CN108985143B - Method for identifying iron tower structure of overhead transmission line based on unmanned aerial vehicle image - Google Patents
Method for identifying iron tower structure of overhead transmission line based on unmanned aerial vehicle image Download PDFInfo
- Publication number
- CN108985143B CN108985143B CN201810443089.7A CN201810443089A CN108985143B CN 108985143 B CN108985143 B CN 108985143B CN 201810443089 A CN201810443089 A CN 201810443089A CN 108985143 B CN108985143 B CN 108985143B
- Authority
- CN
- China
- Prior art keywords
- line segment
- line
- line segments
- tower
- common
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Astronomy & Astrophysics (AREA)
- Remote Sensing (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method for identifying an iron tower structure of an overhead transmission line based on an unmanned aerial vehicle image, which is characterized by comprising the following steps of: 1) extracting line segments in different directions from the unmanned aerial vehicle inspection image and describing the attributes of the line segments; 2) detecting and classifying connection characteristics between adjacent multiple line segments; 3) clustering line segments with different connection characteristics into local contour characteristics of triangles and quadrangles contained in the iron tower; 4) and clustering and combining the local contour characteristics into a structure which conforms to the tower head, the tower body and the tower legs of the power transmission line iron tower. The method can effectively identify the structure of the iron tower of the overhead transmission line, and is suitable for identifying the structure of the iron tower under the condition that the complete outline of the iron tower and the shielding between the outline and the structure of the iron tower cannot be detected.
Description
Technical Field
The invention relates to the technical field of unmanned aerial vehicle image recognition, in particular to a method for recognizing an overhead transmission line iron tower structure based on an unmanned aerial vehicle image.
Background
In recent years, the technology of a multi-rotor unmanned aerial vehicle is rapidly developed, the multi-rotor unmanned aerial vehicle is widely applied to the fields of agricultural plant protection, environmental monitoring, security protection, electric power inspection and the like, and especially the unmanned aerial vehicle inspection can greatly improve the maintenance efficiency of a high-voltage transmission line. However, when the unmanned aerial vehicle flies in a business mode, higher requirements are provided for ground operators, and when the unmanned aerial vehicle patrols and examines an electrified power transmission line iron tower, the unmanned aerial vehicle is operated to fly safely, camera equipment installed on the unmanned aerial vehicle is further operated to effectively shoot the iron tower, and the unmanned aerial vehicle is used for working with higher operation difficulty.
When the multi-rotor unmanned aerial vehicle flies at low altitude, various artificial facilities on the ground, such as power transmission line towers, railways, bridges, roads, buildings and the like, can be shot. The railway, the bridge and the highway have obvious long parallel linear characteristics and can be effectively distinguished from iron towers and buildings. The local structures of the iron tower and the building have similar line characteristics, so that the position area of the iron tower or the building cannot be correctly identified; on the other hand, images shot by the unmanned aerial vehicle are greatly influenced by complex background textures and light rays, and complete edge profiles cannot be extracted frequently, so that the structure of the iron tower cannot be identified correctly.
The iron tower is a 3D line structure of fretwork, and the iron tower of arbitrary angle shooting still has more serious sheltering from between the line structure, further causes the difficulty of discerning the iron tower structure. T Dutta, H Sharm, A Vellaiappan. Image Analysis-Based Automatic Detection of Transmission powers Using an axial image.7 th Iberian Conference Pattern Recognition and Image Analysis (IbPRIA), JUN 17-19,2015. Olivier Steiger, Erwan Lucas, Yannick Mark. Andre Fischer, Thomas h.kolbe, Felicitas Lang, etc. extracting building components from architectural Images Using high effective Aggregation in 2D and 3D. computer Vision and Image understanding.1998(11),72(2) proposes a component model of the roof shape of a building, proposes a four-level clustering method of feature layers, feature clustering layers, building component layers, building model layers to detect the 3D structure of the building. Xiaoobai Liu, Yibian Zhao, Song-Chun Zhu, Single-View 3D Scene matching by configured Grammar, IEEE Conference on Computer Vision and Pattern recognition.2014, a three-layer tree structure syntax description attribute diagram is proposed to identify the artificial Scene with a rectangular shape, and four inference rules (arrangement, nesting, arrangement of multi-face cubes, net shape or arrangement relationship in a slice structure) are described based on features to understand the content of the artificial Scene. Under the condition of relatively complete extraction of bottom-layer features, the method can obtain a relatively good result through the inference rule. In the identification method of the artificial scene, a Gestalt perception theory and local feature detection play an important role. The Gestalt perception theory is used for analyzing stable and regular simplified structural features of artificial objects, Chang Cheng, Andrea Koschan, Chung-Hao Chen, etc. outer Scene Image Segmentation Based on Background Recognition and Percentual organization IEEE Transactions on Image Processing,2012(3),21(3), and for identifying artificial objects in natural scenes, analyzing the arrangement relationship (similarity, symmetry relationship, arrangement relationship, strong and weak dependence relationship and approximation) among the segmented regions to detect artificial facilities. Cheng and bin, zai chao, ding ming jump, zhou cheng ping airport object recognition based on knowledge reasoning [ J ] infrared and laser engineering, 2011(3),40(3), recognizing building areas and straight line structures of airports expressed as 'japanese' font to recognize structures of simple man-made objects by sensing line segment arrangement characteristics. Vitroio Ferrari, Loic Fevrier, Fre 'de' ric Juriee, Cordelia Schmid group of Adjacent content Segments for Object Detection, IEEE Transactions On Pattern Analysis and Machine understanding, 2008(1),30(1) studies local shape features consisting of k connected approximate line segment profiles (KAS) for Object class Detection. When k equals 2 or 3, 35 common local features are generated, but not only these local features are present in the artificial object recognition. For the identification of 3D artificial Objects, V.Verma, R.Kumar, S.Hsu.3D Building Detection and Modeling from Aerial LIDAR Data, IEEE Computer Society Conference on Computer Vision and Pattern Recognition,2006. Building a 3D Object model and a roof Structure model from a roof topology, and then constructing a closed roof Structure diagram to identify a Building or 3D Object, R.Wessel, R.Klein.Learning the Building Structure model of Man-Made Objects for 3D Shape Retrieval [ C ] 3rd EUROGRAPHGRAPHPHYSKshop on 3D Object Retrieval,2010(5) from feature selection and similarity measurements, identifying artificial Objects by using a method of feature bags to establish similarity matches between features. These methods all require complete contour features and do not fully consider complex background texture interference and severe occlusion conditions like iron tower structures.
When the unmanned aerial vehicle patrols and examines every time, some redundant images can exist, such as motion blurred images shot by the unmanned aerial vehicle during maneuvering, camera angles deviate from shot images of power transmission lines or long-distance power transmission lines, and the images are ineffective images for defect detection. The invention aims to enable an unmanned aerial vehicle to automatically shoot an iron tower target by accurately identifying the iron tower structure in a natural scene.
Disclosure of Invention
In order to overcome the defects in the prior art, the invention provides a method for identifying the iron tower structure of the overhead transmission line based on an unmanned aerial vehicle image, and solves the problem that the iron tower structure of the overhead transmission line cannot be effectively detected at present.
In order to achieve the above purpose, the invention adopts the following technical scheme: a method for identifying an iron tower structure of an overhead transmission line based on an unmanned aerial vehicle image is characterized by comprising the following steps:
1) extracting line segments in different directions from the unmanned aerial vehicle inspection image and describing the attributes of the line segments;
2) detecting and classifying connection characteristics between adjacent multiple line segments;
3) clustering line segments with different connection characteristics into local contour characteristics of triangles and quadrangles contained in the iron tower;
4) and clustering and combining the local contour characteristics into a structure which conforms to the tower head, the tower body and the tower legs of the power transmission line iron tower.
The method for identifying the iron tower structure of the overhead transmission line based on the unmanned aerial vehicle image is characterized by comprising the following steps of: the step 1) of extracting line segments in different directions from the unmanned aerial vehicle inspection image and describing the attributes of the line segments comprises the following specific steps:
firstly, graying an unmanned aerial vehicle inspection image to generate a binary image;
secondly, constructing line segments into vector line segments on the generated binary image, and expressing the line segments in different directions as follows: l ═ l1,l2,....,lnN is the total number of line segments, and any line segment liThe attributes are described as: (c)i,si,ei,θi,Li,oi) I is a line serial number, i is 1 to n,
is a line segment liThe coordinates of the center point are calculated,
is a line segment liThe coordinates of the starting point,
is a line segment liCoordinates of the end point, θiIs a line segment liAngle, line segment liLength Li，oiLine segment l is (h, v, o +, o-)iThe orientation of (1); the line segment angle is-10 degrees to 10 degrees and is regarded as a horizontal line segment which is expressed as h; less than-75 ° and greater than-90 ° or greater than 75 ° and less than 90 ° are considered as vertical segments, denoted v; the angle of 10-75 degrees is regarded as an oblique upper line segment and is expressed as o +; the angle of-75 DEG to-10 DEG is regarded as a downward slope line segment and is represented as o-.
The method for identifying the iron tower structure of the overhead transmission line based on the unmanned aerial vehicle image is characterized by comprising the following steps of: the unmanned aerial vehicle inspection image is grayed, a binary image is generated, and the method specifically comprises the following steps:
and processing the inspection image by using 8 Prewitt operators in different directions, extracting edge information of the inspection image, generating a binary image by a maximum inter-class binary difference method, and managing all line segments in different directions by using a Blob communication structure.
The method for identifying the iron tower structure of the overhead transmission line based on the unmanned aerial vehicle image is characterized by comprising the following steps of: the step 2) of detecting and classifying the connection characteristics between the adjacent multiple line segments comprises the following specific steps:
in a line segment set, a line segment l is readiGo through any other line segment ljJ is 1 to n, and first, the line segment l is calculatediStarting point and line segment ljWhether the collineation exists or not is judged by the following method:
if d(s)i,sj)||d(si,ej) T is ≦ T, wherein: d(s)i,sj) Is a line segment liStarting point and line segment ljThe distance of the starting point; d(s)i,ej) Is a line segment liStarting point and line segment ljThe distance of the end point, T is a distance threshold value, the value range of T is more than or equal to 0 and less than or equal to 64, and | represents the relation of OR; and if two line segments li、ljThe length of (a) satisfies:
0.3≤Trif the value is less than or equal to 1, the line segment liAnd line segment ljHaving co-connected endpoints, wherein: line segment liIs expressed as LiLine segment ljIs expressed as Lj；
Collecting line segment liThe common endpoint line segment for all start points is represented as: cs(i) (ii) a Collecting line segment liThe common endpoint line segment for all end points is represented as: ce(i) The common-endpoint line segments are classified according to the following rules:
firstly, detecting two adjacent L-shaped line sections: line segment liCommon endpoint set C ofs(i)、Ce(i) If the following condition is satisfied, the line segment l is detectediWith adjacent line segmentsThe interval is L-shaped, and the line segment with the same end point:
1) if c iss(i)＝0&&ce(i) Not equal to 0 or cs(i)≠0&&ce(i) Line segment l is 0iOnly one end point has a common connecting line segment;
2) all and line segments liThe line segment with the same end point is connected with the starting point or the ending point;
3) line segment liThe angle difference L between the common endpoint line segment and all the common endpoint line segmentsθThe range of (A) is as follows: l is not less than 75 degreesθLess than or equal to 105 degrees; if line segment liThe angle difference L between the common endpoint line segment and all the common endpoint line segmentsθThe range of (A) is as follows: theta is more than or equal to 25 degreesdiffJudging that the two adjacent line segments are in a shape of angle less than or equal to 75 degrees, regarding the triangular structure as a combination of two structures with common edges, enabling the end points of the two structures with non-common edges to meet intersection conditions, and clustering the line segments meeting the conditions into the triangular structure;
II, detecting U-shaped three-phase adjacent line sections:
line segment liCommon endpoint set C ofs(i)、Ce(i) If the following condition is satisfied, the line segment l is detectediAnd a U-shaped common-end line segment is arranged between the adjacent line segments:
1) if c iss(i)≠0&&ce(i) Not equal to 0, line segment liThe starting end point and the ending end point are provided with a common connecting line segment;
2) all and line segments liThe line segment with the same end point is connected with the starting point or the ending point;
3) line segment liDegree difference theta between all line segments at common end pointdiffThe range of (A) is as follows: theta is more than or equal to 75 degreesdiff≤105°；
Third, the detection of Y-shaped line segment, if it is connected with line segment liThe number of the common end point line segments is more than two, at least two included angles between every two adjacent line segments are more than or equal to 90 degrees in the clockwise direction, and then the Y-shaped node model is judged;
fourth, detecting the line segment with the point K, if it is the line segment liThe number of the line segments at the common end point is more than two, and at least two included angles between every two adjacent line segments are less than 90 degrees in the clockwise direction, thenJudging as a K-type node model;
fifthly, cross X-shaped line segment detection is carried out, and any two line segments l are assumedi、ljThe orientations are respectively:
and is
And is
And is
And is
Two by two line segments of these four cases, ojIs a line segment ljThe direction of (2), calculating the intersection r of the two line segments (r ═ r)x,ry)，rx,ryRespectively is the abscissa and the ordinate of the intersection point of the two line segments;
if the following conditions are met:
and
and is
And
two line segments li、ljThe length of (a) satisfies:
wherein: t is more than or equal to 0.3 and less than or equal to 1, and then the two line segments are classified into X-shaped cross shapes;
sixthly, detecting a star-shaped structure, wherein the central points of all line segments forming the structure are intersected at a point P, the point P is taken as an endpoint to re-divide the line segments forming the structure, any line segment always has another collinear line segment taking the point P as the endpoint,
if line segment li,lj,lk.., satisfying:
in the formula, the condition (1) indicates that the midpoints of all the line segments intersect, | | ci,cjI represents a line segment li，ljEuclidean distance between center points, ci,cj,ckRespectively represent line segments li，lj，lkCoordinate of center point of (1), TcRepresents a distance threshold; condition (2) indicates that the minimum to maximum length ratio of all line segments is greater than the threshold ratio of 0.5, Li,Lj,LkRespectively represent line segments li，lj，lkWill satisfy all line segments l of the conditioni，lj，lk..S(S,Llines,Pcenter,Pends) Wherein S is a star type identifier indicating that the line segment set is a star type identifier, LlinesTo satisfy segment numbering of star type, PcenterIs the aggregate center of the points in all line segments, PendsThe start and stop points of all line segments.
The method for identifying the iron tower structure of the overhead transmission line based on the unmanned aerial vehicle image is characterized by comprising the following steps of: the step 3) clusters line segments with different connection characteristics into local contour characteristics of triangles and quadrangles contained in the iron tower, and the specific steps comprise:
iron tower IIIClustering the angular local contour features, and identifying the triangle Tr1,Tr2.., further identifying the relationship between the triangles includes the steps of:
1): with the current triangle Tr1For reference, searching a triangle structure T with collinear edges and common vertices with the current triangler2And classified as follows:
a) triangle Tr1、Tr2There are common edges, both forming a parallelogram or trapezoid;
b) triangle Tr1、Tr2Collinear edges exist, but are not equal, and the areas are not coincident;
c) triangle Tr1、Tr2Common edges exist and the areas coincide;
2): repeating the step 1) until all the triangle structures are verified;
3): the triangular line segments which meet the three conditions in the step 1) are gathered together.
Secondly, clustering the local contour features of quadrangles in the iron tower, if two side lines of two U-shaped parts are the same or two triangles share a line segment, combining the two side lines into a quadrangle, and identifying the quadrangle Q1,Q2.., further identifying the relationship between quadrangles, the method comprises the following steps:
1) with the current quadrilateral Q1As a reference, the search satisfies with the current quadrangle:
a) quadrilateral Q1、Q2One group of opposite sides are collinear, the other group of opposite sides are parallel, and the areas are not overlapped;
b) quadrilateral Q1、Q2No public edge or vertex exists, four edges are respectively correspondingly parallel, and the area satisfies
Namely, the ratio of the smaller area to the larger area is less than 0.8;
c) quadrilateral Q1、Q2No public edge or vertex exists, four edges are respectively correspondingly parallel, and the area satisfies
d) Quadrilateral Q1、Q2A common vertex exists, two edges sharing the vertex are correspondingly collinear, and the other two edges are correspondingly parallel;
2) the classified output meets the four types of quadrilateral combinations, and all quadrilateral structures are traversed;
3) and (4) clustering and merging the quadrilateral line segments meeting the spatial relationship of the classes a) to d).
The method for identifying the iron tower structure of the overhead transmission line based on the unmanned aerial vehicle image is characterized by comprising the following steps of: and 4) clustering and combining the local contour characteristics again to form a structure which conforms to the tower head, the tower body and the tower legs of the power transmission line iron tower, and the concrete steps comprise:
the method specifically comprises the following steps of:
the set of quadrilateral structures identified in step 3): q1,Q2.., set of triangle structures: t isr1,Tr2.., identifying the tower body surface structure F of the iron tower1,F2.., the concrete steps include:
1) searching a long vertical line section on the inspection image, judging whether the searched long vertical line section accords with a trapezoidal relation, judging whether the angle of the leftmost long line section accords with positive 70-90 degrees, judging whether the angle of the rightmost long line section accords with negative 90-negative 70 degrees, wherein the angle is the angle of the slope of an image pixel coordinate fitting outlet line section, and taking the long line section which accords with the trapezoidal relation as an outermost support line segment A, D of the tower body of the iron tower;
2) according to the 3D hollow structure of the tower body, the tower body surface is overlapped and shielded, and the tower body support comprises a plurality of K-shaped nodes SK1The tower body surface comprises a plurality of typical internal star nodes SS2(ii) a Searching rightwards from the leftmost support A of the image, and further searching a group of quadrangles with no overlapping area and collinear upper and lower opposite sides, wherein when the number of the quadrangles meeting the condition is more than 2, the line segments contained in the quadrangles belong to an internal star node SS2Line segments simultaneously belonging to a K-shaped node SK1Fitting out straight line from lower to upper of image coordinate based on K-shaped nodeA line segment; if two straight line segments are fitted, setting the straight line segments as the line segments of the support inside the tower body, and taking the external polygon of the quadrangle as the surface of the iron tower, namely the surface F of the iron tower1Bracket line segment is A, B, iron tower surface F2The stent line segment is A, C, A is the overlapping leftmost collinear stent line segment;
3) based on the steps 1) and 2), searching the rightmost stent line segment D in the image leftwards, searching the quadrangle group connected with the common vertex, and further searching the line segments contained in the quadrangle to belong to the internal star node SS2Line segments simultaneously belonging to a K-shaped node SK2Fitting a straight line segment from the lower part of the image coordinate to the upper part based on the K-shaped node; if two straight line segments are fitted, setting the straight line segments as tower body internal support line segments, and if the fitted internal support line segments are superposed with the internal support line segments fitted in the step 2), identifying a 3D tower body structure and a tower surface F3Bracket line segment is C, D, iron tower surface F4The stent line segment is B, D, D is the overlapping rightmost collinear stent line segment;
4) when the triangular structure on the straight line of the tower support is analyzed, triangles at the end points of the iron tower boundary A, D are in common and coincident, triangles at the end points of the tower body inner support B, C are in common and non-coincident, and the triangular areas represent base points of tower feet of the iron tower.
Secondly, the concrete steps of the identification of the tower head of the iron tower comprise:
in the identified quadrilateral structure set Q1,Q2.., set of triangle structures Tr1,Tr2.., set of tower body and surface structures F1,F2.., identifying the tower head cross arm structure Cr1,Cr2.., the steps include:
1) the method comprises the steps that a parallelogram or trapezoidal triangle formed by a common side is used as a basis for identifying a tower head cross arm, if more than three triangles are included in a trapezoidal set formed by adjacent triangles, a circumscribed polygonal area of the triangular set is used as an iron tower head cross arm Cr1；
2) Quadrangles without common edges or vertexes, with parallel edges and unequal areas are taken as the basis for identifying the tower head cross arm, and if the four edges meet the conditionsThe number of the shapes is more than 3, and the quadrilateral circumscribed area meeting the conditions is taken as a tower head cross arm area Cr1；
3) Searching line segment and tower surface F1,F2.. triangle with collinear boundaries, if the triangle is not in the tower body plane, it is used as the tower head cross arm area Cr2。
The invention achieves the following beneficial effects: the method can effectively identify the iron tower structure of the overhead transmission line, and in order to distinguish different areas where buildings and the iron tower are located, starting from line characteristics of the iron tower and the building structure, common connection areas among line segments in different directions are used as connection characteristic points of artificial objects (the iron tower or the buildings), the connection characteristic points are classified, then connection rules of the line segments among the connection characteristic points are established, local shape characteristics of the artificial objects are identified semantically, connection rules among the local shape characteristics are further deduced, and then the structure of the complicated artificial object iron tower is identified. The method is suitable for identifying the structure of the iron tower under the condition that the complete outline of the iron tower and the shielding between the complete outline and the structure of the iron tower cannot be detected.
Drawings
FIG. 1 is a schematic view of a building rooftop model on an image of an unmanned aerial vehicle according to the present invention;
FIG. 2 is a 3D model of an iron tower according to the present invention;
fig. 3 is a schematic diagram of a main iron tower node model according to the present invention;
fig. 4 is a flow chart of the method for identifying the transmission line iron tower of the invention;
FIG. 5 illustrates a line segment orientation grouping method described in the present invention;
FIG. 6 is a diagram illustrating a line segment common-endpoint model according to the present invention;
FIG. 7 is a schematic view of a U-shaped wire segment shape model of the present invention classification;
FIG. 8 is a schematic view of a classified Y-node model according to the present invention;
FIG. 9 is a schematic diagram of a classified common-endpoint model according to the present invention;
FIG. 10 is a schematic view of a star-shaped partial structure;
FIG. 11 is a schematic diagram of the spatial relationship between triangles;
FIG. 12 is a schematic diagram of the spatial relationship between quadrangles;
fig. 13 is a schematic view of tower face identification;
fig. 14 is a schematic space structure diagram of the tower head and the cross arm.
Detailed Description
The invention is further described below with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and the protection scope of the present invention is not limited thereby.
As shown in fig. 1, different characteristics of connection nodes in a surface structure presented by a building shot by an unmanned aerial vehicle from the air, the roof of the building mainly presents a 'Y' shape model in any direction;
as shown in fig. 2, is an unmanned aerial vehicle aerial photography iron tower 3D model. The whole iron tower mainly comprises a tower head, a tower body and tower legs, wherein the part of the tower head above the section of the tower which is sharply changed (broken line appears) from the tower legs to the tower is the tower head, and if the section is not sharply changed, the part above the lower chord of the lower cross arm is the tower head. Tower legs: the first section of tower above the foundation (in the electric power terminology, the foundation is a cement pier which is formed by burying a steel tower support underground and is protruded on the ground by pouring cement) is called a tower leg; a tower body: the part between the tower leg and the tower head is called the tower body. Fig. 2(a) depicts a 3D structural model of a pylon, where A, B, C, D is referred to as the pylon's pylon body brace, and A, D is referred to as the boundary brace, and B, C is referred to as the occluded internal brace. E. F are all called cross arms of the tower head, E is an upper cross arm, and F is a lower cross arm. The fretwork characteristic of iron tower will cause and to produce a lot of false tie points between the billet of sheltering from on the image and thereby cause certain degree of difficulty for discernment, even if also be difficult to distinguish the concrete tower face that every billet is affiliated to of manual observation. The electric power iron tower that uses in the actual electric wire netting is of a great variety, can divide into according to the shape: the main difference of classification is the tower head, and the tower body often has similar steel bar cross-linking support structure. The oblique steel of the tower body presents a crossed X-shaped model, and meanwhile, the oblique steel of the tower body presents a K-shaped characteristic when being connected to a main supporting frame of the iron tower, and the tower legs present a W-shaped characteristic.
The overall identification of the iron tower is divided into the identification of a tower body and the identification of a tower head, the tower body of the iron tower is supported by four supports to form four tower surfaces of the tower body, the tower head is supported by a transverse tower, and the supports of the tower body and the supports of the tower head are communicated.
As shown in fig. 3, the diagonal steel of the tower body is described as converging to the bracket with the node feature, fig. 3(a) depicts that in the tower head area, the crossed diagonal segments converge to the cross frame; FIGS. 3(b), (e) illustrate the convergence of diagonal segments to the support to the main frame in the tower region; fig. 3(c) and (d) describe the intersection characteristics of line segments inside the iron tower.
As shown in fig. 4, a method for identifying an overhead transmission line iron tower structure based on an unmanned aerial vehicle image includes the steps of:
1) extracting line segments in different directions from the unmanned aerial vehicle inspection image and describing the attributes of the line segments;
2) detecting and classifying connection characteristics between adjacent multiple line segments;
3) clustering the line segments with different connection characteristics into triangular and quadrilateral profile characteristics contained in the iron tower;
4) and clustering and combining the local contour characteristics into a structure which conforms to the tower head, the tower body and the tower legs of the power transmission line iron tower.
Step 1, extracting line segments in different directions from an unmanned aerial vehicle inspection image and describing the attributes of the line segments, specifically:
the method comprises the steps of carrying out gray processing on an unmanned aerial vehicle inspection image, processing the inspection image by adopting 8 Prewitt operators in different directions, extracting edge information of the inspection image, generating a binary image by a maximum inter-class binary difference method, managing line segments in all different directions by using a Blob communication structure, and specifically adopting the 8 Prewitt operators in different directions as follows:
secondly, constructing line segments into vector line segments on the generated binary image, and expressing the line segments in different directions as follows: l ═ l1,l2,....,lnN is the total number of line segments, and any line segment liThe attributes are described as: (c)i,si,ei,θi,Li,oi) I is a line serial number, i is 1 to n, wherein:
is a line segment liThe coordinates of the center point are calculated,
is a line segment liThe coordinates of the starting point,
is a line segment liCoordinates of the end point, θiIs a line segment liAngle, line segment liLength Li，oiLine segment l is (h, v, o +, o-)iThe orientation of (1);
as shown in fig. 5, the line segments are classified into 4 groups according to their angles: -10 ° is taken as a horizontal line segment, denoted h; less than-75 ° and greater than-90 ° or greater than 75 ° and less than 90 ° are considered as vertical segments, denoted v; the angle of 10-75 degrees is regarded as an oblique upper line segment and is expressed as o +; the angle of-75 DEG to-10 DEG is regarded as a downward slope line segment and is represented as o-. Regarding horizontal, oblique upper and oblique lower line segments, taking a left end point as a starting point and taking a right end point as an end point; for a vertical line segment, the top point is taken as the starting point and the bottom point is taken as the ending point.
Step 2, detecting and classifying connection characteristics between adjacent multiple line segments, and the specific steps comprise:
as shown in FIG. 6, in a line segment set, a line segment l is readiGo through any other line segment ljJ is 1 to n, and first, the line segment l is calculatediStarting point and line segment ljWhether the collineation exists or not is judged by the following method:
if d(s)i,sj)||d(si,ej) T is ≦ T, wherein: d(s)i,sj) Is a line segment liStarting point and line segment ljThe distance of the starting point; d(s)i,ej) Is a line segment liStarting point and line segment ljDistance of the end point; t is a distance threshold value, the value range of T is more than or equal to 0 and less than or equal to 64, and | I represents the relation of OR; and if two line segments li、ljThe length of (a) satisfies:
0.3≤Trif the value is less than or equal to 1, the line segment liAnd line segment ljHaving co-connected endpoints, wherein: line segment liIs expressed as LiLine segment ljIs expressed as Lj；
Collecting line segment liThe common endpoint line segment for all start points is represented as: cs(i) (ii) a Collecting line segment liThe common endpoint line segment for all end points is represented as: ce(i) In that respect In the common endpoint line segment set, each line segment attribute is defined as (L)c,Lt,Lθ)，LcLine segment number, LtIndicating a common end point of the starting point or the end point, LθFor the angles of the common-endpoint line segments, the common-endpoint line segments are classified according to the following rules:
firstly, detecting two adjacent L-shaped line sections: line segment liCommon endpoint set C ofs(i)、Ce(i) If the following condition is satisfied, the line segment l is detectediAnd adjacent line segments are L-shaped (marked S)L) Common endpoint line segment:
1) if c iss(i)＝0&&ce(i) Not equal to 0 or cs(i)≠0&&ce(i) Line segment l is 0iOnly one end point has a common connecting line segment;
2) all and line segments liThe line segment with the same end point is connected with the starting point or the ending point;
3) line segment liThe angle difference L between the common endpoint line segment and all the common endpoint line segmentsθThe range of (A) is as follows: l is not less than 75 degreesθLess than or equal to 105 degrees; if line segment liThe angle difference L between the common endpoint line segment and all the common endpoint line segmentsθThe range of (A) is as follows: theta is more than or equal to 25 degreesdiffAnd judging that the angle between two adjacent line segments is less than or equal to 75 degrees. The triangular structure is regarded as the combination of two < shaped structures with common edges,the end points of the two non-common edges of the & lt-shaped structures meet the intersection condition, and the line segments meeting the condition are clustered into a triangular structure.
II, detecting U-shaped three-phase adjacent line sections:
line segment l, as shown in FIG. 7iCommon endpoint set C ofs(i)、Ce(i) If the following condition is satisfied, the line segment l is detectediAnd adjacent line segments are U-shaped (marked S)U) Common endpoint line segment:
1) if c iss(i)≠0&&ce(i) Not equal to 0, line segment liThe starting end point and the ending end point are provided with a common connecting line segment;
2) all and line segments liThe line segment with the same end point is connected with the starting point or the ending point;
3) line segment liDegree difference theta between all line segments at common end pointdiffThe range of (A) is as follows: theta is more than or equal to 75 degreesdiff≤105°；
Third, the detection of Y-shaped line segment, if it is connected with line segment liThe number of the common end point line segments is more than two, at least two included angles between every two adjacent line segments are more than or equal to 90 degrees in the clockwise direction, and the common end point line segments are judged to be a Y-shaped node model (marked S) as shown in figure 8Y)。
Fourth, detecting the line segment with the point K, if it is the line segment liThe number of the line segments at the common end point is more than two; at least two of the included angles between two adjacent segments are less than 90 degrees in the clockwise direction, and the K-type segment is determined (marked S) as shown in FIGS. 9(a), (b), (c) and (d)K) And (4) node models. Of particular interest to the node model of fig. 9(c) (d), all line segments share a common starting point or end point, and a large number of these common end point connection nodes exist on the tower body of the iron tower.
Fifthly, cross X-shaped line segment detection is carried out, and any two line segments l are assumedi、ljThe orientations are respectively:
and is
And is
And is
And is
Two by two line segments of these four cases, ojIs a line segment ljThe direction of (2), calculating the intersection r of the two line segments (r ═ r)x,ry)，rx,ryRespectively is the abscissa and the ordinate of the intersection point of the two line segments;
if the following conditions are met:
and
and is
And
two line segments li、ljThe length of (a) satisfies:
wherein: t is 0.3. ltoreq. T.ltoreq.1, the two line segments are classified as X-shaped cross shapes (symbol S) as in FIG. 9(e)X) Structural features which are abundant on iron towers.
Sixthly, detecting a star-shaped structure, wherein a star-shaped partial structure of a tower is shown in figure 10, the structure has symmetry,that is, the central points of all the line segments forming the structure intersect at P, the line segments forming the structure are re-divided by taking the point P as an end point, and any line segment always has another collinear line segment taking the point P as an end point. Line segment s in FIG. 10(a)jP and line segment Pej。
As shown in FIG. 10(a), if the line segment li,lj,lk.., satisfying:
in the formula, the condition (1) indicates that the midpoints of all the line segments intersect, | | ci,cjI represents a line segment li，ljEuclidean distance between center points, ci,cj,ckRespectively represent line segments li，lj，lkCoordinate of center point of (1), TcRepresents a distance threshold value, and takes a value of 16 or 32; condition (2) indicates that all line segment lengths satisfy a certain threshold ratio, Li,Lj,LkRespectively represent line segments li，lj，lkLength of (d); at this time, all the line segments li，lj，lk..S(S,Llines,Pcenter,Pends) Wherein S is a star type identifier indicating that the line segment set is a star type identifier, LlinesTo satisfy segment numbering of star type, PcenterIs the aggregate center of the points in all line segments, PendsThe coordinates of the start and stop points of all line segments, namely the coordinates of the four points of the circumscribed rectangle of all line segments. In fact, the symmetrical star structure may be deformed due to the change of the photographing angle.
firstly, clustering the relationship contour features between triangles, as shown in fig. 11, divides the relationship between two triangles into three main types of relationships: FIG. 11(a) shows two triangles forming a parallelogram; as shown in fig. 11(b), two triangles present collinear and coincident sides,no overlap exists between the two; as shown in fig. 11(c), two triangles have collinear and overlapping sides, and overlap each other. After recognizing the triangle Tr1,Tr2.., further identifying the relationship between the triangles includes the steps of:
1): with the current triangle Tr1For reference, searching a triangle structure T with collinear edges and common vertices with the current triangler2And classified as follows.
a) Triangle Tr1、Tr2There are common edges, both constituting a parallelogram or trapezoid as shown in FIG. 11 (a);
b) triangle Tr1、Tr2Collinear edges exist, but are not equal, and the areas are not coincident as shown in FIG. 11 (b);
c) triangle Tr1、Tr2There are common edges and the areas coincide as shown in fig. 11 (c).
2): repeating the step 1) until all the triangle structures are verified;
3): the triangular line segments which meet the three conditions in the step 1) are gathered together.
Secondly, clustering relation contour features among the quadrangles:
the spatial relationship between the quadrilaterals is illustrated as fig. 12, where fig. 12(a) two quadrilaterals have one set of opposing sides collinear and the other set of opposing sides also in parallel relationship; FIG. 12(b) shows that the sides of two quadrangles are parallel to each other, and the areas of the quadrangles are different; FIG. 12(c) shows two quadrilaterals with parallel sides and approximately equal areas; in FIG. 12(d) the two quadrilaterals share vertices and the edges sharing vertices have collinear features. The specific process for discussing the spatial relationship of the quadrangle is as follows: if two edge lines of the two U-shaped lines are the same or two triangles share the same line, the two U-shaped lines are combined into a quadrangle, and the quadrangle Q is identified1,Q2.., further identifying the relationship between quadrangles, the method comprises the following steps:
1) with the current quadrilateral Q1As a reference, the search satisfies with the current quadrangle:
a) as shown in FIG. 12(a), IVEdge shape Q1、Q2One set of opposite edges are collinear (GF and BC, EH and AD), the other set of opposite edges are parallel (GH and CD, EF and AB), and the areas are not overlapped;
b) as shown in FIG. 12(b), a quadrangle Q1、Q2No public edge or vertex exists, four edges are respectively correspondingly parallel, and the area satisfies
Namely, the ratio of the smaller area to the larger area is less than 0.8;
c) as shown in FIG. 12(c), a quadrangle Q1、Q2No public edge or vertex exists, four edges are respectively correspondingly parallel, and the area satisfies
d) As shown in FIG. 12(d), a quadrangle Q1、Q2There are common vertices, two of which are co-linear (AD and EF, CD and EH) and two parallel (HG and BC, GF and AB).
2) The classified output meets the four types of quadrilateral combinations, and all quadrilateral structures are traversed;
3) and clustering and merging quadrilateral line segments meeting the spatial relations of the classes a), b), c) and d).
And 4, clustering and combining the local contour characteristics into a structure which conforms to the tower head, the tower body and the tower legs of the power transmission line iron tower, and specifically comprising the following steps:
the method specifically comprises the following steps of:
fig. 13 is a schematic diagram illustrating the exploded and recognized surface of the tower body of the iron tower. The set of quadrilateral structures identified by step 3: q1,Q2.., set of triangle structures Tr1,Tr2.., identifying the tower body surface structure F of the iron tower1,F2.., the steps include:
1) searching a long vertical line section on the inspection image, judging whether the searched long vertical line section accords with a trapezoidal relation, wherein the angle of the leftmost long line section accords with positive 70-90 degrees, the angle of the rightmost long line section accords with negative (-90-70 degrees), and the angle is the angle of the slope of an image pixel coordinate fitting outlet line section. As shown in fig. 13, the long line segment conforming to the trapezoidal relationship is taken as an outermost support line segment A, D of the tower body of the iron tower;
2) according to the 3D hollow structure of the tower body, the tower body surface is overlapped and shielded, and the tower body support comprises a plurality of K-shaped nodes SK1The tower body surface comprises a plurality of typical internal star nodes SS2. As shown in fig. 13(a) and (b), searching from the leftmost frame a of the image to the right, and satisfying a set of quadrangles with no overlapping area and collinear upper and lower opposite sides, i.e. the result of class a) output in step 3 two, when the number of quadrangles satisfying the condition is greater than 2, as shown in fig. 13(a), specifically as shown in fig. 13(b), a quadrangle Q1,Q2,Q3,Q4Further searching for those segments contained in the quadrilateral to belong to the internal star node SS2Line segments simultaneously belonging to a K-shaped node SK1And fitting a straight line segment from the lower part of the image coordinate to the upper part based on the K-shaped node. If two straight line segments are fitted, the two straight line segments are set as the line segments of the support in the tower body, and based on the external polygon of the quadrangle as the surface of the iron tower, the figure 13(a) is called as an iron tower surface F1FIG. 13(b) is referred to as a tower surface F2Tower surface F1Bracket line segment is A, B, iron tower surface F2The stent line segment is A, C, and A is the overlapping leftmost collinear stent line segment.
3) Based on the above steps 1) and 2), searching the rightmost stent segment D in the image leftwards, and enabling the quadrangle connected with the common vertexes to be as Q shown in FIG. 13(c)1,Q2,Q3I.e. satisfying the class d) in step 3 two, further search for those segments contained in the quadrilateral which belong to the internal star node SS2Line segments simultaneously belonging to a K-shaped node SK2And fitting a straight line segment from the lower part of the image coordinate to the upper part based on the K-shaped node. And if two straight line segments are fitted, setting the straight line segments as the line segments of the internal support of the tower body, and if the fitted line segments of the internal support coincide with the line segments of the internal support fitted in the step 2), identifying the 3D tower body structure. FIG. 13(c) is referred to as a tower surface F3FIG. 13(d) is referred to as tower surface F4Tower surface F3Bracket line segment is C, D, iron tower surface F4The stent line segment is B, D, and D is the rightmost collinear stent line segment that overlaps.
4) Analyzing the triangular structure on the straight line of the tower bracket, and then analyzing the triangular T at the end point of the iron tower boundary A, Dr1， Tr2The edges are shared and overlapped to present the shape of S1, namely, the relation of c) in the first step 3 is satisfied; triangle T at end point of tower body inner support B, Cr3、Tr4The common edges are not overlapped, namely the relation of b) in the step 3I is satisfied, and the triangular areas represent the base points of the tower legs of the iron tower.
Secondly, the concrete steps of the identification of the tower head of the iron tower comprise:
for the identification of the tower head cross arm structure, two types of local structure constraint relations are shown in (a) and (b) in fig. 14, the following rules are operated for accurately identifying the cross arm structure: set of quadrilateral structures Q identified in step 31,Q2.., set of triangle structures Tr1,Tr2.., set of tower body and surface structures F1,F2.., identifying the tower head cross arm structure Cr1,Cr2.., the steps include:
1) the triangles satisfying the condition of a) in the first step 3 are taken as the basis of identification, and if the condition satisfies that the adjacent triangles form more than three triangles in the trapezoid set, as shown by T in FIG. 14(a)r1,Tr2,Tr3,Tr4And (3) as shown, taking the circumscribed polygonal area of the triangle set as the cross arm C of the tower head of the iron towerr1…；
2) Paralleling the sides satisfying the relationship of b) in step 3 two and forming quadrangles with unequal areas (usually, the area ratio between adjacent quadrangles is less than 0.8), such as Q in FIG. 14(b)1,Q2,Q3,Q4As shown, as an input for identifying the tower head and the cross arm, if the number of quadrangles meeting the condition is more than 3, the quadrangle circumscribed area meeting the condition is taken as a tower head cross arm area Cr1；
3) Searching line segment and tower surface F1,F2.. triangle with collinear boundaries, if the triangle is not in the tower body plane, it is used as the tower head cross arm area Cr2. Pass verification Cr1Region and Cr2Reliable detection of cross-arm for overlapping of regionsAnd the area is used as judgment for increasing reliability identification.
The method can effectively identify the iron tower structure of the overhead transmission line, and in order to distinguish different areas where buildings and the iron tower are located, starting from line characteristics of the iron tower and the building structure, common connection areas among line segments in different directions are used as connection characteristic points of artificial objects (the iron tower or the buildings), the connection characteristic points are classified, then connection rules of the line segments among the connection characteristic points are established, local shape characteristics of the artificial objects are identified semantically, connection rules among the local shape characteristics are further deduced, and then the structure of the complicated artificial object iron tower is identified. The method is suitable for identifying the structure of the iron tower under the condition that the complete outline of the iron tower and the shielding between the complete outline and the structure of the iron tower cannot be detected.
The above description is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, several modifications and variations can be made without departing from the technical principle of the present invention, and these modifications and variations should also be regarded as the protection scope of the present invention.
Claims (4)
1. A method for identifying an iron tower structure of an overhead transmission line based on an unmanned aerial vehicle image is characterized by comprising the following steps:
1) extracting line segments in different directions from the unmanned aerial vehicle inspection image and describing the attributes of the line segments;
2) detecting and classifying connection characteristics between adjacent multiple line segments;
3) clustering line segments with different connection characteristics into local contour characteristics of triangles and quadrangles contained in the iron tower;
4) clustering and combining the local contour characteristics into structures which accord with the tower head, the tower body and the tower legs of the power transmission line iron tower;
the step 3) clusters line segments with different connection characteristics into local contour characteristics of triangles and quadrangles contained in the iron tower, and the specific steps comprise:
firstly, clustering the local contour features of triangles in the iron tower, and identifying the triangle Tr1,Tr2.., three are further identifiedAngular relationships, the steps comprising:
1): with the current triangle Tr1For reference, searching a triangle structure T with collinear edges and common vertices with the current triangler2And classified as follows:
a) triangle Tr1、Tr2There are common edges, both forming a parallelogram or trapezoid;
b) triangle Tr1、Tr2Collinear edges exist, but are not equal, and the areas are not coincident;
c) triangle Tr1、Tr2Common edges exist and the areas coincide;
2): repeating the step 1) until all the triangle structures are verified;
3): gathering the triangular line segments meeting the three conditions in the step 1);
secondly, clustering the local contour features of quadrangles in the iron tower, if two side lines of two U-shaped parts are the same or two triangles share a line segment, combining the two side lines into a quadrangle, and identifying the quadrangle Q1,Q2.., further identifying the relationship between quadrangles, the method comprises the following steps:
1) with the current quadrilateral Q1As a reference, the search satisfies with the current quadrangle:
a) quadrilateral Q1、Q2One group of opposite sides are collinear, the other group of opposite sides are parallel, and the areas are not overlapped;
b) quadrilateral Q1、Q2No public edge or vertex exists, four edges are respectively correspondingly parallel, and the area satisfies
Namely, the ratio of the smaller area to the larger area is less than 0.8;
c) quadrilateral Q1、Q2No public edge or vertex exists, four edges are respectively correspondingly parallel, and the area satisfies
d) Quadrilateral shapeQ1、Q2A common vertex exists, two edges sharing the vertex are correspondingly collinear, and the other two edges are correspondingly parallel;
2) the classified output meets the four types of quadrilateral combinations, and all quadrilateral structures are traversed;
3) clustering and merging the quadrilateral line segments meeting the spatial relationship of a) -d);
and 4) clustering and combining the local contour characteristics again to form a structure which conforms to the tower head, the tower body and the tower legs of the power transmission line iron tower, and the concrete steps comprise:
the method specifically comprises the following steps of:
the set of quadrilateral structures identified in step 3): q1,Q2.., set of triangle structures: t isr1,Tr2.., identifying the tower body surface structure F of the iron tower1,F2.., the concrete steps include:
1) searching a long vertical line section on the inspection image, judging whether the searched long vertical line section accords with a trapezoidal relation, judging whether the angle of the leftmost long line section accords with positive 70-90 degrees, judging whether the angle of the rightmost long line section accords with negative 90-negative 70 degrees, wherein the angle is the angle of the slope of an image pixel coordinate fitting outlet line section, and taking the long line section which accords with the trapezoidal relation as an outermost support line segment A, D of the tower body of the iron tower;
2) according to the 3D hollow structure of the tower body, the tower body surface is overlapped and shielded, and the tower body support comprises a plurality of K-shaped nodes SK1The tower body surface comprises a plurality of typical internal star nodes SS2(ii) a Searching rightwards from the leftmost support A of the image, and further searching a group of quadrangles with no overlapping area and collinear upper and lower opposite sides, wherein when the number of the quadrangles meeting the condition is more than 2, the line segments contained in the quadrangles belong to an internal star node SS2Line segments simultaneously belonging to a K-shaped node SK1Fitting a straight line segment from the lower part of the image coordinate to the upper part based on the K-shaped node; if two straight line segments are fitted, setting the straight line segments as the line segments of the support inside the tower body, and taking the external polygon of the quadrangle as the surface of the iron tower, namely the surface F of the iron tower1Bracket line segment is A, B, iron tower surface F2The stent line segment is A, C, A is heavyThe leftmost collinear leg segment of the stack;
3) based on the steps 1) and 2), searching the rightmost stent line segment D in the image leftwards, searching the quadrangle group connected with the common vertex, and further searching the line segments contained in the quadrangle to belong to the internal star node SS2Line segments simultaneously belonging to a K-shaped node SK2Fitting a straight line segment from the lower part of the image coordinate to the upper part based on the K-shaped node; if two straight line segments are fitted, setting the straight line segments as tower body internal support line segments, and if the fitted internal support line segments are superposed with the internal support line segments fitted in the step 2), identifying a 3D tower body structure and a tower surface F3Bracket line segment is C, D, iron tower surface F4The stent line segment is B, D, D is the overlapping rightmost collinear stent line segment;
4) analyzing the triangular structure on the straight line of the tower support, the triangles at the end points of the iron tower boundary A, D are in common and overlapped, the triangles at the end points of the tower body internal support B, C are in common and not overlapped, and the triangular areas represent the base points of the tower feet of the iron tower;
secondly, the concrete steps of the identification of the tower head of the iron tower comprise:
in the identified quadrilateral structure set Q1,Q2.., set of triangle structures Tr1,Tr2.., set of tower body and surface structures F1,F2.., identifying the tower head cross arm structure Cr1,Cr2.., the steps include:
1) the method comprises the steps that a parallelogram or trapezoidal triangle formed by a common side is used as a basis for identifying a tower head cross arm, if more than three triangles are included in a trapezoidal set formed by adjacent triangles, a circumscribed polygonal area of the triangular set is used as an iron tower head cross arm Cr1；
2) Quadrangles which have no common edges or vertexes, are parallel to each other and have unequal areas are taken as the basis for identifying the tower head cross arm, if the number of the quadrangles meeting the conditions is more than 3, the quadrangle external connection area meeting the conditions is taken as a tower head cross arm area Cr1；
3) Searching line segment and tower surface F1,F2.., triangles with collinear boundaries, if the triangles are not within the tower body,using it as tower head cross arm area Cr2。
2. The method for identifying the iron tower structure of the overhead transmission line based on the unmanned aerial vehicle image as claimed in claim 1, wherein the method comprises the following steps: the step 1) of extracting line segments in different directions from the unmanned aerial vehicle inspection image and describing the attributes of the line segments comprises the following specific steps:
firstly, graying an unmanned aerial vehicle inspection image to generate a binary image;
secondly, constructing line segments into vector line segments on the generated binary image, and expressing the line segments in different directions as follows: l ═ l1,l2,....,lnN is the total number of line segments, and any line segment liThe attributes are described as: (c)i,si,ei,θi,Li,oi) I is a line serial number, i is 1 to n,
is a line segment liThe coordinates of the center point are calculated,
is a line segment liThe coordinates of the starting point,
is a line segment liCoordinates of the end point, θiIs a line segment liAngle, line segment liLength Li，oiLine segment l is (h, v, o +, o-)iThe orientation of (1); the line segment angle is-10 degrees to 10 degrees and is regarded as a horizontal line segment which is expressed as h; less than-75 ° and greater than-90 ° or greater than 75 ° and less than 90 ° are considered as vertical segments, denoted v; the angle of 10-75 degrees is regarded as an oblique upper line segment and is expressed as o +; the angle of-75 DEG to-10 DEG is regarded as a downward slope line segment and is represented as o-.
3. The method for identifying the iron tower structure of the overhead transmission line based on the unmanned aerial vehicle image as claimed in claim 2, wherein the method comprises the following steps: the unmanned aerial vehicle inspection image is grayed, a binary image is generated, and the method specifically comprises the following steps:
and processing the inspection image by using 8 Prewitt operators in different directions, extracting edge information of the inspection image, generating a binary image by a maximum inter-class binary difference method, and managing all line segments in different directions by using a Blob communication structure.
4. The method for identifying the iron tower structure of the overhead transmission line based on the unmanned aerial vehicle image as claimed in claim 2, wherein the method comprises the following steps: the step 2) of detecting and classifying the connection characteristics between the adjacent multiple line segments comprises the following specific steps:
in a line segment set, a line segment l is readiGo through any other line segment ljJ is 1 to n, and first, the line segment l is calculatediStarting point and line segment ljWhether the collineation exists or not is judged by the following method:
if d(s)i,sj)||d(si,ej) T is ≦ T, wherein: d(s)i,sj) Is a line segment liStarting point and line segment ljThe distance of the starting point; d(s)i,ej) Is a line segment liStarting point and line segment ljThe distance of the end point, T is a distance threshold value, the value range of T is more than or equal to 0 and less than or equal to 64, and | represents the relation of OR; and if two line segments li、ljThe length of (a) satisfies:
then line segment liAnd line segment ljHaving co-connected endpoints, wherein: line segment liIs expressed as LiLine segment ljIs expressed as Lj；
Collecting line segment liThe common endpoint line segment for all start points is represented as: cs(i) (ii) a Collecting line segment liThe common endpoint line segment for all end points is represented as: ce(i) The common-endpoint line segments are classified according to the following rules:
firstly, detecting two adjacent L-shaped line sections: line segment liCommon endpoint set C ofs(i)、Ce(i) If the following condition is satisfied, the detection is performedLine outgoing section liAnd an L-shaped common-end line segment is arranged between the adjacent line segments:
1) if c iss(i)＝0&&ce(i) Not equal to 0 or cs(i)≠0&&ce(i) Line segment l is 0iOnly one end point has a common connecting line segment;
2) all and line segments liThe line segment with the same end point is connected with the starting point or the ending point;
3) line segment liThe angle difference L between the common endpoint line segment and all the common endpoint line segmentsθThe range of (A) is as follows: l is not less than 75 degreesθLess than or equal to 105 degrees; if line segment liThe angle difference L between the common endpoint line segment and all the common endpoint line segmentsθThe range of (A) is as follows: theta is more than or equal to 25 degreesdiffJudging that the two adjacent line segments are in a shape of angle less than or equal to 75 degrees, regarding the triangular structure as a combination of two structures with common edges, enabling the end points of the two structures with non-common edges to meet intersection conditions, and clustering the line segments meeting the conditions into the triangular structure;
II, detecting U-shaped three-phase adjacent line sections:
line segment liCommon endpoint set C ofs(i)、Ce(i) If the following condition is satisfied, the line segment l is detectediAnd a U-shaped common-end line segment is arranged between the adjacent line segments:
1) if c iss(i)≠0&&ce(i) Not equal to 0, line segment liThe starting end point and the ending end point are provided with a common connecting line segment;
2) all and line segments liThe line segment with the same end point is connected with the starting point or the ending point;
3) line segment liDegree difference theta between all line segments at common end pointdiffThe range of (A) is as follows: theta is more than or equal to 75 degreesdiff≤105°；
Third, the detection of Y-shaped line segment, if it is connected with line segment liThe number of the common end point line segments is more than two, at least two included angles between every two adjacent line segments are more than or equal to 90 degrees in the clockwise direction, and then the Y-shaped node model is judged;
fourth, detecting the line segment with the point K, if it is the line segment liThe number of the line segments at the common end point is more than two, and the included angle between every two adjacent line segments is at least two according to the clockwise directionIf the number of the nodes is less than 90 degrees, judging the node as a K-type node model;
fifthly, cross X-shaped line segment detection is carried out, and any two line segments l are assumedi、ljThe orientations are respectively:
and is
And is
And is
And is
Two by two line segments of these four cases, ojIs a line segment ljThe direction of (2), calculating the intersection r of the two line segments (r ═ r)x,ry)，rx,ryRespectively is the abscissa and the ordinate of the intersection point of the two line segments;
if the following conditions are met:
and
and is
And
two line segments li、ljThe length of (a) satisfies:
wherein: t is more than or equal to 0.3 and less than or equal to 1, and then the two line segments are classified into X-shaped cross shapes;
sixthly, detecting a star-shaped structure, wherein the central points of all line segments forming the structure are intersected at a point P, the point P is taken as an endpoint to re-divide the line segments forming the structure, any line segment always has another collinear line segment taking the point P as the endpoint,
if line segment li,lj,lk.., satisfying:
in the formula, the condition (1) indicates that the midpoints of all the line segments intersect, | | ci,cjI represents a line segment li，ljEuclidean distance between center points, ci,cj,ckRespectively represent line segments li，lj，lkCoordinate of center point of (1), TcRepresents a distance threshold; condition (2) indicates that the minimum to maximum length ratio of all line segments is greater than the threshold ratio of 0.5, Li,Lj,LkRespectively represent line segments li，lj，lkWill satisfy all line segments l of the conditioni，lj，lk..S(S,Llines,Pcenter,Pends) Wherein S is a star type identifier indicating that the line segment set is a star type identifier, LlinesTo satisfy segment numbering of star type, PcenterIs the aggregate center of the points in all line segments, PendsThe start and stop points of all line segments.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810443089.7A CN108985143B (en) | 2018-05-10 | 2018-05-10 | Method for identifying iron tower structure of overhead transmission line based on unmanned aerial vehicle image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810443089.7A CN108985143B (en) | 2018-05-10 | 2018-05-10 | Method for identifying iron tower structure of overhead transmission line based on unmanned aerial vehicle image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108985143A CN108985143A (en) | 2018-12-11 |
| CN108985143B true CN108985143B (en) | 2022-03-29 |
Family
ID=64541972
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810443089.7A Active CN108985143B (en) | 2018-05-10 | 2018-05-10 | Method for identifying iron tower structure of overhead transmission line based on unmanned aerial vehicle image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108985143B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109711368B (en) * | 2018-12-29 | 2023-04-14 | 华东理工大学 | High-altitude cable dangerous area image identification method |
| CN110119161B (en) * | 2019-06-14 | 2022-03-01 | 河海大学常州校区 | Unmanned aerial vehicle power transmission line inspection visual obstacle avoidance method and system |
| CN110889827A (en) * | 2019-11-06 | 2020-03-17 | 国网山西省电力公司吕梁供电公司 | Transmission line tower online identification and inclination detection method based on vision |
| CN112034470B (en) * | 2020-09-02 | 2022-10-18 | 亿嘉和科技股份有限公司 | Cable identification and positioning method based on solid-state area array laser radar |
| CN112904896B (en) * | 2021-01-21 | 2022-11-04 | 中国南方电网有限责任公司超高压输电公司柳州局 | Unmanned aerial vehicle autonomous driving route multiplexing method |
| CN112883840B (en) * | 2021-02-02 | 2023-07-07 | 中国人民公安大学 | Power transmission line extraction method based on key point detection |
| CN114626131B (en) * | 2022-03-23 | 2023-12-19 | 新风光电力科技(北京)有限公司 | Power grid power infrastructure safety evaluation analysis method, system and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102819743A (en) * | 2012-08-14 | 2012-12-12 | 常州大学 | Detection method for quickly identifying straight-line segments in digital image |
| 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 | Method and device for detecting toppling of tower |
| 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 |
-
2018
- 2018-05-10 CN CN201810443089.7A patent/CN108985143B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102819743A (en) * | 2012-08-14 | 2012-12-12 | 常州大学 | Detection method for quickly identifying straight-line segments in digital image |
| 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 | Method and device for detecting toppling of tower |
| 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 |
Non-Patent Citations (7)
| Title |
|---|
| Extracting Buildings from Aerial Images using Hierarchical Aggregation in 2D and 3D;Andre Fisher 等;《COMPUTER VISION AND IMAGE UNDERSTANDING》;19981130;第72卷(第2期);185–203 * |
| Pylon Line Spatial Correlation Assisted Transmission Line Detection;JUN ZHANG 等;《IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS》;20141031;第50卷(第4期);2890-2905 * |
| 一种基于多尺度自适应形态学的塔吊提取方法;于博 等;《遥感技术与应用》;20130430;第28卷(第2期);240-244 * |
| 一种无人机图像的铁塔上鸟巢检测方法;徐晶 等;《计算机工程与应用》;20160104;1-7 * |
| 基于无人机图像的电力杆塔倾斜检测;王榆夫 等;《计算机仿真》;20170731;第34卷(第7期);426-431 * |
| 架空输电线路中设备轮廓提取方法的研究;李辉 等;《电子测量技术》;20150228;第41卷(第4期);87-92 * |
| 采用单幅草图的正交多面体模型生成方法;宋沫飞 等;《计算机辅助设计与图形学学报》;20120131;第24卷(第1期);50-59 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108985143A (en) | 2018-12-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108985143B (en) | Method for identifying iron tower structure of overhead transmission line based on unmanned aerial vehicle image | |
| CN110570428B (en) | Method and system for dividing building roof sheet from large-scale image dense matching point cloud | |
| CN112801022B (en) | Method for rapidly detecting and updating road boundary of unmanned mining card operation area | |
| Yang et al. | Hierarchical extraction of urban objects from mobile laser scanning data | |
| WO2020052530A1 (en) | Image processing method and device and related apparatus | |
| Awrangjeb et al. | Automatic building extraction from LiDAR data covering complex urban scenes | |
| CN106780524A (en) | A kind of three-dimensional point cloud road boundary extraction method | |
| Liu et al. | Building extraction from high resolution imagery based on multi-scale object oriented classification and probabilistic Hough transform | |
| Wei | Building boundary extraction based on lidar point clouds data | |
| CN107680095A (en) | The electric line foreign matter detection of unmanned plane image based on template matches and optical flow method | |
| Fukano et al. | Detection and classification of pole-like objects from mobile mapping data | |
| Awrangjeb et al. | Rule-based segmentation of LIDAR point cloud for automatic extraction of building roof planes | |
| CN107462217A (en) | Unmanned aerial vehicle binocular vision barrier sensing method for power inspection task | |
| CN111652241B (en) | Building contour extraction method integrating image features and densely matched point cloud features | |
| CN113704381B (en) | Road network data processing method and device, computer equipment and storage medium | |
| Maltezos et al. | Automatic extraction of building roof planes from airborne lidar data applying an extended 3d randomized hough transform | |
| Le Saux et al. | Railway detection: From filtering to segmentation networks | |
| Tu et al. | Extraction of street pole-like objects based on plane filtering from mobile LiDAR data | |
| Hu et al. | Geometric feature enhanced line segment extraction from large-scale point clouds with hierarchical topological optimization | |
| Qiu et al. | A fast and robust algorithm for road edges extraction from lidar data | |
| CN107918775A (en) | The zebra line detecting method and system that a kind of auxiliary vehicle safety drives | |
| Cong et al. | Research on a point cloud registration method of mobile laser scanning and terrestrial laser scanning | |
| Wang et al. | Traffic sign three‐dimensional reconstruction based on point clouds and panoramic images | |
| Ogawa et al. | Joining street-view images and building footprint gis data | |
| CN114779794B (en) | Street obstacle identification method based on unmanned patrol vehicle system in typhoon scene |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |