The content of the invention
To overcome problem present in correlation technique, the present invention to provide a kind of transmission line of electricity based on unmanned plane binocular vision
Ice covering thickness measuring method, to solve prior art in electric power line ice-covering thickness measure inaccurate problem.
In order to solve above-mentioned technical problem, the embodiment of the invention discloses following technical scheme:
This application discloses a kind of electric power line ice-covering thickness measuring method based on unmanned plane binocular vision, including:
Unmanned plane binocular vision Aerial Images are obtained, wherein, the Aerial Images include left and right two width Aerial Images;
Scale invariant features transform SIFT feature Point matching is carried out to the left and right two width Aerial Images;
Judge whether the characteristic point is abundant;
If the SIFT feature point fully, according to the SIFT feature point trigonometric calculations is carried out, obtain tested
The three dimensional space coordinate of object point;
If the SIFT feature point is insufficient, manually add SIFT feature point, according to the SIFT feature point and people
The SIFT feature point of work addition carries out trigonometric calculations, obtains the three dimensional space coordinate of testee point;
According to the three dimensional space coordinate, with reference to unmanned plane positioning and orientation system information and head attitude information, navigated
Image position coordinates are clapped, wherein, the Aerial Images position coordinates is the ground that object is located in the left and right two width Aerial Images
Reason coordinate;
The Aerial Images are carried out with image rate Ratio rim detection, threaded rim is obtained;
Hough lines detections are carried out to the threaded rim, the straight line in the threaded rim is obtained;
Enter row conductor screening to the threaded rim cathetus, obtain wire edge;
Icing identification is carried out according to the wire edge, icing region is marked;
Ice covering thickness measurement characteristic point is carried out according to the icing region to choose;
Ice covering thickness value calculating is carried out according to the Aerial Images position coordinates and ice covering thickness measurement characteristic point.
Alternatively, it is described SIFT feature Point matching is carried out to the Aerial Images to include:
Calculate the SIFT feature point of the Aerial Images;
DOG metric space pyramids are set up, extremum extracting is carried out to the SIFT feature point;
The SIFT feature point position is accurately positioned according to the result of the extremum extracting and SIFT feature point is located
Yardstick;
According to the pinpoint SIFT feature point position and SIFT feature point place yardstick, SIFT feature is calculated
Vector;
Characteristic matching is carried out according to the SIFT feature vector, wherein, the characteristic matching is automated characterization matching.
It is alternatively, described according to the pinpoint SIFT feature point position and SIFT feature point place yardstick,
Calculating SIFT feature vector includes:
According to the pinpoint SIFT feature point position, using the gradient direction distribution of SIFT feature vertex neighborhood pixel
It is characterized as that each SIFT feature point determines directioin parameter;
According to the window calculation SIFT feature vector that 8 × 8 are taken centered on the pinpoint SIFT feature point position.
Alternatively, the data that the SIFT feature vector is included have:Position, yardstick, direction and SIFT description.
Alternatively, it is described to judge whether the characteristic point fully includes:
Obtain the SIFT feature point quantity of automatic characteristic matching;
The SIFT feature point quantity that automated characterization is matched is compared with preset value;
If the SIFT feature point quantity of automated characterization matching is less than the preset value, then it is assumed that the SIFT feature
Point is insufficient;
If the SIFT feature point quantity of automated characterization matching is more than or equal to the preset value, then it is assumed that described
SIFT feature point is abundant.
Alternatively, the artificial addition characteristic point includes:
The same place in the left and right two width Aerial Images is chosen, the coordinate of the same place is obtained;
The coordinate of the same place is manually added into matching, wherein, the same place is the left and right two width Aerial Images
The pixel of the same object that middle different angles or position photograph.
Alternatively, the Mathematical Modeling of the trigonometric calculations is:
Wherein, f1, f2 are left and right two camera focus, and Xi, Yi are the image coordinate in the Aerial Images, and r is spin moment
Battle array, t is translation matrix.
Alternatively, described to enter row conductor screening to the straight line, obtaining wire edge includes:
Determine whether more than or equal to two straight lines and straight line parallel to be detected;
If being not greater than being equal to two straight lines and straight line parallel to be detected, straight line to be detected is not wire;
If more than or equal to two straight lines and straight line parallel to be detected, then judging straight length to be detected whether less than institute
State Aerial Images length;
If straight line to be detected is less than the Aerial Images length, straight line to be detected is not wire;
If straight line to be detected is not less than the Aerial Images length, straight line to be detected is wire.
Alternatively, described to carry out icing identification according to the wire edge, marking icing region includes:
Obtain the color of the wire and brightness on the Aerial Images;
The color of the color and brightness and non-ice coating wire and brightness are compared;
If the color and brightness are identical with the color of non-ice coating wire and brightness, the non-icing of wire is not covered
Ice is marked;
If the color of the color and brightness and non-ice coating wire and brightness are differed, wire icing marks icing
Edge, wherein, the icing edge is the threaded rim that image Ratio rim detections are arrived.
Alternatively, it is described according to the icing region carry out ice covering thickness measurement characteristic point choose include:
Obtain the position at the icing edge;
Choose any point on icing edge side;
Choose on icing edge opposite side with any point along the relative point of straight line.
The technical scheme that embodiments of the invention are provided can include following beneficial effect:The present invention is based on disclosed in implementing
The electric power line ice-covering thickness measuring method of unmanned plane binocular vision obtains left and right Aerial Images using binocular vision technology, right
Left and right Aerial Images carry out the process such as SIFT feature Point matching, trigonometric calculations, Ratio rim detections and Hough transform
The ice covering thickness of wire is calculated, be can solve the problem that because such environmental effects, cost of labor and time cost are high in prior art,
Electric power line ice-covering thickness measures inaccurate problem caused by measurement is inconvenient, effectively improve electrical network the condition of a disaster emergency response when
Effect property.
It should be appreciated that the general description of the above and detailed description hereinafter are only exemplary and explanatory, not
The present invention can be limited.
Specific embodiment
Here exemplary embodiment will be illustrated in detail, its example is illustrated in the accompanying drawings.Explained below is related to
During accompanying drawing, unless otherwise indicated, the same numbers in different accompanying drawings represent same or analogous key element.Following exemplary embodiment
Described in embodiment do not represent and the consistent all embodiments of the present invention.Conversely, they be only with it is such as appended
The example of the consistent apparatus and method of some aspects described in detail in claims, the present invention.
Fig. 1 is a kind of electric power line ice-covering thickness measuring method flow chart based on unmanned plane binocular vision, such as Fig. 1 institutes
Show, the electric power line ice-covering thickness measuring method based on unmanned plane binocular vision includes:
Unmanned plane binocular vision Aerial Images are obtained, wherein, the Aerial Images include left and right two width Aerial Images;
SIFT feature Point matching is carried out to the left and right two width Aerial Images;
Judge whether the characteristic point is abundant;
If the SIFT feature point fully, according to the SIFT feature point trigonometric calculations is carried out, obtain tested
The three dimensional space coordinate of object point;
If the SIFT feature point is insufficient, manually add SIFT feature point, according to the SIFT feature point and people
The SIFT feature point of work addition carries out trigonometric calculations, obtains the three dimensional space coordinate of testee point;
According to the three dimensional space coordinate, with reference to unmanned plane positioning and orientation system information and head attitude information, navigated
Image position coordinates are clapped, wherein, the Aerial Images position coordinates is the ground that object is located in the left and right two width Aerial Images
Reason coordinate;
The Aerial Images are carried out with image Ratio rim detections, threaded rim is obtained;
Hough lines detections are carried out to the threaded rim, the straight line in the threaded rim is obtained;
Enter row conductor screening to the threaded rim cathetus, obtain wire edge;
Icing identification is carried out according to the wire edge, icing region is marked;
Ice covering thickness measurement characteristic point is carried out according to the icing region to choose;
Ice covering thickness value calculating is carried out according to the Aerial Images position coordinates and ice covering thickness measurement characteristic point.
The electric power line ice-covering thickness measuring method based on unmanned plane binocular vision as described in Figure 1, specific implementation method
It is as follows:
Using the Visible Light Camera of two different focals of UAV flight, two cameras are shared the same light axle, and optical axis and longitudinal axis angle
Equal, camera focus are respectively 35mm and 105mm, wherein, 35mm focal lengths viewing field of camera is big, for finding and being aligned transmission line of electricity
Wire, 105mm focal length camera high definitions, for detecting and recognizing wire icing, control synchronization of the unmanned plane to two cameras.Nothing
The man-machine distance away from transmission line of electricity 40-100 rice ensures Aerial Images along planning airline operation, flying speed and filming frequency
Course, sidelapping degree are more than 60%, and the POS data and aspect information during record storage unmanned plane during flying is completed
Unmanned plane binocular vision is taken photo by plane the collection of data.
Selection to be measured the image taken photo by plane containing shaft tower and wire binocular vision of icing, and image is carried out accordingly first
Pretreatment, in image pair SIFT feature point is calculated respectively, and due to Aerial Images high resolution, the depth of field is deeper, background in image
Accounting is great, and wire proportion is few, thus select set up DOG (Difference of Gaussian, DOG, Gaussian function
Difference) metric space pyramid, concrete method for building up is as follows:
In different scale parameter σ consecutive variations, the metric space of G (x, y, σ) pie graph picture, for two-dimensional image I
(x, y), the metric space under different scale represents that L (x, y, σ) can be by image I (x, y) and the convolution of Gaussian kernel G (x, y, σ)
Obtain:
L (x, y, σ)=G (x, y, σ)=G (x, y, σ) * I (x, y),
Wherein, L represents metric space, and (x, y) represents the point on I, and σ is scale factor, and its value is more big, characterizes the image
It is smoothed bigger;Its value is more little, characterizes the image and is smoothed less.Large scale corresponds to the general picture feature of image, little chi
Spend the minutia corresponding to image.Therefore, it is the key for setting up metric space to select suitable scale factor to smooth.
Set up gaussian pyramid:In order to obtain the invariant feature point under different scale space, by image I (x, y) and not
Convolution operation is carried out with Gaussian kernel G (x, y, σ) under the metric space factor, gaussian pyramid is constituted.
Set up DOG metric space pyramids:DOG is the difference of adjacent two metric spaces function, is represented with D (x, y, σ), public
Formula is as described below:
G (x, y, σ)=(G (x, y, k σ)-G (x, y, σ)) * I (x, y)=L (x, y, k σ)-L (x, y, σ)
DOG pyramids are subtracted each other by adjacent metric space function in gaussian pyramid.
In the DOG metric space pyramids of above-mentioned foundation, in order to detect the maximum and minimum of a value in DOG spaces, DOG
In metric space, each pixel of intermediate layer (except the bottom and top) needs 8 neighbor pixels with same layer
And its last layer and the 9 of next layer neighbor pixels altogether 26 neighbor pixels are compared, it is empty in yardstick to guarantee
Between and two dimensional image space all detect local extremum.
The low point of contrast and unstable marginal point are removed so as to accurately determine position and the yardstick at characteristic point place, profit
Directioin parameter is formulated for each characteristic point with the gradient direction distribution characteristic of characteristic point neighborhood territory pixel, makes operator that there is invariable rotary
Property, 8 × 8 window calculation SIFT feature vector is taken centered on characteristic point, characteristic matching is carried out, wherein, characteristic point
The data that SIFT feature vector is included have:Position, yardstick, direction and SIFT description.If characteristic point very little, is needed into pedestrian
Work selected characteristic point, it is ensured that characteristic matching precision, that is, choose the same place in the left and right two width Aerial Images, obtains described
The coordinate of same place, by the coordinate of the same place matching is manually added, wherein, the same place is the left and right two width boat
Clap the pixel of the same object that different angles or position photograph in image.
Then trigonometric calculations are carried out according to the SIFT feature point of the SIFT feature point and artificial addition, is obtained tested
The three dimensional space coordinate of object point, concrete calculation is as follows:
A kind of binocular vision of electric power line ice-covering thickness measuring method based on unmanned plane binocular vision as shown in Figure 2
Three-dimensional coordinate measurement model, if left camera is located at sensor and measures coordinate o-x1y1z1Origin at and without spin, image coordinate system
For O1X1Y1, effective focal length is f1;Right camera coordinates system is o-x2y2z2, image coordinate system is O2X2Y2, effective focal length is f2, by phase
Machine Perspective transformation model has:
And measurement coordinate system o-xyz and o-x1y1z1Correlation space conversion matrices M between coordinate system12It is expressed as:
Wherein, R, T are respectively o-xyz coordinate systems and o-x1y1z1It is flat between spin matrix and origin between coordinate system
The corresponding relation moved between transformation vector, then sensor measurement coordinate system is represented spatial point and 2 camera image planes points is:
Then, the Mathematical Modeling of binocular vision sensor three-dimensional coordinate measurement can be expressed as:
Known focal length f1、f2With spatial point in the magazine image coordinate in left and right, as long as obtaining spin matrix R and translation vector
Amount T can be obtained by the three dimensional space coordinate of testee point.
Coordinate scale information under the image coordinate system of jobbie point is random, is obtained by two testee points
Space coordinates, recycle aspect and head attitude information to resolve and obtain the relative position of two testee points and close
System.
According to the attitude information and the attitude information of head of aircraft, camera can be calculated when shooting left relative to big
The spin matrix of ground coordinate system, the vector of two testee points compositions can be calculated in earth coordinates by spin matrix
Under form, it is as follows:
If aircraft spin matrix is:
The spin matrix of head is:
Then geodetic coordinates is to the spin matrix of image space auxiliary coordinates:
R=Rplane*Rplat
If the vector under the auxiliary coordinates of image space is:
Then the vectorial coordinate under earth coordinates is:
The Aerial Images are carried out with image rate Ratio rim detection, threaded rim is obtained, Ratio operators are Tupin
Deng the line edge detection algorithm based on statistical model of proposition:A given region R comprising n pixeli, each pixel
Gray value is Pk, then region RiAverage gray beFor one fixed width, center pixel is x0Region R1, its
The region on both sides is R2, R3, then detect in R1It is middle to cross x0Both sides are divided into into the straight line of close size, by taking level as an example, Ratio is calculated
The template of son is as shown in figure 3, in center pixel x0Neighborhood open a rectangular window, and mark off as shown in Figure 3 three area
Domain, judges respectively R1And R2, R1And R3Between whether there is edge, if all existed, then center pixel is considered as straight line
On point, the concrete steps of Ratio operators are expressed as follows:
If region R1、R2、R3Pixel average be respectively μ1, μ2, μ3, definition region i and j rim detection receptance function rij:
The Line feature receptance function r of Ratio operators:
R=min (r12,r13)
Wherein, r12And r13Respectively region R1、R2With region R1、R3Between average ratio response, give a threshold value
rth, when r is more than the threshold value, it is believed that this center pixel x0For the pixel on straight line.
First parameter of Ratio operators is the size of template.The width of wire substantially between 1 to 2 pixels, due to
The impact of the various interference such as forest, house, mountain range so that complex background lower wire pixel may be submerged among noise, be
At utmost suppress noise, need to select suitable Ratio operators template size, and consider the increase of template size to amount of calculation
Impact, experiment test can obtain Ratio operator template sizes optimal size be 5 × 5.
The second parameter of Ratio operators is the number in detection of straight lines direction.Unmanned plane edge during patrolling transmission line
The flight of wire oblique upper certain altitude, wire is substantially horizontal in taken image, according to this priori, can be with
Threaded rim detection is carried out only with the Ratio operators template of horizontal direction.
3rd parameter of Ratio operators is threshold value rth.Because the resolution ratio of infrared image is relatively low, background is complicated, it is difficult to
Choose a fixed threshold value and ask for edge, therefore rthWant adapting to image.In general, wire is that line feature is most in image
Obvious region, rthCalculating:The threaded rim of infrared image is calculated using Ratio operators;Ask for the average of edge image;Order
rthEqual to the average of three times.
Hough lines detections are carried out to the threaded rim, the straight line in the threaded rim is obtained, Hough straight lines are carried
Take specific as follows:
The line segment of two-dimensional space differently can carry out guantification description to it, give one group of point (xi,yi) (i=1,
2 ..., n), straight line L can be fitted according to slope and intercept, i.e.,:
yi=mxi+ci
Wherein, m is the slope of straight line L, and c is the intercept of straight line L, and then the straight line L of image space can be expressed as parameter
A point in space (m, c), each point on straight line can be expressed as the straight line of parameter space again.But as L and y
When axle is parallel, slope is infinitely great, cannot represent in parameter space, therefore can represent straight line using polar form, i.e.,
xicosθ+yiSin θ=ρ
Wherein, ρ is distance of the rectangular coordinate system origin to straight line L, and θ is the angle of straight line and x-axis.Thus, through Hough
Conversion, the straight line L of image space is a point in (ρ, θ) space, and the point (x on straight line Li,yi) in (ρ, θ) space it is one
Sine curve, for the set of n given image point obtains n bar ρ-θ curves, most curve intersections in (ρ ', θ '), then these phases
Handing over the point of the corresponding image space of curve can be fitted with straight line L.If there is j parameter curve phase in (ρ, θ) space
Intersection point, then these intersection points correspond to the j bar straight lines in image space.
PPHT (accumulated probability Hough transform) is a kind of classical improved form of Hough transform, can arrange line segment most
Little length, the largest interval between 2 points of same straight line, and the start-stop position of straight line is obtained, therefore, intended using PPHT algorithms
Close wire.
Enter row conductor screening to the threaded rim cathetus, obtain wire edge, including step as described below:
Choose angle after PPHT to existLine segment constitute line-segment sets M1;
Choose M1The maximum line segment L of middle lengthmax;
In M1In carry out line segment connection:Slope is identical, and intercept is identical, and the different two lines section in start-stop position is connected to become
Same slope, the most long line segment of same intercept, obtains line-segment sets M2;
In M2Middle selection is all to be approximately parallel to LmaxLine segment (angle difference existsWithin), it is believed that it is wire, and in front and back
Extend 10 length in pixels as finally detected wire M3。
Icing identification is carried out according to the wire edge, marking icing region includes:
Obtain the color of the wire and brightness on the Aerial Images;
The color of the color and brightness and non-ice coating wire and brightness are compared;
If the color and brightness are identical with the color of non-ice coating wire and brightness, the non-icing of wire is not covered
Ice is marked;
If the color of the color and brightness and non-ice coating wire and brightness are differed, wire icing marks icing
Edge, wherein, icing edge is the threaded rim that image Ratio rim detections are arrived.
Ice covering thickness value calculating is carried out according to the Aerial Images position coordinates and ice covering thickness measurement characteristic point:
Take photo by plane on picture at two, by intelligent Matching or according to artificial visual object point is selected, closed using the constraint of core line
System being capable of indirect labor's reconnaissance;Determine image space auxiliary coordinates yardstick;Can be calculated according to the corresponding POS information of aerial photograph
The shooting point distance of two photos, can determine true between two articles point under the auxiliary coordinates of image space using the actual distance
Actual distance from, the distance as 2 points of clearance, while vertical range between 2 points and horizontal direction can be calculated
The ice covering thickness of distance, i.e. wire.
It should be noted that herein, term " including ", "comprising" or its any other variant are intended to non-row
His property is included, so that a series of process, method, article or equipment including key elements not only include those key elements, and
And also include other key elements being not expressly set out, or also include for this process, method, article or equipment institute inherently
Key element.In the absence of more restrictions, the key element for being limited by sentence "including a ...", it is not excluded that including institute
Also there is other identical element in process, method, article or the equipment of stating key element.
Those skilled in the art will readily occur to its of the present invention after considering specification and putting into practice disclosure of the invention here
Its embodiment.The application is intended to any modification of the present invention, purposes or adaptations, these modifications, purposes or
Person's adaptations follow the general principle of the present invention and including the undocumented common knowledge in the art of the present invention
Or conventional techniques.Description and embodiments are considered only as exemplary, and true scope and spirit of the invention are by following
Claim is pointed out.
It should be appreciated that the precision architecture for being described above and being shown in the drawings is the invention is not limited in, and
And can without departing from the scope carry out various modifications and changes.The scope of the present invention is only limited by appended claim.