CN103177241B - The method utilizing video image processing technology location power transmission line spacer - Google Patents

Abstract

The invention discloses and belong to digital video image process and a kind of power transmission line spacer identification utilizing video image processing technology in transmission line of electricity on-line monitoring technique field and localization method.The method by original image is standardized, the field of direction is asked for, the method such as template matching carries out target-region locating and conductor spacer centralized positioning to image.In target-region locating, use squared gradient algorithm to calculate image direction field, to the consistency information, image is carried out transmission line of electricity target-region locating in conjunction with each.Use and carry out conductor spacer centralized positioning based on the Fast template matching algorithm that rectangular histogram is relevant with normalization, algorithm sets Histogram Matching associating threshold value and normalization relevant matches maximum as the foundation judging coupling optimum position, choose, through repeatedly coupling, the coordinate setting conductor spacer central point that in matching result, probability of occurrence is maximum, provide reference information for power transmission state monitoring with fault pre-alarming.

Description

The method utilizing video image processing technology location power transmission line spacer

Technical field

The invention belongs to digital video image process and transmission line of electricity on-line monitoring technique field.Relate to especially And a kind of method utilizing video image processing technology location power transmission line spacer, specifically The identification of digital remote video image when transmission line of electricity is carried out on-line monitoring and location, including Normalization, the field of direction of power transmission line spacer image are asked for, a series of figures such as template matching As treatment technology, it is possible to power transmission line spacer image is automatically processed and identifies, extract The positional information of conductor spacer in transmission line of electricity.

Background technology

Power transmission line spacer is the support equipment between split conductor, makes to protect between the sub-conductor of split conductor Hold certain spacing distance.One of visual plant as transmission line of electricity, conductor spacer is to obtain to lead Bank hangs down, waves and the reliable reference of the state parameter such as growth-gen under line.

At present power transmission line spacer positional information is extracted used technological means, mainly Use theodolite or ground laser radar technology.And existing monitoring means needs staff deep Entering on the spot operation, the bumpy section being difficult to arrive operating personnel is difficult to accurately complete monitoring , and the testing time is long.Use image processing techniques that power transmission line spacer is identified fixed Position is a kind of the most directly perceived, easy method, and Chinese scholars has had begun to correlational study. Wherein, the transmission line of electricity that Wai Ho Li et al. shoots from transmission line of electricity side with ordinary camera Image is goal in research, transmission pressure is considered as straight line and uses template matching method to carry out it Location, in conjunction with the feature of the contact that is connected between conductor spacer and wire in image, uses Gabor filtering Realizing conductor spacer location, this algorithm can obtain ideal for low coverage, the simple image of background Positioning result, but poor effect during and background complexity big in power transmission line sag.Li Junfang etc. People propose be arranged on shaft tower video camera shooting video image as object of study, employing is returned One algorithm changing relevant matches, carries out the coupling location of conductor spacer in conjunction with Pyramid technology search, The method uses both full-pixel normalization Matching primitives, algorithm is computationally intensive, the highest, causes matching algorithm poor in timeliness.

The present invention, based on digital remote video image technology, uses the video camera shooting being arranged on shaft tower Transmission line of electricity live video, to be sent in the transmission line of electricity video flowing of Surveillance center the number intercepted Word image is object of study, and use that a kind of new algorithm realizes power transmission line spacer is automatic Identifying and location, the method positions transmission line of electricity target area by the field of direction, and by rectangular histogram Combine with Normalized Cross Correlation Algorithm, for conductor spacer centralized positioning.This algorithm can be more Add and position the conductor spacer in transmission line of electricity in real time, accurately, for ensureing that power generation safely provides The means of a kind of more precise and high efficiency, have very important realistic meaning.

Summary of the invention

It is an object of the invention to provide one utilizes video image processing technology to position power transmission line spacer Method, it is characterised in that the method is a kind of based on digital remote video image processing technology Power transmission line spacer identification and localization method；By the camera collection being erected on shaft tower Transmission line of electricity video signal, passed through the transmission channel side with video flowing through video server Formula sends back Surveillance center in real time, intercepts from the transmission line of electricity video flowing being sent to Surveillance center Containing the digital picture of conductor spacer, Fast template matching algorithm based on the field of direction is used to carry out defeated Electric line image target area location and conductor spacer coupling location, to simplify the complexity of location algorithm Degree, improves accuracy and the real-time of location, and the extraction for transmission line status information provides ginseng Examine information；Specifically comprise the following steps that

1) by the camera collection transmission line of electricity video signal being erected on shaft tower, from transmission line of electricity Video signal intercepts the original image containing conductor spacer；

2) from original image, conductor spacer template image to be positioned is intercepted；

3) standardize original image successively, the field of direction is asked for, respectively to concordance calculating and region The image procossing of labelling, obtains image in zone marker and is respectively more than or equal to maximum to concordance value The region of 96%；

4) marked region carries out carrying out by the mode of rank scanning the template matching of conductor spacer, with the 2nd step Cut The conductor spacer image obtained is template, and matching factor with Matching sub-image top left corner pixel coordinate is Index, the pixel coordinate choosing maximum match coefficient place is best match position, with optimal The centre coordinate centre spot as conductor spacer of place subimage is put in coordination；

5) repeat the 3rd, 4 steps, after repeatedly processing, mating, choose matching result occurs general The coordinate of rate maximum is conductor spacer anchor point.

Described normalization is that the pixel specification that original image carries out routine processes, and eliminates gradation of image poor Different impact, is to ask for, for the follow-up field of direction, the preprocessing process that carries out.

The described field of direction is asked for being to use squared gradient algorithm, uses image block method to calculate image The gradient vector of block and deflection, and then ask for the image entirety squared gradient field of direction.

Described respectively calculate and zone marker to concordance, be each to concordance numerical value to image direction field Calculate, with image respectively to the 96% of concordance maximum as threshold value, choose each to concordance Value locks more than or equal to the region of this threshold value, and is labeled as target area.

Described template matching is to use the Fast template matching algorithm relevant with normalization based on rectangular histogram, Rectangular histogram is combined threshold value coupling and normalization relevant matches combines, reduce answering of matching primitives Miscellaneous degree.Rectangular histogram associating threshold value matching algorithm is to use Euclidean distance and crossing function that template is straight Side's figure carries out Similarity Measure with the rectangular histogram mating subregion, with improve the real-time of algorithm with Accuracy.

The invention has the beneficial effects as follows and first the transmission line of electricity image section at conductor spacer place is carried out target Zone location, simplifies the scope of coupling location scanning；Then first rectangular histogram associating threshold value is used Joining, the method for the relevant optimal coupling of rear normalization carries out conductor spacer centralized positioning；Last from repeatedly Positioning result is chosen and occurs that the coordinate of maximum probability is best located point.The method ensure that While the real-time of location, reduce the error rate in position fixing process.

Accompanying drawing explanation

Fig. 1 is transmission line of electricity original image.

Fig. 2 is the image after original image normalization processes.

Fig. 3 is the field of direction figure of normalization image.

Fig. 4 is each to concordance figure of transmission line of electricity image.

Fig. 5 is the zone location result figure of transmission line of electricity image.

Fig. 6 is the Histogram Matching result queue figure of transmission line of electricity image.

Fig. 7 is the optimal coupling positioning result figure of transmission line of electricity image.

Fig. 8 is the coupling location three-dimensional table diagram of transmission line of electricity image.

Fig. 9 is the flow chart of transmission line of electricity images match location.

Detailed description of the invention

The present invention provides a kind of method utilizing video image processing technology location power transmission line spacer, The method is a kind of new power transmission line spacer based on digital remote video image processing technology Identify and localization method.The method is the conductor spacer in order to identify and position transmission line of electricity accurately , the digital picture intercepted in the transmission line of electricity video flowing being sent to Surveillance center is as object of study , by the digital picture collected being carried out video procession, extract transmission line of electricity interval Rod center location information, obtains sag for monitoring transmission line of electricity, the running state parameter such as waves and carry Information for reference.Below in conjunction with the accompanying drawings the present invention is further described, and with transmission line of electricity As a example by the processing procedure of conductor spacer image.

One, picture specification

Fig. 1 be power transmission line spacer original image (Fig. 1 top is fixed on conductor spacer on each electric wire, And be the external environment background image of camera collection beyond each electric wire).Adopt to eliminate image The figure caused due to noise or the external environment of sensor (photographic head) self during collection As the difference in gray scale, the first original image to Fig. 1 carry out normalization process, eliminate image ash Difference on degree, Fig. 2 is the image after normalization.Normalization does not change the definition of image, Simply contrast and the gray scale of different target in image are adjusted in a fixing rank, for Subsequent treatment provides a image the most unified Specification.Normalization formulae is:

$G(i,j)=\left\{\begin{array}{c}{M}_0+\sqrt{\frac{{\mathrm{VAR}}_0{(I(i,j)-M)}^2}{\mathrm{VAR}}},I(i,j)>M\\ M_0-\sqrt{\frac{\mathrm{VA}{R}_0{(I(i,j)-M)}^2}{\mathrm{VAR}}},I(i,j)\≤M\end{array}\right.$

Wherein, (i is j) that in original image, (i, j) gray value of coordinate points, M and VAR is respectively image I to I Gray average and variance, M₀And VAR₀It is respectively the average of image and the expected value of variance, G ( I, j) for image pixel after normalization.

Two, transmission line of electricity target-region locating is carried out by the field of direction

In order to simplify the computation complexity of location algorithm, use target-region locating side based on the field of direction Transmission pressure region in image is positioned by method.

The field of direction is to describe image texture direction and the two dimensional surface field of correspondence position thereof.Utilization orientation field The partial structurtes characteristic of image can be obtained, thus realize target-region locating.Based on gradient to The method of amount field can directly utilize the gradient fields of local data, and the statistical nature of comprehensive neighborhood is raw Become the stable field of direction, the concordance generating direction can also be assessed simultaneously.Squared gradient direction It is as follows that step is asked in field:

1) pixel in image G (x, gradient vector [G y) are asked for_x(x,y),G_y(x,y)]^T；

2) image G is divided into the block (herein selecting w=8) that a series of nonoverlapping size is w*w, each piecemeal Gradient vector be:

${[G_{\mathrm{Bx}},G_{\mathrm{By}}]}^T=\left[\begin{array}{c}\underset{x=1}{\overset{W}{\mathrm{\Σ}}}\underset{y=1}{\overset{W}{\mathrm{\Σ}}}{G}_x^2(x,y)-G_y^2(x,y)\\ \underset{x=1}{\overset{W}{\mathrm{\Σ}}}\underset{y=1}{\overset{W}{\mathrm{\Σ}}}2G_x(x,y)G_y(x,y)\end{array}\right]$ ,

In formula, [G_Bx,G_By]^TIt is the gradient vector of piecemeal.

3) Block direction angle θ is asked for by gradient vector:

$\mathrm{\&theta;}=\frac{1}{2}\mathrm{\&pi;}+\frac{1}{2}\left\{\begin{array}{c}\arctan \left(\frac{G_{\mathrm{By}}}{G_{\mathrm{Bx}}}\right),G_{\mathrm{Bx}}\&GreaterEqual;0;\\ \arctan \left(\frac{G_{\mathrm{By}}}{G_{\mathrm{Bx}}}\right)+\mathrm{\&pi;},G_{\mathrm{Bx}}<0\&cap;G_{\mathrm{By}}\&GreaterEqual;0;\\ {\arctan }^{-1}\left(\frac{G_{\mathrm{By}}}{G_{\mathrm{Bx}}}\right)-\mathrm{\&pi;},G_{\mathrm{Bx}};0\&cap;G_{\mathrm{By}}<0.\end{array}\right.$

In piecemeal, the deflection of each pixel is identical, and θ ∈ [0, π).θ is carried out discretization , obtain the direction number of correspondence, thus form the field of direction.

The reliability of Block direction can be weighed by concordance coh of Block direction, the biggest theory of this value Bright piece of inside gradient field of direction concordance is the best.[G in following formula_sx(x,y),G_sy(x,y)]^T It it is picture in block The squared gradient vector of element.

$\mathrm{coh}=\frac{\left|\underset{x=1}{\overset{W}{\mathrm{\&Sigma;}}}\underset{y=1}{\overset{W}{\mathrm{\&Sigma;}}}(G_{\mathrm{sx}}(x,y),G_{\mathrm{sy}}(x,y))\right|}{\underset{x=1}{\overset{W}{\mathrm{\&Sigma;}}}\underset{y=1}{\overset{W}{\mathrm{\&Sigma;}}}\left|(G_{\mathrm{sx}}(x,y),G_{\mathrm{sy}}(x,y))\right|}$

Squared gradient vector is asked for according to the following formula: $\left[\begin{array}{c}{G}_{\mathrm{sx}}(x,y)\\ G_{\mathrm{sy}}(x,y)\end{array}\right]=\left[\begin{array}{c}{G^2}_x(x,y)-{G_y}^2(x,y))\\ 2G_x(x,y)G_y(x,y)\end{array}\right]$

In order to obtain the more specific location information of transmission pressure part, to Fig. 2 travel direction field and respectively to one Cause calculates, and result is the most as shown in Figure 3,4.In Fig. 4, select conductor part one pocket Block direction and image in the Block direction of other parts carry out concordance calculating, select concordance Value is target area higher than the region of maximum 96%, and is labeled, and result is as shown in Figure 5.

Three, the Fast template matching algorithm positioning interval rod relevant with normalization based on rectangular histogram is used

On the basis of target-region locating, the present invention uses based on fast relevant with normalization of rectangular histogram Conductor spacer center is positioned by speed template matching algorithm.

Two-dimensional image information is reduced to one-dimensional process by Histogram Matching algorithm, greatly reduces meter The complexity calculated, but this algorithm accuracy is the highest.For utilizing the feature description image of image, can Statistic histogram by gradation of image:

$p\left(r_k\right)=n_k/n,k=\mathrm{1,2},...,L$

Wherein, total number of image pixels is n, and the excursion of all grey scale pixel values is 1 ~ L, gray value For r_kNumber of pixels be n_k, p (r_k) it is rectangular histogram.Generally use Euclidean distance function or straight Side's figure intersects function to directly Side's figure similarity is measured, and definition is shown below respectively, P in formula_G(i)、P_T(i) difference Represent the rectangular histogram of region to be matched and template.

$M_E(G,T)=\sqrt{\underset{i=1}{\overset{L}{\mathrm{\&Sigma;}}}{[P_G\left(i\right)-P_T\left(i\right)]}^2}$

$Q_E=(G,T)=\frac{\underset{i=1}{\overset{L}{\mathrm{\&Sigma;}}}\min [P_G\left(i\right),P_T\left(i\right)]}{\underset{i=0}{\overset{L}{\mathrm{\&Sigma;}}}{P}_G\left(i\right)}$

Normalized Cross Correlation Algorithm is based on the classic algorithm in Image Matching technology, this algorithm Accuracy is high, but computationally intensive, is defined as follows:

$G_{\mathrm{NCC}}=\frac{\underset{(x,y)\&Element;W}{\mathrm{\&Sigma;}}G(x,y)*T(x,y)}{\sqrt{\underset{(x,y)\&Element;W}{\mathrm{\&Sigma;}}G{(x,y)}^2\underset{(x,y)\&Element;W}{\mathrm{\&Sigma;}}T{(x,y)}^2}}$

In order to make full use of rectangular histogram and the advantage of normalization relevant matches, overcome its defect, the present invention Use the method that both rectangular histogram and normalization relevant matches combine, set rectangular histogram associating threshold Value is relevant with normalization most preferably mates standard, carries out power transmission line spacer fast and accurately Location.Algorithm steps is as follows:

1) read in original transmission line of electricity image, leave in relief area, including artwork each point pixel value, Artwork width, height etc.；

2) intercepting conductor spacer image to be matched from input picture is matching template, calculates its rectangular histogram；

3) input picture is carried out target-region locating, target area is carried out from left to right, to Under search, and be respectively adopted Euclidean distance and crossing function by THE TEMPLATE HYSTOGRAM with mate sub-district The rectangular histogram in territory carries out Similarity Measure；

4) associating threshold alpha=130 of Euclidean distance and crossing function are determined by repetition test, β=0.99, Select more than the pixel of threshold value and be marked；

5) image-region of labelling is normalized relevant matches calculating, matching factor with template image With coupling subregion top left corner pixel coordinate for index, the coordinate points choosing correlation coefficient maximum is Return value, is optimal match point；

6) according to the axisymmetric feature of conductor spacer, the central point of position matching subregion, it is conductor spacer Center.

On the basis of target-region locating, the target area of mark in Fig. 5 is carried out rectangular histogram and slightly mates , return the matching result in threshold value, as shown in Figure 6.Histogram Matching result is normalized Relevant matches, finally realizes the location of conductor spacer as shown in square frame in Fig. 7.The three-dimensional of experimental result Representing figure as shown in Figure 8, optimum position, three-dimensional table diagram display location is accurate and unique.

Fig. 9 is the flow chart of power transmission line spacer coupling location.

The method is the conductor spacer in order to identify and position transmission line of electricity accurately, to be sent in monitoring The digital picture intercepted in the transmission line of electricity video flowing of the heart is object of study, by collecting Digital picture carries out video procession, extracts power transmission line spacer center location information, Obtain sag for monitoring transmission line of electricity, the running state parameter such as wave reference information is provided.

Claims (2)

1. the method utilizing video image processing technology location power transmission line spacer, the method is one Plant power transmission line spacer identification based on digital remote video image processing technology and localization method；Pass through The camera collection transmission line of electricity video signal that is erected on shaft tower, through video server, it is logical by transmission Road sends back Surveillance center in the way of video flowing in real time, from the transmission line of electricity video flowing being sent to Surveillance center Middle intercepting contains the digital picture of conductor spacer, uses Fast template matching algorithm based on the field of direction to carry out defeated Electric line image target area location and conductor spacer coupling location, to simplify the complexity of location algorithm, carry The accuracy of high location and real-time, the extraction for transmission line status information provides reference information；It is special Levy and be, specifically comprise the following steps that

1) by the camera collection transmission line of electricity video signal being erected on shaft tower, believe from transmission line of electricity video The original image containing conductor spacer is intercepted in number；

2) from original image, conductor spacer template image to be positioned is intercepted；

3) standardize original image successively, the field of direction is asked for, respectively to concordance calculating and zone marker Image procossing, zone marker is to calculate, each of image direction field with image to concordance numerical value Each is threshold value to the 96% of concordance maximum, chooses each region being more than or equal to this threshold value to concordance value Lock, and be labeled as target area；Carry out marked region being spaced by the mode of rank scanning Rod template matching, with choose maximum match coefficient place pixel coordinate as best match position；Obtain In zone marker, image is respectively more than or equal to the region of maximum 96% to concordance；Wherein, described normalization is The pixel specification that original image carries out routine processes, and eliminates the impact of gradation of image difference, is for follow-up The field of direction asks for the preprocessing process carried out；

4) to step 3) zone marker carry out carrying out the template matching of conductor spacer by the mode of rank scanning, Template matching is to use the Fast template matching algorithm relevant with normalization based on rectangular histogram, rectangular histogram is joined Close threshold value coupling and normalization relevant matches combines, reduce the complexity of matching primitives, use Euclidean distance With crossing function, THE TEMPLATE HYSTOGRAM is carried out Similarity Measure with the rectangular histogram mating subregion；Again with step 2 from Intercepting the conductor spacer image obtained in original image is template, and matching factor with the pixel coordinate of Matching sub-image is Index, the pixel coordinate choosing maximum match coefficient place is best match position, with best match position The centre coordinate of place subimage is as the centre spot of conductor spacer；

5) repeat step 3,4, after repeatedly processing, mating, choose probability of occurrence in matching result Maximum coordinate is conductor spacer anchor point.

One the most according to claim 1 utilizes transmission line of electricity interval, video image processing technology location The method of rod, it is characterised in that the described field of direction is asked for being to use squared gradient algorithm, and application image divides Block method calculates gradient vector and the deflection of image subblock, and then asks for image entirety squared gradient direction ?.