阅读说明：本技术 基于激光点云的杆塔特征点自动识别及倾斜参数自动测量方法 (Tower characteristic point automatic identification and inclination parameter automatic measurement method based on laser point cloud ) 是由 曹晖 田泽英 王涛 薛霜思 袁廷翼 楼润枫 解明辉 于 2020-06-29 设计创作，主要内容包括：本发明公开了一种基于激光点云的杆塔特征点自动识别及倾斜参数自动测量方法,首先将获取的原始点云数据进行预处理,之后通过采用归一化数字表面模型、最大类间方差法、连通域分割、点云密度分割从处理后的点云数据中提取出杆塔,最后通过水平截面法得出杆塔的倾斜角度,从而实现了基于激光点云的杆塔特征点自动识别及倾斜参数自动测量。本发明能够对杆塔进行自动识别和倾斜度测量,大大地节省了人力物力,对于电力系统的稳定运行具有重要的研究意义。(The invention discloses a method for automatically identifying pole tower characteristic points and automatically measuring inclination parameters based on laser point cloud. The method can automatically identify the tower and measure the inclination, greatly saves manpower and material resources, and has important research significance for stable operation of the power system.)

1. A tower feature point automatic identification and inclination parameter automatic measurement method based on laser point cloud is characterized in that: the method comprises the following steps:

step 1: using an unmanned aerial vehicle to carry a laser radar to obtain point cloud data of the transmission tower;

step 2: preprocessing point cloud data:

carrying out abnormal value elimination preprocessing operation on the original point cloud data, and then separating out ground points and non-ground points through filtering and sorting processing;

and step 3: the method comprises the following steps of automatically identifying a pole tower target and automatically extracting key point positions of a pole tower structure:

according to the three-dimensional projection characteristics of the typical tower shape and the geometric structure of the tower target characteristic points of the laser point cloud, the automatic identification of the tower target and the automatic extraction of the key point positions of the tower structure are realized; the specific process is as follows:

(1) noise rejection on ground using normalized digital surface model

In a pair of complete tower point cloud images, the point cloud data comprises: pole tower point cloud, power line point cloud, rugged ground point cloud, tall building point cloud, tree point cloud, ground vegetation point cloud and noise point cloud; the ground noise can influence the integral point cloud processing, and the influence of the terrain can be eliminated and the ground noise can be filtered by adopting the normalized digital surface model;

(2) determining an optimal threshold value

Automatically determining an optimal threshold value by a maximum inter-class variance method in the process of filtering ground noise points; if the elevation value of the point cloud is lower than the optimal threshold value, the point cloud is removed, if the elevation value of the point cloud is larger than the optimal threshold value, the point cloud is left, at the moment, tall building point cloud and tree point cloud in the point cloud image are completely removed, and power line and tower candidate point cloud are left;

(3) tower power line extraction method based on connected domain analysis method

Only the original point cloud data of the towers and the power lines are left after the normalized digital surface model and the maximum inter-class variance method are adopted, and a connected domain analysis method is adopted in the area with more vegetation, so that the vegetation point cloud on the ground can be removed, and the towers and the power lines in the complex area with more vegetation can be extracted;

(4) extracting tower point cloud according to point cloud density

Because the point cloud density of the power line is small, and the point cloud density of the tower is large, the point cloud data of the area is counted, the point cloud with large density is the point cloud of the tower, the point cloud with small density is the point cloud of the power line, and the accurate point cloud of the tower can be extracted according to the difference of the point cloud densities to determine the position of the tower;

and 4, step 4: automatically calculating the inclination angle of the tower:

in an electric power system, a tower can incline due to climate, environment, natural disasters or human factors, and the operation of a power transmission line can be directly influenced under the serious condition; measuring the inclination angle of the tower by adopting a horizontal section method;

the specific measurement method is as follows:

(1) high voltage tower stratification

Horizontally slicing the three-dimensional tower point cloud, enabling the sliced point cloud to be parallel to the horizontal ground, projecting the sliced point cloud on the horizontal ground, and extracting the sliced boundary point cloud;

(2) boundary segmentation

Detecting boundary point clouds by adopting Hough transformation, converting lines of an image space into a parameter space, detecting extreme points of the parameter space, and extracting boundary lines;

(3) fitting of boundary straight line

The boundary point cloud extracted through the Hongh transformation has an abnormal value, and a straight line is fitted by adopting an improved least square method to obtain an accurate boundary; the calculation process is as follows:

let the equation of a straight line be

y＝ax+b

Wherein a and b are linear parameters;

the standard deviation of the distances of the points to the fitted line is as follows: :

wherein d is_iRepresents the distance of the ith point to the fitted line,represents the average distance of all points to the fitted straight line, n is the total number of all points;

average distance of all points to fitted straight lineWhen d is_iWhen the value is more than 2 sigma, the point is an abnormal point, and the point is removed; otherwise, the point is reserved; then, calculating the parameters a and b again; performing cyclic operation on the steps when d_iStopping operation when the standard deviation is less than 2 times;

(4) boundary angular point and central point calculation

Let the intersection point of two straight lines be

Then the projection coordinates of the corner points of the patch on the XOY plane are

The intersection point calculation is carried out on all the four straight lines to obtain four intersection points of the four edges of the plane, namely four boundary corner points which are respectively (x)₁,y₁)，(x₂,y₂)，(x₃,y₃)，(x₄,y₄) Wherein (x)₁,y₁) Is a plane(x) is the upper left corner point coordinate of (c)₂,y₂) Upper right corner point coordinate of bit plane, (x)₃,y₃) Is the coordinate of the lower left corner point of the plane, (x)₄,y₄) Coordinates of a lower right corner point of the bit plane;

the center point of the plane is

Calculating to finally obtain the central point of the plane through the above formula;

(5) gradient calculation method

And taking out a horizontal section of the tower point cloud, calculating the height difference between the center point of the horizontal section and the center point of the horizontal plane of the bottom layer and the height difference between the horizontal section and the horizontal plane of the bottom layer, wherein the result of the ratio is the inclination of the tower.

Technical Field

The invention relates to the technical field of computer vision, in particular to a method for automatically identifying pole tower characteristic points and automatically measuring inclination parameters based on laser point cloud.

Background

Our country mainly examines the transmission line manually, for example, it depends on the ground transportation means or hand-held instrument to detect, also detects the deformation of the pole tower through satellite remote sensing, at present, many exploratory researches are being done in this respect. The existing tower inclination detection means still has a plurality of defects and short plates in practical application: the problems of insufficient reliability of communication power supply, difficulty in installation and adjustment and the like exist when the tower inclination online monitoring device is installed, extra operation and maintenance burden is brought to line inspection personnel, the tower inclination online monitoring device is only suitable for fixed-point monitoring of a small number of towers, and the benefit cost ratio is not high for regional tower inclination monitoring; the manual measurement is carried out tower by using surveying and mapping means such as theodolite and total station, the workload is large, the efficiency is low, and the measurement difficulty of special towers (such as towers in mountainous areas) is great, so that the time consumption, the cost and the difficulty are great for sections needing to be measured generally; the remote sensing monitoring based on the SAR satellite has the advantages of wide monitoring range, large area and the like, but has higher requirements on the acquisition period and the image resolution of a remote sensing image, and has the disadvantages of huge data acquisition cost, complex data processing and difficult comprehensive popularization. Therefore, a practical technology capable of monitoring and early warning the inclination state of the tower in a geological unstable area in a large range is urgently needed.

In the electric power system, the use of the unmanned aerial vehicle brings a new prospect for the development of the industry: the unmanned aerial vehicle is used for carrying laser radar, visible light, multispectral, infrared and other equipment, rapid inspection of the power transmission line is achieved, and the unmanned aerial vehicle is already applied and popularized in the power inspection department. Compared with the mode that the unmanned aerial vehicle carries visible light and infrared load to patrol and examine, the unmanned aerial vehicle electric power line patrol system based on the laser radar has the absolute advantage of high horizontal and vertical position precision, can effectively improve the accuracy and reliability of patrolling and examining, and therefore becomes one of the main technical approaches for solving the monitoring problem of the inclined state of the tower.

Disclosure of Invention

In order to overcome the defects of positioning identification and tower deformation monitoring equipment in the prior art, the invention provides a tower characteristic point automatic identification and inclination parameter automatic measurement method based on laser point cloud.

In order to achieve the purpose, the invention adopts the following technical scheme:

the method comprises the following specific steps:

step 1: using an unmanned aerial vehicle to carry a laser radar to obtain point cloud data of the transmission tower;

step 2: preprocessing point cloud data:

carrying out abnormal value elimination preprocessing operation on the original point cloud data, and then separating out ground points and non-ground points through filtering and sorting processing;

and step 3: the method comprises the following steps of automatically identifying a pole tower target and automatically extracting key point positions of a pole tower structure:

according to the three-dimensional projection characteristics of the typical tower shape and the geometric structure of the tower target characteristic points of the laser point cloud, the automatic identification of the tower target and the automatic extraction of the key point positions of the tower structure are realized; the specific process is as follows:

(1) noise rejection on ground using normalized digital surface model

In a pair of complete tower point cloud images, the point cloud data comprises: pole tower point cloud, power line point cloud, rugged ground point cloud, tall building point cloud, tree point cloud, ground vegetation point cloud and noise point cloud; the ground noise can influence the integral point cloud processing, and the influence of the terrain can be eliminated and the ground noise can be filtered by adopting the normalized digital surface model;

(2) determining an optimal threshold value

Automatically determining an optimal threshold value by a maximum inter-class variance method in the process of filtering ground noise points; if the elevation value of the point cloud is lower than the optimal threshold value, the point cloud is removed, if the elevation value of the point cloud is larger than the optimal threshold value, the point cloud is left, at the moment, tall building point cloud and tree point cloud in the point cloud image are completely removed, and power line and tower candidate point cloud are left;

(3) tower power line extraction method based on connected domain analysis method

Only the original point cloud data of the towers and the power lines are left after the normalized digital surface model and the maximum inter-class variance method are adopted, and a connected domain analysis method is adopted in the area with more vegetation, so that the vegetation point cloud on the ground can be removed, and the towers and the power lines in the complex area with more vegetation can be extracted;

(4) extracting tower point cloud according to point cloud density

Because the point cloud density of the power line is small, and the point cloud density of the tower is large, the point cloud data of the area is counted, the point cloud with large density is the point cloud of the tower, the point cloud with small density is the point cloud of the power line, and the accurate point cloud of the tower can be extracted according to the difference of the point cloud densities to determine the position of the tower;

and 4, step 4: automatically calculating the inclination angle of the tower:

in an electric power system, a tower can incline due to climate, environment, natural disasters or human factors, and the operation of a power transmission line can be directly influenced under the serious condition; measuring the inclination angle of the tower by adopting a horizontal section method;

the specific measurement method is as follows:

(1) high voltage tower stratification

Horizontally slicing the three-dimensional tower point cloud, enabling the sliced point cloud to be parallel to the horizontal ground, projecting the sliced point cloud on the horizontal ground, and extracting the sliced boundary point cloud;

(2) boundary segmentation

Detecting boundary point clouds by adopting Hough transformation, converting lines of an image space into a parameter space, detecting extreme points of the parameter space, and extracting boundary lines; (Hough transform is a well-known element for detecting straight lines)

(3) Fitting of boundary straight line

The boundary point cloud extracted through the Hongh transformation has an abnormal value, and a straight line is fitted by adopting an improved least square method to obtain an accurate boundary; the calculation process is as follows:

let the equation of a straight line be

y＝ax+b

Wherein a and b are linear parameters;

the standard deviation of the distances of the points to the fitted line is as follows: :

wherein d is_iRepresents the distance of the ith point to the fitted line,represents the average distance of all points to the fitted straight line, n is the total number of all points;

average distance of all points to fitted straight line

When d is_iWhen the value is more than 2 sigma, the point is an abnormal point, and the point is removed; otherwise, the point is reserved; then, calculating the parameters a and b again; performing cyclic operation on the steps when d_iStopping operation when the standard deviation is less than 2 times;

(4) boundary angular point and central point calculation

Let the intersection point of two straight lines be

Then the projection coordinates of the corner points of the patch on the XOY plane are

The intersection point calculation is carried out on all the four straight lines to obtain four intersection points of the four edges of the plane, namely four boundary corner points which are respectively (x)₁,y₁)，(x₂,y₂)，(x₃,y₃)，(x₄,y₄) Wherein (x)₁,y₁) Is the upper left corner point coordinate of the plane, (x)₂,y₂) Upper right corner point coordinate of bit plane, (x)₃,y₃) Is the coordinate of the lower left corner point of the plane, (x)₄,y₄) Coordinates of a lower right corner point of the bit plane;

the center point of the plane is

Calculating to finally obtain the central point of the plane through the above formula;

(5) gradient calculation method

And taking out a horizontal section of the tower point cloud, calculating the height difference between the center point of the horizontal section and the center point of the horizontal plane of the bottom layer and the height difference between the horizontal section and the horizontal plane of the bottom layer, wherein the result of the ratio is the inclination of the tower.

According to the method, the unmanned aerial vehicle carries the laser radar to obtain the point cloud data of the transmission tower, the original point cloud data is preprocessed, the target tower point cloud is identified, and the tower inclination angle is calculated through the horizontal section method, so that automatic identification of the tower characteristic points and automatic measurement of the inclination parameters based on the laser point cloud are realized. The method effectively saves the time for manual judgment, can automatically measure the inclination angle of the tower so as to judge the state of the tower, and has important research significance for the stable operation of the power transmission line.

Drawings

Fig. 1 is an original tower point cloud image shot by an unmanned aerial vehicle.

Fig. 2 is a tower point cloud image from which ground noise is removed.

Fig. 3 is an image in which only the tower point cloud remains after other interference is removed.

Fig. 4 is a schematic diagram of tower inclination measurement based on a three-dimensional laser radar.

Detailed Description

The present invention will be described in more detail below with reference to the accompanying drawings and specific embodiments.

The invention relates to a tower characteristic point automatic identification and inclination parameter automatic measurement method based on laser point cloud, which comprises the following steps:

step 1: and shooting the tower by using an unmanned aerial vehicle in a range with the horizontal distance of 20-40m and the vertical distance of 20-40m from the outer edge of the tower to obtain point cloud data of the tower.

Step 2: after the original point cloud data is obtained, the point cloud data is preprocessed to generate a point cloud model with a three-dimensional space, as shown in fig. 1.

And step 3: according to the three-dimensional projection characteristics of the typical tower shape and the geometric structure of the tower target characteristic points of the laser point cloud, the automatic identification of the tower target and the automatic extraction of the key point positions of the tower structure are realized; in a pair of complete tower point cloud images, the point cloud data comprises: pole tower point cloud, power line point cloud, rugged ground point cloud, tall building point cloud, tree point cloud, ground vegetation point cloud and noise point cloud; the concrete process of extracting the tower point cloud is as follows: (1) noise on the ground can affect the overall point cloud processing, so that the influence of the ground noise can be eliminated by adopting a normalized digital surface model for the tower point cloud data, the ground noise is filtered, and a foundation is laid for subsequent processing; (2) automatically determining an optimal threshold value by a maximum inter-class variance method; if the elevation value of the point cloud is lower than the optimal threshold value, the point cloud point is removed, and if the elevation value of the point cloud is larger than the optimal threshold value, the point cloud points are left as candidate points of the power line and the tower; (3) only the original point cloud data of the towers and the power lines are left after the normalized digital surface model and the maximum inter-class variance method are adopted, a connected domain analysis method is adopted in the area with more vegetation, the vegetation point cloud on the ground can be removed, and the towers and the power lines in the complex area with more vegetation are extracted, as shown in the attached figure 2; (4) since the point cloud density of the power line is small and the point cloud density of the tower is large, the point cloud of the tower is extracted according to the difference of the point cloud densities, and accurate point cloud of the tower can be extracted by counting the point cloud data of the area to determine the position of the tower, as shown in fig. 3.

And 4, step 4: the method adopts a horizontal section method to measure the inclination angle of the tower, and the processing flow is as follows:

(1) high voltage tower stratification

Carrying out horizontal slicing operation on the three-dimensional tower point cloud, enabling the sliced point cloud to be parallel to the horizontal ground and be projected on the horizontal ground, and then extracting the sliced boundary point cloud;

(2) boundary segmentation

Detecting boundary point clouds by adopting Hough transformation, and detecting extreme points of a parameter space by converting lines of an image space into the parameter space so as to extract a boundary line;

(3) fitting of boundary straight line

Fitting a straight line by adopting an improved least square method to obtain an accurate boundary; the calculation process is as follows:

let the equation of a straight line be

y＝ax+b

Wherein a and b are linear parameters;

the standard deviation of the distances of the points to the fitted line is as follows:

wherein d is_iRepresents the distance of the ith point to the fitted line,

represents the average distance of all points to the fitted straight line, n is the total number of all points;

average distance of all points to fitted straight lineWhen d is_iWhen the value is more than 2 sigma, the point is an abnormal point, and the point is removed; otherwise, the point is reserved; then, calculating the parameters a and b again; performing cyclic operation on the steps when d_iStopping operation when the standard deviation is less than 2 times;

(4) boundary angular point and central point calculation

Let the intersection point of two straight lines be

Then the projection coordinates of the corner points of the patch on the XOY plane are

The intersection point calculation is carried out on all the four straight lines to obtain four intersection points of the four edges of the plane, namely four boundary corner points which are respectively (x)₁,y₁)，(x₂,y₂)，(x₃,y₃)，(x₄,y₄) Wherein (x)₁,y₁) Is the upper left corner point coordinate of the plane, (x)₂,y₂) Upper right corner point coordinate of bit plane, (x)₃,y₃) Is the coordinate of the lower left corner point of the plane, (x)₄,y₄) Coordinates of a lower right corner point of the bit plane;

the center point of the plane is

Calculating to finally obtain the central point of the plane through the above formula;

(5) gradient calculation method

Taking out a horizontal section of the point cloud of the tower, and calculating the ratio of the height difference between the center point of the horizontal section and the center point of the horizontal plane of the bottom layer to the height difference between the horizontal section and the horizontal plane of the bottom layer, wherein the result of the ratio is the inclination of the tower; as shown in figure 4, the height of the tower is H, the center point of the bottom of the tower is O, and the projection of the top of the tower is O'₁S is O and O'₁And finally calculating the inclination angle of the tower according to the formula q which is S/H.

According to the specification of construction and acceptance of 110-750 KV overhead transmission lines, the maximum allowable inclination value of a normal tower is shown in the following table 1;

TABLE 1 Tilt Classification

Categories	Reinforced concrete pole	Iron tower
			Tower gradient	150％	05% (suitable for tower with height of 50m and above)
		1.0% (suitable for tower with height below 50 m)
			Cross arm skew degree	1.00％	1％

Based on the above measurement method, the final measurement results were obtained in combination with the specifications of table 1, as shown in table 2 below.

TABLE 2