CN101950357A - Method for identifying towers, drainage threads and wires of high-voltage line based on position relations - Google Patents
Method for identifying towers, drainage threads and wires of high-voltage line based on position relations Download PDFInfo
- Publication number
- CN101950357A CN101950357A CN2010102900035A CN201010290003A CN101950357A CN 101950357 A CN101950357 A CN 101950357A CN 2010102900035 A CN2010102900035 A CN 2010102900035A CN 201010290003 A CN201010290003 A CN 201010290003A CN 101950357 A CN101950357 A CN 101950357A
- Authority
- CN
- China
- Prior art keywords
- image
- little
- long straight
- lead
- shaft tower
- 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
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a method for identifying towers, drainage threads and wires of a high-voltage line based on position relations, which comprises the following steps: (1) preprocessing an input image of the high-voltage line; (2) respectively extracting horizontal, vertical and inclined small lines on the image subjected to preprocessing; (3) identifying the towers according to the numbers of the horizontal, vertical and inclined small lines in each block on the image; (4) in case the towers do not exist on the image, identifying wires on a class I image; (5) in case the towers exist on the image, generating a curve by a way of small-line fitting on the image, and judging whether the drainage threads exist on the image; (6) in case the drainage threads exist on the image, identifying the wires on a class II image; and (7) in case the drainage threads exist on the image, identifying the wires on a class III image. The method for identifying the wires according to the position relations between the towers, the drainage threads and the wires can eliminate the interference in the background so as to make the identification result more reliable and accurate.
Description
Technical field
The present invention relates to the method for a kind of position-based relation recognition high-tension line shaft tower, drainage thread, lead.This method can be applied to shaft tower, drainage thread and the lead that accurately identifies circuit on infrared image, ultraviolet image, the visible images.
Background technology
Adopt helicopter that high voltage circuit is patrolled and examined and have higher efficient.Have multispectral image capture device by on helicopter, installing, be used for the defective of transmission line of electricity is detected.Because it is big to patrol and examine the acquisition of image data amount, adopts human eye analyzing and diagnosing circuit image defective fully at every turn, working strength is very big, can cause the omission or the erroneous judgement of defective simultaneously because of the fatigue of human eye.Therefore the method that adopts dysnusia to diagnose is analyzed the image of collection automatically, finds the suspicious defective in the circuit, effectively reduces the artificial data volume of diagnosing, and then reduces the intensity of manual analysis, improves the efficient and the reliability of line upkeep simultaneously.The basis of dysnusia diagnosis at first is the automatic identification of each parts of high-tension line on the image, mainly studies the identification of shaft tower, drainage thread, three parts of lead here, secondly is on the basis of component identification parts to be diagnosed.
The document of existing high-tension line component identification is less, and relates generally to the identification of lead.People such as Guangjian Yan at first use the radon conversion to extract line segment from line of electric force in " Automatic Extraction of Power Lines From Aerial Images " literary composition, the method of use group then connects these line segments, use Kalman filter that these line segments are connected into complete lead at last, thereby realize the identification of lead; People such as Zhengrong Li at first use the neural wave filter of pulse coupling to reject ground unrest in " Towards automatic power line detection for a UAV surveillance system using pulse coupled neural filter and an improved Hough transform " literary composition, produce edge clearly, use Hough change detection straight line, in the Hough space, use line clustering method based on knowledge, improve the hough conversion, and then obtain the result of identification.The problem that these two kinds of recognition methodss exist is: two kinds of methods have been used radon conversion and hough conversion, so calculated amount is bigger, owing to the line object on the high-tension line image under the lowered in field environment is more, mean more to be not suitable for using hough or radon algorithm simultaneously; Secondly, because these two kinds of methods are only discerned lead according to the characteristic of lead itself,, will not judge into lead by accident with the similar background interference of wire characteristic easily, cause follow-up generation wrong diagnosis in conjunction with the position relation recognition lead of miscellaneous part on the high-tension line and lead.
Summary of the invention
In order to solve the problems of the technologies described above, the object of the present invention is to provide the method for a kind of position-based relation recognition overhead high voltage line shaft tower, drainage thread, lead.This method can accurately be discerned shaft tower, drainage thread and lead, unique characteristics by making full use of shaft tower, drainage thread, three parts of lead and position concern to be discerned, calculated amount is big when having overcome because of use hough, radon conversion, and get rid of the parts self character and the interference of the background brought, reduced recognizer complexity, improved the accuracy of identification.
For achieving the above object, the present invention adopts following technical scheme:
The method of a kind of position-based relation recognition high-tension line shaft tower, drainage thread, lead, its step is as follows:
(1), the overhead high voltage line image to input carries out pre-service;
(2), on pretreated image, extract respectively level, vertical, little line segment tilts;
(3), judge whether there is the tower bar in the image with level, distribution vertical, the little line segment quantity that tilts in each piecemeal of image.Image vertically is divided into piece, has level, vertical, the little line segment that tilts in each piecemeal respectively,, judge whether there is shaft tower in the image with the quantity of level in each piecemeal in the image, vertical, the little line segment that tilts.If there is not shaft tower in the image, then the image when pre-treatment is a first kind image, then changes step (4); If there is shaft tower on the image, then change step (5);
(4), do not have shaft tower on the image, then present image is a first kind image, and the method that adopts little line segment to follow the tracks of, merge in first kind image generates long straight line, discerns lead in long straight line;
(5), there is shaft tower on the image, select the little line segment of the little segment of curve of match according to the position relation of drainage thread and shaft tower, adopt the little segment of curve of little line-fitting, follow the tracks of, merge the little segment of curve formation curve of match section, in the segment of curve that generates, discern drainage thread, judge whether have drainage thread, if there is no drainage thread in the image, then image is the second class image, changes step (6); If there is drainage thread, then image is the 3rd class image, then skips to step (7);
(6), do not have drainage thread on the image, in the second class image, adopt little line segment to follow the tracks of, merge and generate long straight line, in long straight line, discern lead;
(7), have drainage thread on the image, in the 3rd class image, adopt little line segment to follow the tracks of, merge and generate long straight line, in long straight line, discern lead.
In the described step (1) the high-tension line image being carried out pre-service comprises: the gray processing of image, with Prewitt operator extraction image border, with the image behind the comprehensive two-value method binaryzation edge extracting, the image after adopting linear refinement to binaryzation carries out refinement at last.
Described step (6) is discerned lead in long straight line: in the long straight line that generates, according to slope long straight line is included in each sets of parallel, find two groups of parallel lines, one group of parallel lines adjacent image left side wherein, another group adjacent image the right, if two groups of parallel lines have intersecting area, and the angle of two groups of parallel lines is when being the obtuse angle, and then these two groups of parallel lines are the leads that identify in long straight line in the second class image.
Described step (7) is discerned lead in long straight line: in the long straight line that generates, according to slope long straight line is included in each sets of parallel, have intersecting area between the drainage thread that finds one group of parallel lines and identified, then these group parallel lines are the leads that identify in long straight line in the 3rd class image.
The beneficial effect that position-based relation recognition high-tension line shaft tower of the present invention, drainage thread, the method for lead compared with prior art have is: it is according to discerning that this method utilizes the position between shaft tower, drainage thread, the lead to close, can get rid of the interference in the background, make that the result of identification is more reliable, accurate.This method can be applied on infrared image, ultraviolet image, the visible images, accurately identifies shaft tower, drainage thread and the lead of circuit.Further accurate in thermal-induced imagery, intelligent diagnostics goes out thermal defect on the circuit; At the discharge defect of diagnosing out on the ultraviolet image on the circuit; Diagnosing out defectives such as disconnected strand, foreign matter on the lead adheres on the visible images.
Description of drawings
Fig. 1 is the pictorial diagram of first kind image that runs through the lead of entire image;
Fig. 2 is shaft tower, lead to have occurred and is the pictorial diagram of the second class image of tangent tower circuit;
Fig. 3 is shaft tower, drainage thread, lead to have occurred and is the pictorial diagram of the 3rd class image of anchor support circuit;
Fig. 4 is the process flow diagram of the method for position-based relation recognition high-tension line shaft tower of the present invention, drainage thread, lead;
Fig. 5 is in the first kind image, and lead enters from the image left side, and the right is left, and runs through the synoptic diagram of entire image;
Fig. 6 is in the first kind image, and lead enters from image bottom, and the top is left, and runs through the synoptic diagram of entire image;
Fig. 7 is in the first kind image, and lead enters from image bottom, and the right is left, and runs through the synoptic diagram of entire image;
Fig. 8 is in the first kind image, and lead enters from the image left side, and the top is left, and runs through the synoptic diagram of entire image;
Fig. 9 is a lead identification synoptic diagram in the second class image;
Figure 10 is the lead identification synoptic diagram of shaft tower when the image left side in the 3rd class image;
Figure 11 is the lead identification synoptic diagram when shaft tower is in the middle of image in the 3rd class image;
Figure 12 is the lead identification synoptic diagram of shaft tower when image the right in the 3rd class image;
Figure 13 is the process flow diagram of drainage thread identification in the step of the present invention (5).
Embodiment
Below in conjunction with the drawings and specific embodiments embodiments of the invention are described in further detail.
Present embodiment is to be the explanation that example is carried out with the true high-tension line visible images that collects from helicopter, and the resolution of image is 1280*762.The recognizer of high-tension line shaft tower, drainage thread, lead is designed to the dynamic link libraries of standard, one group of input-output function of definition in the storehouse.Initialization comprises: allocation space, definition picture format, the threshold value that more default follow-up judgements need be used; Deinitialization comprises: the space of distributing when being released in initialization; Call recognition function, input picture, output diagnostic result; Be applied in airborne real-time online diagnostic system, afterwards in the line defct analytic system.
Image is divided three classes: the lead of entire image has only appearred running through in first kind image as shown in Figure 1 among the figure; The second class image as shown in Figure 2, shaft tower, lead have occurred among the figure, shaft tower is a tangent tower; The 3rd class image as shown in Figure 3, shaft tower, drainage thread, lead have occurred among the figure, shaft tower is an anchor support.
As shown in Figure 4, be the method for position-based relation recognition high-tension line shaft tower of the present invention, drainage thread, lead, its step is as follows:
(1), the overhead high voltage line image to input carries out pre-service
The overhead high voltage line visible images that reads in is carried out pre-service, and its step comprises: image gray processing, with Prewitt operator extraction image border, with the method two value edge image of comprehensive binaryzation, adopt image after the method refinement binaryzation of linear refinement at last.
(2), on pretreated image, extract respectively level, vertical, little line segment tilts
On the pretreated image according to image on attribute extraction level respectively, vertical, the little line segment that tilts of each pixel be: at first on the intact image of pre-service according to each pixel and its 8 field in the position relation of each pixel be defined as horizontal pixel point, vertical pixel point, isolated pixel point respectively, extract respectively then level, vertically, little line segment tilts.The step of extracting is as follows:
(2-1), on the intact image of pre-service the little line segment of extraction level, its concrete steps are as follows:
(2-1-1), to find attribute in image be the pixel of horizontal pixel point, is that starting point begins search with this point, and this point is designated as seed.If there is the horizontal pixel point, then change step (2-1-2); Otherwise change step (2-1-5);
(2-1-2), search for the right side, bottom right, upper right three directions of seed respectively, if wherein there is the horizontal pixel point on any one direction, then continue to note the current direction of search simultaneously for to the right at the enterprising line search of this direction, still to the bottom right, still to upper right;
(2-1-3), when not having horizontal pixel point on the direction of search that current searched pixel is being noted, stop search, with the starting point of seed as the little line segment of level that extracts, pixel when stopping with search is a terminal point, calculate the slope of little line segment, intercept, length, and define the information that little line segment structure is used for preserving the little line segment of extraction, this structure comprises:
1) starting point of little line segment;
2) little line segment terminal point;
3) little line segment slope;
4) little line segment intercept;
5) little line segment length;
6) which image block little line segment drops in.
The definition of little line segment structure is expressed as with false code:
Segment{
Point?startP；
Point?endP；
Double?slope；
Double?intercept；
Int?length；
Int?locate；
}
(2-1-4), get back to step (2-1-1), the extraction of the little line segment of continuation level.
(2-1-5), not have the attribute of not searching in the image be the pixel of horizontal pixel point, stops the extraction of the little line segment of level.
(2-2), on the intact image of pre-service, extract vertical little line segment, its concrete steps adopt and the essentially identical step of above-mentioned steps (2-1), just " horizontal pixel point " in the above-mentioned steps (2-1) changed into " vertical pixel point ", extract vertical little line segment through step (2-2-1), (2-2-2), (2-2-3), (2-2-4), (2-2-5);
(2-3), on the intact image of pre-service, extract little line segment, its concrete steps adopt and the essentially identical step of above-mentioned steps (2-1), just " horizontal pixel point " in the above-mentioned steps (2-1) are changed into " isolated pixel point " and extract little line segment through step (2-3-1), (2-3-2), (2-3-3), (2-3-4), (2-3-5);
(3), judge whether there is shaft tower in the image, its concrete steps are as follows with level, distribution vertical, the little line segment quantity that tilts in each image block in the image:
(3-1), image vertically is divided into 10, add up the quantity of level in each image block, vertical, the little line segment that tilts respectively.
The information that definition image block structure is used for preserving each image block, this structure comprises:
1) quantity of the little line segment of level in the piecemeal;
2) quantity of vertical little line segment in the piecemeal;
3) quantity of the little line segment of piecemeal medium dip;
4) whether piecemeal is the sign in shaft tower zone.
The definition of image block structure is expressed as with false code:
BlockInfo{
Int?Hnum；
Int?Vnum；
Int?Snum；
BOOL?Tower_Area；
}
(3-2), calculate the average and the variance of level in each piecemeal, vertical, the little line segment quantity that tilts, be expressed as with false code:
Hmean＝1/10*∑BlockInfo
i.Hnum (1)
Vmean＝1/10*∑BlockInfo
i.Vnum (2)
Smean＝1/10*∑BlockInfo
i.Snum (3)
Hvar＝1/10*∑(BlockInfo
i.Hnum-Hmean)
2 (4)
Vvar＝1/10*∑(BlockInfo
i.Vnum-Vmean)
2 (5)
Svar＝1/10*∑(BlockInfo
i.Snum-Smean)
2 (6)
Wherein Hmean, Vmean, Smean are the average of level on the entire image, vertical, the little line segment quantity that tilts,
H var, V var, S var are the variance of level on the entire image, vertical, the little line segment quantity that tilts.
If during simultaneously greater than default three threshold values, then there is shaft tower in the variance of the little line segment of level, vertical little line segment, the little line segment quantity that tilts in the image.The false code that this deterministic process adopts is expressed as follows:
if(Hvar＞α
1&&Vvar＞α
2&&Svar＞α
3)
Tower_Exist＝1 (7)
else
Tower_Exist＝0
α wherein
iBe threshold value.
If when (3-3) having shaft tower, then judge shaft tower place piecemeal according to level, quantity vertical, the little line segment that tilts in the piecemeal.If certain piecemeal satisfies in the image:
1) difference of the average of the quantity of the little line segment of level and the little line segment quantity of entire image level is greater than the variance of the little line segment quantity of entire image level;
2) difference of the average of the vertical little line segment quantity with entire image of the quantity of vertical little line segment is greater than the variance of the vertical little line segment quantity of entire image;
3) the tilt quantity of little line segment and entire image tilts the difference of average of little line segment quantity greater than the tilt variance of little line segment quantity of entire image.
This Rule of judgment is expressed as follows with false code:
BlockInfo
i.Hnum-Hmean＞Hvar (8)
BlockInfo
i.Vnum-Vmean＞Vvar (9)
BlockInfo
i.Snum-Smean＞Svar (10)
At this moment, then think shaft tower in this image block, the field Tower_Area among the partitioned organization BlockInfo of shaft tower place is changed to TRUE.
(3-4), be foundation with the result of shaft tower identification, judge that the image when pre-treatment still is second, third class image for first kind image.If there is no image is the first kind image during shaft tower, continues step (4); If image is the second class image or the 3rd class image when having shaft tower, skips to step (5) and continue to handle.
(4), do not have shaft tower on the image, then the image when pre-treatment is a first kind image, follows the tracks of, merges little line segment and generate long straight line on image, discerns lead in long straight line,
Define long linear structure:
1) long straight line starting point;
2) long straight line terminal point;
3) long straight slope;
4) long straight line intercept;
5) long straight length;
6) being used for the long straight line of mark belongs to the field of which sets of parallel.
The definition of long linear structure is expressed as with false code:
line{
Point?startP；
Point?endP；
Double?slope；
Double?intercept；
Int?length；
Int?Group；
}
Definition sets of parallel structure:
1) group number of sets of parallel;
2) the bar number of straight line in the sets of parallel;
3) starting point of sets of parallel;
4) terminal point of sets of parallel;
5) slope of sets of parallel;
6) be used for the mark sets of parallel whether near the field of image left hand edge;
7) be used for the mark sets of parallel whether near the field of image right hand edge;
8) be used for the mark sets of parallel whether near the field of image lower limb;
9) be used for the mark sets of parallel whether near the field of image coboundary.
The definition of sets of parallel structure is expressed as with false code:
Parallel{
Int?Group；
Int?size；
Point?startP；
Point?endP；
Double?slope；
Int?left_edge；
Int?right_edge；
Int?bottom_edge；
Int?up_edge；
}
Set up a long straight line object chained list, be designated as Power_line, be used for depositing tracking, the long straight line after merging.A newly-built sets of parallel chained list is designated as Parallel, is used for depositing the identical long straight line object of slope among the Power_line.
Follow the tracks of, merge little line segment and generate long straight line on image, the concrete steps of identification lead are as follows in long straight line:
(4-1), follow the tracks of respectively, the merging level, vertical, little line segment tilts
(4-1-1), tracking, the little line segment of merging level, its step is as follows:
(4-1-1-1), at first take out a little line segment of level in the little line segment of the level that extracts, be designated as seed_line, this little line segment need satisfy condition: the length field length in the little line segment structure is not-1.After taking-up, need seed_line.length is changed to-1.If there is such straight-line segment, then forwards step (4-1-1-2) to, otherwise forward step (4-1-1-4) to.
Seek the little line segment of next bar level (4-1-1-2) and then in the little line segment of the level that has extracted, be designated as curr_line, this little line segment satisfies following condition: 1) the x coordinate of starting point is greater than the x coordinate of seed_line terminal point; 2) the curr_line.length field is not-1.With the starting point of seed_line, the terminal point of curr_line is two end points of a new straight line, calculates the slope of new straight line, is expressed as follows with false code:
Calculate two intercepts with the terminal point of this slope, seed_line and the starting point of curr_line, be expressed as follows with false code:
int?ercept
1＝seed_line.endP.y-seed_line.endP.x*newslope (12)
int?ercept
2＝curr_line.startP.y-curr_line.startP.y*newslope (13)
If above-mentioned two little line segment conllinear, so above-mentioned two intercepts should be identical, is expressed as with false code:
|intercept
1-intercept
2|＜1 (14)
If seed_line and curr_line meet above-mentioned collinear condition, then two straight lines are merged, the terminal point of seed_line is reset to the terminal point of curr_line, recomputate length, slope and the intercept of seed_line simultaneously.The length field of curr_line is changed to-1.
If (4-1-1-3) in the little line segment of the level of having extracted, still have the little line segment of level that satisfies condition in the step (4-1-1-1), then return step (4-1-1-2), otherwise finish this secondary tracking, seed_line is added in the structure, and this structure is used for depositing tracking, the long straight line object after merging.Handle if also exist the little line segment of level not carry out following the tracks of among the Hsegment, then return step (4-1-1-1).Otherwise change step (4-1-1-4).
(4-1-1-4), there is not the straight-line segment that does not have tracked processing in the Hsegment structure.Finish the tracking of the little line segment of all levels.
(4-1-2), follow the tracks of, merge vertical little line segment, step is with (4-1-1), just " the little line segment of level " in the step (4-1-1) made into " vertical little line segment ", finish tracking, the merging of vertical little line segment through step (4-1-2-1), (4-1-2-2), (4-1-2-3), (4-1-2-4).
(4-1-3), follow the tracks of, merge to tilt little line segment, step is with (4-1-1), just " the little line segment of level " in the step (4-1-1) made into " little line segment tilts ", finish tracking, the merging of little line segment through step (4-1-3-1), (4-1-3-2), (4-1-3-3), (4-1-3-4).
(4-2), discern lead in long straight line, its concrete steps are as follows:
(4-2-1) sets of parallel among the statistics Power_line.The long straight line that slope among the Power_line is identical is included into same sets of parallel, the Group field of the long straight line object of correspondence is changed to same value, and in Parallel the record this group parallel lines information, with long vertical element among the Parallel count size not the long straight line in 10～4 scopes from Power_line, delete.
(4-2-2), judge whether every group of parallel lines among the Parallel run through entire image.If the sets of parallel among the Parallel can satisfy in the following condition of representing with false code any one, then these group parallel lines are the lead on the first kind image, and these group parallel lines run through entire image, and the condition that following false code is represented is:
(1)Parallel
i.startP.x→0，Parallel
i.endP.x→width
or
(2)Parallel
i.startP.y→0，Parallel
i.endP.y?→height
or (17)
(3)Parallel
i.startP.y→height，Parallel
i.endP.x→width
or
(4)Parallel
i.startP.x→0，Parallel
i.endP.y→0
Finish the lead identification when not having shaft tower on the image.
(5), there is shaft tower on the image, select the little line segment of the little segment of curve of match according to the position relation of drainage thread and shaft tower, adopt the little segment of curve of little line-fitting, follow the tracks of, merge the little segment of curve formation curve of match section, in the segment of curve that generates, discern drainage thread, judge whether have drainage thread, if there is no drainage thread in the image, then image is the second class image, changes step (6); If there is drainage thread, then image is the 3rd class image, then skips to step (7);
Definition segment of curve structure:
1) end points of the little line segment of composition segment of curve;
2) being used for mark forms in the little line segment of segment of curve whether comprise vertical little line segment;
3) being used for mark forms in the little line segment of segment of curve whether comprise the little line segment of level;
4) being used for mark forms in the little line segment of segment of curve whether comprise to tilt little line segment.
The false code of segment of curve organization definition is expressed as:
Curve{
List?Point；
Int?Vexist；
Int?Hexist；
Int?Sexist；
}；
Adopting the mode of little line-fitting to generate little segment of curve in image, follow the tracks of, merge little segment of curve formation curve section, is that example describes with the drainage thread of discerning the shaft tower left side, in like manner can discern the drainage thread on shaft tower the right.
As shown in figure 13, its concrete steps are as follows:
(5-1), choose those according to the position of shaft tower and be positioned at the shaft tower left side, and the little line segment of position next-door neighbour's shaft tower is used for the little segment of curve of match.
(5-2), come the little segment of curve of match with selecteed little line segment, concrete steps are as follows:
(5-2-1), in selecteed little line segment, find the vertical little line segment of article one, be designated as seed, be seed with seed, in these little line segments, find the vertical little line segment of second again, be designated as seg, seed and seg satisfy condition:
1) the x coordinate of the terminal point of seed is less than the x coordinate of the starting point of seg;
2) distance between the starting point of the terminal point of seed and seg is less than certain threshold value;
3) interior angle of seed and seg between 135 to 180 the degree between,
Be expressed as follows with false code:
seg.startP.x＞seed.endP.x
dis(seg.startP，seed.endP)＜δ (18)
135°＜diag(seg，seed)＜180°
(5-2-2), merge seed and seg, the end points of seed and seg is deposited into interim curvilinear structures, this curvilinear structures is designated as Vcurve, the Vexist field with Vcurve is changed to 1 simultaneously.Be new seed again with seg, in the selected little line segment that comes out, search again and whether still exist little line segment to satisfy the condition that above-mentioned false code is represented with seg.If there is so vertical little line segment, then the end points with the vertical little line segment that satisfies condition adds among the Vcurve.Can carry out identical match operation to the selected little line segment of level that comes out, the little line segment that tilts equally, the result of match is stored in respectively among Hcurve, the Scurve;
(5-3), the little segment of curve among Vcurve, Hcurve, the Scurve is followed the tracks of, merged, the formation curve section.Its concrete steps are as follows:
(5-3-1), from Vcurve, take out a curve (being designated as curve1), search and whether in Hcurve, have curve (being designated as curve2), little line segment of the last item in the little line segment of the feasible curve1 of composition and the satisfied condition of using false code (18) expression of the little line segment of article one in the little line segment of forming curve2, if there is such segment of curve, then above-mentioned two little segment of curve are merged, the end points of forming the straight-line segment of curve2 is added among the curve1, simultaneously the Hexist among the curve1 is changed to 1;
(5-3-2), equally, in Scurve, search whether there is curve and curve1 curve altogether again;
(5-3-3), will follow the tracks of the result who merges little segment of curve is saved in the curve chained list (being designated as FCurve) at every turn;
(5-4), the definition drainage thread is made up of vertical little line segment, the little line segment of level, the little line segment that tilts, in the segment of curve that generates, discern drainage thread, if there is curve among the FCurve, this curve is made up of vertical little line segment, the little line segment of level, the little line segment that tilts respectively, and then this curve is a drainage thread;
(5-5), judge whether there is drainage thread in the image, concrete steps are as follows:
If (5-5-1) do not have drainage thread among the FCurve, the image that does not then have drainage thread is the second class image, continues step (6);
If (5-5-2) have drainage thread among the FCurve, the image that then has drainage thread is the 3rd class image, skips to step (7).
(6), do not have drainage thread on the image, adopt little line segment to follow the tracks of, merge and generate long straight line in the second class image, discern lead in long straight line, be elaborated with reference to Fig. 9, its concrete steps are as follows:
(6-1), tracking, merging level, vertical, the little line segment that tilts generate long straight line object, and long straight line are divided into groups the formation sets of parallel according to slope;
(6-2), calculate the starting point of every group of sets of parallel, the x of terminal point, the average of y coordinate, be recorded in the Origin And Destination field of corresponding Parallel structure.In the Parallel structure, judge these group parallel lines and the image relation at four edges up and down according to x, the y coordinate of starting point and terminal point.Be described below with false code:
(1)Parallel
i.startP.x→0
(2)Parallel
i.startP.y→0
(3)Paralel
i.startP.y→height (19)
(4)Parallel
i.endP.x→width
(5)Parallel
i.endP.y→height
Simultaneously there is one group of sets of parallel to satisfy above-mentioned false code condition (4) again if exist one group of sets of parallel to satisfy above-mentioned false code condition (1), distance between while two groups of sets of parallel is less than certain threshold value, in Fig. 9, it is in the red square frame of δ that the starting point of the terminal point of those group parallel lines of the left side and those group parallel lines of the right has all dropped on the length of side, when the angle of two groups of sets of parallel was the obtuse angle, then these two groups of leads were the lead that identifies in long straight line on the second class image;
(7) there is drainage thread on the image, on the 3rd class image, adopts the mode of little line-fitting to generate long straight line, in long straight line, discern lead.Describe with reference to Figure 10,11,12, its concrete steps are as follows:
(7-1), follow the tracks of, merge three kinds little line segments and generate long straight line objects, will long straight line classifying according to slope obtains sets of parallel;
(7-2), according to the position of shaft tower, determine that the position relation between lead and the drainage thread is as follows:
If (7-2-1) shaft tower is when the left side of image, in sets of parallel, search one group of parallel lines on shaft tower the right, and and drainage thread between have intersecting area, then these group parallel lines are lead.As shown in figure 10, lead begins to extend until the right hand edge of image from the near end of distance shaft tower in Figure 10, having intersecting area away from shaft tower between an end far away and nearly shaft tower one end of lead at drainage thread, is that red box indicating drainage thread and the lead of δ exists intersecting area with the length of side in the drawings;
If (7-2-2) shaft tower is when image middle, in sets of parallel, search one group of parallel lines on the shaft tower left side, and and the drainage thread on the shaft tower left side between have intersecting area; Search another group parallel lines on shaft tower the right, and and there is intersecting area between the drainage thread on shaft tower the right, then these two groups of parallel lines are lead. as shown in figure 11, the lead on the shaft tower left side enters from the image left hand edge and extends to drainage thread away from shaft tower one end always in Figure 11, and there is intersecting area between lead and the drainage thread, be that red box indicating drainage thread and the lead of δ exists intersecting area with the length of side in the drawings, the situation on shaft tower the right is identical;
If (7-2-3) shaft tower is when image the right, in sets of parallel, search one group of parallel lines on the shaft tower left side, and and there is intersecting area between the drainage thread on the shaft tower left side, these group parallel lines are lead, as shown in figure 12, the lead on the shaft tower left side enters from the left hand edge of image and extends to the end of drainage thread away from shaft tower always in Figure 12, has intersecting area with drainage thread, is that red box indicating drainage thread and the lead of δ exists intersecting area with the length of side in the drawings.
Claims (4)
1. the method for a position-based relation recognition high-tension line shaft tower, drainage thread, lead is characterized in that this method step is as follows:
(1), the overhead high voltage line image to input carries out pre-service;
(2), on pretreated image, extract respectively level, vertical, little line segment tilts;
(3), judge whether there is the tower bar in the image with level, distribution vertical, the little line segment quantity that tilts in each piecemeal in the image, image vertically is divided into piece, have respectively in each piecemeal level, vertical, little line segment tilts, quantity with level in each piecemeal in the image, vertical, the little line segment that tilts, judge and whether have shaft tower in the image, if there is not shaft tower in the image, then the image when pre-treatment is a first kind image, then changes step (4); If there is shaft tower on the image, then change step (5);
(4), do not have shaft tower on the image, then present image is a first kind image, and the method that adopts little line segment to follow the tracks of, merge in first kind image generates long straight line, discerns lead in long straight line;
(5), there is shaft tower on the image, select the little line segment of the little segment of curve of match according to the position relation of drainage thread and shaft tower, adopt the little segment of curve of little line-fitting, follow the tracks of, merge the little segment of curve formation curve of match section, in the segment of curve that generates, discern drainage thread, judge whether have drainage thread, if there is no drainage thread in the image, then image is the second class image, changes step (6); If there is drainage thread, then image is the 3rd class image, then skips to step (7);
(6), do not have drainage thread on the image, in the second class image, adopt little line segment to follow the tracks of, merge and generate long straight line, in long straight line, discern lead;
(7), have drainage thread on the image, in the 3rd class image, adopt little line segment to follow the tracks of, merge and generate long straight line, in long straight line, discern lead.
2. the method for position-based relation recognition high-tension line shaft tower according to claim 1, drainage thread, lead is characterized in that: in the described step (1) the high-tension line image is carried out pre-service and comprise: the gray processing of image, with Prewitt operator extraction image border, with the image behind the comprehensive two-value method binaryzation edge extracting, the image after adopting linear refinement to binaryzation carries out refinement at last.
3. the method for position-based relation recognition high-tension line shaft tower according to claim 1, drainage thread, lead, it is characterized in that: described step (6) is discerned lead and is in long straight line: in the long straight line that generates, according to slope long straight line is included in each sets of parallel, find two groups of parallel lines, one group of parallel lines adjacent image left side wherein, another group adjacent image the right, if two groups of parallel lines have intersecting area, and when the angle of two groups of parallel lines was the obtuse angle, then these two groups of parallel lines were the leads that identify in long straight line in the second class image.
4. the method for position-based relation recognition high-tension line shaft tower according to claim 1, drainage thread, lead, it is characterized in that: described step (7) is discerned lead and is in long straight line: in the long straight line that generates, according to slope long straight line is included in each sets of parallel, find one group of parallel lines and the drainage thread that has identified that intersecting area is arranged, then these group parallel lines are the leads that identify in long straight line in the 3rd class image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010102900035A CN101950357B (en) | 2010-09-21 | 2010-09-21 | Method for identifying towers, drainage threads and wires of high-voltage line based on position relations |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010102900035A CN101950357B (en) | 2010-09-21 | 2010-09-21 | Method for identifying towers, drainage threads and wires of high-voltage line based on position relations |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101950357A true CN101950357A (en) | 2011-01-19 |
| CN101950357B CN101950357B (en) | 2012-11-07 |
Family
ID=43453852
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2010102900035A Expired - Fee Related CN101950357B (en) | 2010-09-21 | 2010-09-21 | Method for identifying towers, drainage threads and wires of high-voltage line based on position relations |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101950357B (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102353355A (en) * | 2011-06-14 | 2012-02-15 | 西安工程大学 | Method for measuring power transmission line pole and tower inclination based on video differences |
| CN102508110A (en) * | 2011-10-10 | 2012-06-20 | 上海大学 | Texture-based insulator fault diagnostic method |
| CN102930280A (en) * | 2012-10-05 | 2013-02-13 | 中国电子科技集团公司第十研究所 | Method for identifying overhead high-voltage wire automatically from infrared image |
| CN103020601A (en) * | 2012-12-13 | 2013-04-03 | 北京航空航天大学 | Method and device for high voltage wire visual detection |
| CN104835152A (en) * | 2015-04-27 | 2015-08-12 | 国家电网公司 | Processing method and system of power transmission line inspection images |
| CN104867134A (en) * | 2015-05-04 | 2015-08-26 | 国家电网公司 | Identification method for transmission line tower inspection by unmanned aerial vehicle |
| CN106355580A (en) * | 2016-09-22 | 2017-01-25 | 云南电网有限责任公司电力科学研究院 | Detection method and device of tilting pole tower |
| CN107767366A (en) * | 2017-09-22 | 2018-03-06 | 全球能源互联网研究院有限公司 | A kind of transmission line of electricity approximating method and device |
| CN107784652A (en) * | 2017-10-30 | 2018-03-09 | 广东电网有限责任公司机巡作业中心 | A kind of shaft tower quick determination method based on unmanned plane image |
| CN107886096A (en) * | 2016-09-29 | 2018-04-06 | 成都思晗科技股份有限公司 | A kind of insulator of transmission line of electricity comes off defect inspection method |
| CN108334844A (en) * | 2018-02-06 | 2018-07-27 | 贵州电网有限责任公司 | A kind of automatic tracking method along the line of polling transmission line |
| CN108734687A (en) * | 2017-04-21 | 2018-11-02 | 游诚曦 | A kind of lateral conductor does not stress defect identification method and device |
| CN110865978A (en) * | 2019-11-06 | 2020-03-06 | 国网福建省电力有限公司漳州供电公司 | Intelligent digital image management system for transmission and distribution network equipment |
| CN112380944A (en) * | 2020-11-06 | 2021-02-19 | 中国电力科学研究院有限公司 | Method and system for evaluating structural state of transmission tower |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101625723A (en) * | 2009-07-02 | 2010-01-13 | 浙江省电力公司 | Rapid image-recognizing method of power line profile |
| CN101793501A (en) * | 2010-04-14 | 2010-08-04 | 华中科技大学 | Transmission line ice coating status detection method based on image |
| CN101806888A (en) * | 2010-03-19 | 2010-08-18 | 清华大学 | Image processing-based high-tension line identification method |
-
2010
- 2010-09-21 CN CN2010102900035A patent/CN101950357B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101625723A (en) * | 2009-07-02 | 2010-01-13 | 浙江省电力公司 | Rapid image-recognizing method of power line profile |
| CN101806888A (en) * | 2010-03-19 | 2010-08-18 | 清华大学 | Image processing-based high-tension line identification method |
| CN101793501A (en) * | 2010-04-14 | 2010-08-04 | 华中科技大学 | Transmission line ice coating status detection method based on image |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102353355A (en) * | 2011-06-14 | 2012-02-15 | 西安工程大学 | Method for measuring power transmission line pole and tower inclination based on video differences |
| CN102508110A (en) * | 2011-10-10 | 2012-06-20 | 上海大学 | Texture-based insulator fault diagnostic method |
| CN102508110B (en) * | 2011-10-10 | 2014-07-09 | 上海大学 | Texture-based insulator fault diagnostic method |
| CN102930280A (en) * | 2012-10-05 | 2013-02-13 | 中国电子科技集团公司第十研究所 | Method for identifying overhead high-voltage wire automatically from infrared image |
| CN103020601A (en) * | 2012-12-13 | 2013-04-03 | 北京航空航天大学 | Method and device for high voltage wire visual detection |
| CN103020601B (en) * | 2012-12-13 | 2015-08-05 | 北京航空航天大学 | Hi-line visible detection method and device |
| CN104835152A (en) * | 2015-04-27 | 2015-08-12 | 国家电网公司 | Processing method and system of power transmission line inspection images |
| CN104867134A (en) * | 2015-05-04 | 2015-08-26 | 国家电网公司 | Identification method for transmission line tower inspection by unmanned aerial vehicle |
| CN104867134B (en) * | 2015-05-04 | 2018-08-10 | 国家电网公司 | A kind of recognition methods for unmanned plane inspection electric power line pole tower |
| CN106355580A (en) * | 2016-09-22 | 2017-01-25 | 云南电网有限责任公司电力科学研究院 | Detection method and device of tilting pole tower |
| CN106355580B (en) * | 2016-09-22 | 2020-07-14 | 云南电网有限责任公司电力科学研究院 | Method and device for detecting toppling of tower |
| CN107886096A (en) * | 2016-09-29 | 2018-04-06 | 成都思晗科技股份有限公司 | A kind of insulator of transmission line of electricity comes off defect inspection method |
| CN108734687B (en) * | 2017-04-21 | 2020-04-28 | 游诚曦 | Method and device for identifying unstressed defects of diagonal cables |
| CN108734687A (en) * | 2017-04-21 | 2018-11-02 | 游诚曦 | A kind of lateral conductor does not stress defect identification method and device |
| CN107767366A (en) * | 2017-09-22 | 2018-03-06 | 全球能源互联网研究院有限公司 | A kind of transmission line of electricity approximating method and device |
| CN107784652A (en) * | 2017-10-30 | 2018-03-09 | 广东电网有限责任公司机巡作业中心 | A kind of shaft tower quick determination method based on unmanned plane image |
| CN107784652B (en) * | 2017-10-30 | 2021-08-10 | 广东电网有限责任公司机巡作业中心 | Tower rapid detection method based on unmanned aerial vehicle image |
| CN108334844A (en) * | 2018-02-06 | 2018-07-27 | 贵州电网有限责任公司 | A kind of automatic tracking method along the line of polling transmission line |
| CN110865978A (en) * | 2019-11-06 | 2020-03-06 | 国网福建省电力有限公司漳州供电公司 | Intelligent digital image management system for transmission and distribution network equipment |
| CN112380944A (en) * | 2020-11-06 | 2021-02-19 | 中国电力科学研究院有限公司 | Method and system for evaluating structural state of transmission tower |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101950357B (en) | 2012-11-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101950357B (en) | Method for identifying towers, drainage threads and wires of high-voltage line based on position relations | |
| CN104036323B (en) | A kind of vehicle checking method based on convolutional neural networks | |
| CN103366374B (en) | Based on the passageway for fire apparatus obstacle detection method of images match | |
| CN101539991B (en) | Effective image-region detection and segmentation method for iris recognition | |
| CN102854191B (en) | Real-time visual detection and identification method for high speed rail surface defect | |
| CN102332089B (en) | Railway wagon brake shoe key going-out fault recognition method based on artificial neural network | |
| CN104091171B (en) | Vehicle-mounted far infrared pedestrian detecting system and method based on local feature | |
| CN110068818A (en) | The working method of traffic intersection vehicle and pedestrian detection is carried out by radar and image capture device | |
| CN102998316B (en) | Transparent liquid impurity detection system and detection method thereof | |
| CN106780486A (en) | A kind of Surface Defects in Steel Plate image extraction method | |
| CN103366156A (en) | Road structure detection and tracking | |
| CN103832433A (en) | Lane departure and front collision warning system and achieving method thereof | |
| CN103473530A (en) | Adaptive action recognition method based on multi-view and multi-mode characteristics | |
| CN105260749A (en) | Real-time target detection method based on oriented gradient two-value mode and soft cascade SVM | |
| CN102902974A (en) | Image based method for identifying railway overhead-contact system bolt support identifying information | |
| CN108009574B (en) | Track fastener detection method | |
| CN105426894A (en) | Image detection method and device for railroad plug nails | |
| CN105354529A (en) | Vehicle converse running detection method and apparatus | |
| CN104915642A (en) | Method and apparatus for measurement of distance to vehicle ahead | |
| Li et al. | Recognizing white line markings for vision-guided vehicle navigation by fuzzy reasoning | |
| CN107545244A (en) | Speed(-)limit sign detection method based on image processing techniques | |
| CN111476804A (en) | Method, device and equipment for efficiently segmenting carrier roller image and storage medium | |
| CN103679183A (en) | Defect identification method for plain weave gray cloth | |
| Shafie et al. | Smart video surveillance system for vehicle detection and traffic flow control | |
| Song et al. | High-speed Railway Fastener Detection and Localization Method based on convolutional neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20121107 Termination date: 20150921 |
|
| EXPY | Termination of patent right or utility model |