining image edge extraction is as follows:
Step S11: original image is converted to gray scale picture;
Step S12: the part edge of excision gray scale picture, obtains new gray scale picture;
Step S13: the complexity new gray scale picture being asked for picture;
Step S14: the total average gradient value of the 3*3 pixel of picture is asked for new gray scale picture.
Image edge extracting method is as carrying out spatial transform to original video picture and improving the process of image stripe edge characteristic, adopt image edge enhancement method, sliding window edge extracting method, the enter factor of combining image edge extracting, extract the edge in original image, final generation corresponding edge picture.
Image edge extraction step is as follows:
Step S21: the average gradient calculating the 3*3 pixel of each pixel for new gray scale picture;
Step S22: extract enter factor according to image complexity, overall average Grad edge calculation;
Step S23: according to the average gradient value of each pixel, obtains dynamic pixel and strengthens coefficient;
Step S24: according to the average gradient of edge extracting enter factor, dynamic pixel enhancing coefficient, pixel, obtain each pixel value of edge picture.
The conversion of Fourier's frequency domain is as spatial domain and frequency domain conversion process, and the edge picture after spatial processing, adopts inverse discrete Fourier transform that edge picture is converted to frequency domain picture.
The evaluation of estimate that video pictures striped evaluation of estimate acquires is as the final assessed value of the striped fault to video pictures, picture bright spot after conversion is all distributed in four angles of picture, after being changed by symmetry, bright spot is all concentrated on central point, the sliding window of 5*5 is adopted to detect picture bright spot, draw bright spot numerical value, and in conjunction with image edge average gradient, image edge complexity as multiple parameters such as factors of influence, draw final video pictures striped evaluation points, draw evaluation of estimate according to evaluation points characteristic distributions.
Video pictures striped appraisal procedure is as follows:
Step S41: the picture bright spot after conversion is all distributed in four angles of picture, after being changed, bright spot is all concentrated on central point by symmetry;
Step S42: adopt the sliding window of 5*5 to detect picture bright spot, draw bright spot numerical value;
Step S43: image edge average gradient, image edge complexity, as multiple parameters such as factors of influence, draw final video pictures striped evaluation points;
Step S44: draw evaluation of estimate according to evaluation points characteristic distributions.
In a word, the present invention proposes a kind of electric power video image striped fault detection method combined based on spatial method and time-domain analysis.A large amount of experimental verifications validity of the present invention and stability is carried out by live video picture actual in electric system.The present invention is easy to realize, and stable and reliable for performance.The present invention effectively achieves the detection of the striped fault to video equipment, for the maintenance of electric power video equipment provides foundation, the every application for electric power video is provided more effective technical support.
The above; be only the embodiment in the present invention; but protection scope of the present invention is not limited thereto; any people being familiar with this technology is in the technical scope disclosed by the present invention; the conversion or replacement expected can be understood; all should be encompassed in and of the present inventionly comprise in scope, therefore, protection scope of the present invention should be as the criterion with the protection domain of claims.
Claims (3)
1., based on the electric power video image striped fault detection method that spatial method and time-domain analysis combine, it is characterized in that: it is as follows that the method comprising the steps of:
S1: be converted to gray level image to original video picture, and the complexity of computed image, average gradient, as the enter factor of Edge extraction;
S2: adopt image edge enhancement method, sliding window edge extracting method, the enter factor of combining image edge extracting, extracts the edge in original image, final generation corresponding edge picture;
S3: the edge picture after spatial processing, adopts inverse discrete Fourier transform that edge picture is converted to frequency domain picture;
S4: the picture bright spot after conversion is all distributed in four angles of picture, after being changed by symmetry, bright spot is all concentrated on central point, the sliding window of 5*5 is adopted to detect picture bright spot, draw bright spot numerical value, and in conjunction with image edge average gradient, image edge complexity as factor of influence, draw final video pictures striped evaluation points, draw evaluation of estimate according to evaluation points characteristic distributions.
2. the electric power video image striped fault detection method combined based on spatial method and time-domain analysis according to claim 1, is characterized in that: it is as follows that the method for the enter factor of Edge extraction described in acquisition S1 comprises step:
S11: original image is converted to gray scale picture;
S12: the part edge of excision gray scale picture, obtains new gray scale picture;
S13: the complexity new gray scale picture being asked for picture;
S14: the total average gradient value of the 3*3 pixel of picture is asked for new gray scale picture.
3. the electric power video image striped fault detection method combined based on spatial method and time-domain analysis according to claim 1, is characterized in that: in S2, to comprise step as follows for the extracting method of image border:
S21: the average gradient calculating the 3*3 pixel of each pixel for new gray scale picture;
S22: extract enter factor according to image complexity, overall average Grad edge calculation;
S23: according to the average gradient value of each pixel, obtains dynamic pixel and strengthens coefficient;
S24: according to the average gradient of edge extracting enter factor, dynamic pixel enhancing coefficient, pixel, obtain each pixel value of edge picture.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410525862.6A CN104766297B (en) | 2014-10-08 | 2014-10-08 | A kind of electric power video image striped fault detection method combined based on spatial domain and time-domain analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410525862.6A CN104766297B (en) | 2014-10-08 | 2014-10-08 | A kind of electric power video image striped fault detection method combined based on spatial domain and time-domain analysis |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104766297A CN104766297A (en) | 2015-07-08 |
| CN104766297B true CN104766297B (en) | 2016-03-02 |
Family
ID=53648107
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410525862.6A Active CN104766297B (en) | 2014-10-08 | 2014-10-08 | A kind of electric power video image striped fault detection method combined based on spatial domain and time-domain analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104766297B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105282542B (en) * | 2015-10-21 | 2017-06-30 | 上海熙菱信息技术有限公司 | The detection method and system of abnormal striped in a kind of video image |
| CN106372584B (en) * | 2016-08-26 | 2019-06-11 | 浙江银江研究院有限公司 | A kind of video image mosaic detection method |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103093179A (en) * | 2011-10-28 | 2013-05-08 | 浙江大华技术股份有限公司 | Video strip quantitative calculation method |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4107029B2 (en) * | 2002-09-20 | 2008-06-25 | 富士ゼロックス株式会社 | Image reading device |
| CN102421008A (en) * | 2011-12-07 | 2012-04-18 | 浙江捷尚视觉科技有限公司 | Intelligent video quality detecting system |
-
2014
- 2014-10-08 CN CN201410525862.6A patent/CN104766297B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103093179A (en) * | 2011-10-28 | 2013-05-08 | 浙江大华技术股份有限公司 | Video strip quantitative calculation method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104766297A (en) | 2015-07-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106940886B (en) | A kind of electrical equipment electric discharge ultraviolet imagery quantization parameter extracting method based on gray scale | |
| CN105957088B (en) | Transformer composite insulator casing monitoring method and system based on computer vision | |
| CN106023185B (en) | A kind of transmission facility method for diagnosing faults | |
| CN103487729B (en) | Based on the power equipments defect detection method that ultraviolet video and infrared video merge | |
| CN104318582B (en) | Detection method for bad state of rotating double-lug component pin of high-speed rail contact network | |
| CN106680285B (en) | Method for recognizing insulator contamination state based on infrared image assisted visible light image | |
| CN105447471A (en) | Infrared detection based device gas leakage identification method and apparatus | |
| CN104282011B (en) | The method and device of interference stripes in a kind of detection video image | |
| CN103954897A (en) | Intelligent power grid high voltage insulation damage monitoring system and method based on ultraviolet imagery | |
| CN107831409A (en) | The method and method for detecting abnormality of superfrequency partial discharge detection TuPu method parameter extraction | |
| CN111402215B (en) | Contact net insulator state detection method based on robust principal component analysis method | |
| CN103247044B (en) | Based on the defective mode detection method of high ferro contact net insulator curve-like and point-like singularity characteristics | |
| CN102831393A (en) | Rapid image recognizing method of power tower pole outline | |
| CN105303158A (en) | Line-fitting video intelligent analysis algorithm for disconnecting switch of transformer station | |
| CN104766297B (en) | A kind of electric power video image striped fault detection method combined based on spatial domain and time-domain analysis | |
| CN108009547A (en) | A kind of nameplate recognition methods of substation equipment and device | |
| CN107179479A (en) | Transmission pressure broken lot defect inspection method based on visible images | |
| CN104849180A (en) | Particle image real-time processing system and particle image real-time processing method | |
| CN112666426A (en) | Equipment fault detection method and system based on multispectral three-in-one image | |
| CN108734704A (en) | Based on the normalized transmission line breakage detection technique of gray variance | |
| CN116258282B (en) | Smart grid resource scheduling and distributing method based on cloud platform | |
| CN109919936B (en) | Method, device and equipment for analyzing running state of composite insulator | |
| CN104794702B (en) | A kind of based on the image blurring detection method of electric power video under the special scenes of image complexity analytical approach | |
| CN108573233B (en) | Power grid ceramic insulator identification method based on image processing technology | |
| CN110207783A (en) | A kind of sensed water level method based on video identification |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information |
Address after: 5, building 1, 210003 Pavel Road, Jiangning District, Jiangsu, Nanjing Applicant after: Electric Power Research Institute of Jiangsu Electric Power Company Applicant after: Nanjing Yinshi Software Company Address before: 5, 210003 floor, No. 40, Nanchang Road, Gulou District, Jiangsu, Nanjing Applicant before: Nanjing Yinshi Software Company Applicant before: Electric Power Research Institute of Jiangsu Electric Power Company |
|
| COR | Change of bibliographic data | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |