CN115655098A - Method for measuring and calculating earth and stone excavation and filling in power grid engineering by high-density laser point cloud technology - Google Patents

Abstract

The invention discloses a method for measuring and calculating excavated and filled earthwork of a power grid project by using a high-density laser point cloud technology, and belongs to the technical field of actual calculation of earthwork of soil and water conservation monitoring work. The unmanned aerial vehicle carries on the laser acquisition equipment, gathers high density laser point cloud data, according to the region of taking a photograph by plane, applies for and handles the flight wholesale. The method comprises the steps of selecting a suitable field along the line as a take-off and landing base, signing a guarantee protocol with an air traffic control unit, coordinating an airspace, performing batch flight according to the related air traffic control unit, designing, self-checking and checking, ensuring the accuracy of data in the design process of an aerial zone, and being better applied to engineering application.

Description

Method for measuring and calculating earth and stone excavation and filling in power grid engineering by high-density laser point cloud technology

Technical Field

The invention relates to the technical field of actual calculation of earth and stone space in water and soil conservation monitoring work, in particular to a method for measuring and calculating earth and stone space excavated and filled in power grid engineering by using a high-density laser point cloud technology.

Background

The volume of earth and stone is a very important numerical value in the water and soil conservation work of a construction project and is also very important in the main engineering process. The accurate measurement of the earth and stone volume has important significance for water conservation acceptance, earth and stone disposal and transportation of construction projects. However, the earth and stone volume of the current construction project cannot be accurately measured and calculated, and can only be estimated through the number of earth transportation vehicles, and the result is often greatly different from the actual result.

The high-density laser point cloud (unmanned aerial vehicle Lidar) technology can extract elevation information very accurately and is usually used for topographic survey, power transmission and transformation route selection, channel data extraction and the like. The method is less applied to the measurement and calculation of the engineering construction water and soil conservation soil and stone.

Therefore, a high-density laser point cloud technology is provided for measuring and calculating the excavation and filling stone space of the power grid engineering.

Disclosure of Invention

Technical problem to be solved

The invention aims to provide a method for measuring and calculating the excavation and filling stone volume of a power grid project by using a high-density laser point cloud technology, so as to solve the problems in the background technology.

Technical scheme

The method for measuring and calculating the excavation and filling stone volume of the power grid engineering by using the high-density laser point cloud technology comprises the following steps:

s1: carrying a laser radar by an unmanned aerial vehicle, and collecting high-density laser point cloud data before and after a project excavation and filling area;

s2: automatically classifying the point cloud, and then manually rechecking and performing data internal processing;

s3: utilizing a GIS to define a calculation range;

s4: and inputting the data into a data system, calculating the volume change by using the data before and after excavation, and calculating the earth and stone volume.

As a preferable scheme of the artificial intelligence-based material level sensing device, the step S1 further comprises the following steps: step S1 further comprises the steps of:

s101: unmanned aerial vehicle Lidar data processing carries out noise point through the data to airborne laser radar obtains and gets rid of, coordinate conversion back, carries out point cloud automatic classification, and the manual work is examined, is revised to the achievement after classifying, at last according to the classification achievement that requires output corresponding.

Based on the technical characteristics: in the design process of the flight band, three steps of design, self-checking and checking are needed to ensure the accuracy of data, and the flight band is better applied to engineering application.

As a preferable solution of the artificial intelligence based level sensing apparatus according to the present invention, the step S101 further includes the steps of:

s1011: point cloud data preprocessing;

s1012: removing point cloud noise;

s1013: classifying point cloud data;

s1014: and (4) point cloud data is mapped, a ground point model is processed, and elevation information is obtained.

Based on the technical characteristics: the laser point cloud data are preprocessed to obtain the space coordinates of the point cloud data and provide data for post-processing DEM. The data required for processing includes: original laser point cloud data, POS resolving data and the like. The processing content comprises laser point cloud, laser point cloud coordinate conversion and the like.

As a preferable aspect of the artificial intelligence-based level sensing apparatus of the present invention, wherein S1012: removing point cloud noise;

based on the technical characteristics: in the process of collecting three-dimensional point cloud data, the laser radar is influenced by various factors, so that certain noise can be generated when the data are obtained. In actual work, besides self-measurement errors, the device can be influenced by external environments, such as the shielding of a measured target, obstacles, the surface material of the measured target and other influence factors; in addition, some local large-scale noise cannot be filtered by the same method because the local large-scale noise is far away from the target point cloud.

The noise is a point which is not associated with the target information description and is useless for the reconstruction of the subsequent whole three-dimensional scene; the working process of the laser point cloud data processing comprises the following steps: eliminating an eccentric angle error by using a POS resolving result and calibration field point cloud, eliminating a torsion error by using a calibration field linear measuring point and a distance error by using a height point, and finally performing coordinate conversion on the calibrated point cloud data to generate point cloud data meeting the requirements;

in an actual point cloud data processing algorithm, it is very difficult to distinguish noise points from target points with characteristic information, and some characteristic information is inevitably lost due to many external factors in the denoising process. A good point cloud filtering algorithm not only has high real-time requirement, but also well retains the characteristic information of the model when denoising. The denoising algorithm with better effect can be provided only by applying the characteristics of the noise points of the point cloud data thoroughly.

The point cloud data is in an unstructured data format, the point cloud data obtained by scanning the laser radar is influenced by the distance between an object and the radar, the distribution has nonuniformity, the point cloud data of the object close to the radar is densely distributed, and the point cloud data of the object far away from the radar is sparsely distributed. In addition, the point cloud data has characteristics of disorder and asymmetry, which causes the point cloud data to lack a clear and uniform data structure during data representation, and increases the difficulty of processing such as segmentation and identification of subsequent point clouds. The neural network is used as an end-to-end network structure, data which is usually processed is conventional input data, such as sequences, images, videos, 3D data and the like, and disordered data such as a point set cannot be directly processed.

As a preferable aspect of the artificial intelligence-based level sensing apparatus of the present invention, wherein S1033: classifying point cloud data;

based on the technical characteristics: in terras solid software, the point cloud data classification map layer is as follows: 1) Defaults; 2) Group; 3) vegetation; 4) Building; 5) Lowpoint; 6) Modelkypoids. Wherein, the Default layer is a point cloud data layer temporarily stored in the operation process; ground points are mainly points for storing and reflecting the real landform of the Ground (deposits such as manually built embankments, earth dikes, step roads, naturally formed large-scale earth pits, earth piles and the like), and parts of manually built hydraulic structures such as earth ridges, water retaining dams, dry embankments, water gates and the like connected with the Ground are regarded as Ground points; vegetation (Vegetation points) are points where surface Vegetation is mainly stored, grasslands, shrubs, bamboo forests, nurseries, young forests, gardens, woodlands and the like are regarded as Vegetation points; building points are mainly points for storing earth surface buildings, and houses, greenhouses and the like are regarded as Building points.

And determining an optimal filter function according to the change of the terrain gradient, wherein for a given height difference value, the probability that the laser foot point with a large height value belongs to the ground point is smaller along with the reduction of the distance between the two points.

The gradient-based filtering algorithm has the characteristics of simplicity in calculation, strong adaptability and the like, but the terrain gradient needs to be known in advance and the size of a window to be opened needs to be determined, a selected point needs to be compared with all other points to determine whether the point is a ground point or not, gradient calculation needs to be carried out on each point in the whole data set, therefore, the calculated amount is increased, the speed is reduced, original point cloud data are partitioned according to the terrain statistical characteristics, then each partition is processed according to the gradient-based filtering algorithm to obtain each data ground point set, and finally, all blocks are spliced according to the characteristic points of an overlapped area to obtain the complete ground point set. Thus, different blocks can obtain different filtering thresholds, so that the singularity of the thresholds is avoided, and the classification error is reduced.

TIN is an important model representing digital elevation, often used to store the proximity relationships between spatially discrete points. For irregularly distributed elevation points, an unordered point set described as a plane can be formally formed, and a common method for converting the point set into the TIN is to construct a Delaunay triangulation network of the point set, where adjacent points in the Delaunay triangulation network are also adjacent in the TIN model.

The filtering algorithm is to filter the ground object points by applying parameters such as a height difference critical value condition, the number of the adjacent points meeting the condition and the like by utilizing the elevation mutation relation of the ground object adjacent point clouds in the TIN model;

as a preferable aspect of the artificial intelligence based level sensing apparatus of the present invention, wherein S1014: and (4) mapping the point cloud data, processing a ground point model and acquiring elevation information.

Based on the technical characteristics: in the actual point cloud filtering operation, a proper constraint condition parameter combination needs to be selected according to specific terrain, ground features and space distribution characteristics of the terrain and the ground features, a more satisfactory result can be achieved by adopting a progressive filtering method, under quite a lot of conditions, continuous ground needs to be divided into a plurality of blocks according to the ground fluctuation rule, and respective parameter combinations are selected according to the characteristics of the blocks.

As a preferable scheme of the artificial intelligence based level sensing apparatus according to the present invention, the step S1011 further comprises the steps of:

s10111: and filtering the point cloud preprocessing information, wherein the filtering comprises slope-based filtering based on mathematical morphology, unmanned aerial vehicle Lidar point cloud filtering based on TIN, pseudo scanning line filtering and wavelet layered filtering.

Based on the technical characteristics: when the mathematical morphology theory is used for filtering unmanned aerial vehicle Lidar point cloud data, in order to extract ground points, the traditional mathematical morphology 'opening' operator is improved, and the processing is carried out according to the following procedures:

and (5) discrete point corrosion treatment. Traversing unmanned aerial vehicle Lidar point cloud data, opening a window with the size of w multiplied by w by taking any point as a center, comparing the elevation of each point in the window, and taking the minimum elevation value in the window as the elevation after corrosion;

and (4) discrete point expansion treatment. And traversing the point cloud data of the unmanned aerial vehicle Lidar, and expanding the corroded data by using a structural window with the same size. The method comprises the following steps of opening a window with the size of w multiplied by w by taking any point as a center, replacing an original elevation value with an etched elevation value at the moment, comparing the elevations of all points in the window, and taking the maximum elevation value in the window as the expanded elevation;

and (4) extracting ground points. And (4) if Zp is the original elevation of the point p and t is a threshold, judging whether the point is a ground point or not when the expansion operation of each point is finished. And if the absolute value of the difference between the expanded elevation value of the point p and the original elevation value Zp of the point p is less than or equal to the threshold value t, the point p is considered as a ground point, otherwise, the point p is considered as a non-ground point.

Determining an optimal filtering function according to the change of the terrain slope, wherein for a given height difference value, the probability that a laser foot point with a large height value belongs to a ground point is smaller along with the reduction of the distance between two points; the gradient-based filtering algorithm has the characteristics of simplicity in calculation, strong adaptability and the like, but the terrain gradient needs to be known in advance and the size of a window to be opened needs to be determined, a selected point needs to be compared with all other points to determine whether the point is a ground point or not, gradient calculation needs to be carried out on each point in the whole data set, therefore, the calculated amount is increased, the speed is reduced, original point cloud data are partitioned according to the terrain statistical characteristics, then each partition is processed according to the gradient-based filtering algorithm to obtain each data ground point set, and finally, all blocks are spliced according to the characteristic points of an overlapped area to obtain the complete ground point set. Thus, different blocks can obtain different filtering thresholds, so that the singularity of the thresholds is avoided, and the classification error is reduced;

TIN is an important model representing digital elevation, often used to store the proximity relationships between spatially discrete points. For irregularly distributed elevation points, an unordered point set described as a plane can be formally formed, and a common method for converting the point set into the TIN is to construct a Delaunay triangulation network of the point set, where adjacent points in the Delaunay triangulation network are also adjacent in the TIN model.

The filtering algorithm is to filter the ground object points by utilizing parameters such as a height difference critical value condition, the number of the adjacent points meeting the condition and the like by utilizing the elevation mutation relation of the ground object adjacent point clouds in the TIN model.

Suppose a non-empty point Cloud Pt _ Cloud is given, two threshold conditions of height difference (threshold _ h) and the number of adjacent points (threshold _ vn) are given according to the distribution of regional terrain, buildings, vegetation and the like and the elevation change situation, and two arrays of Filtered and Unfiltered are defined to record Filtered points and Unfiltered points respectively.

The irregular triangulation network (TIN) based method is a two-dimensional neighborhood search based method, and the calculation amount and the algorithm complexity are relatively large. In general, the filtering effect of the algorithm is good for the ground objects with sudden elevation changes due to obvious sudden elevation changes between tall buildings and vegetation and the adjacent ground points, but an excessive error is generated when filtering the shrubs or the short ground objects. The experimental data (data including coverage types of buildings, roads, vegetation, dry lands, reservoirs and the like) of Majianhua and the like in the undulating mountainous area are filtered based on a TIN filtering algorithm, and the result shows that the filtering effect of the ground objects with obvious high-range mutation such as the vegetation, the buildings and the like is good, in the obtained DEM, the undulating change of the terrain is basically maintained, the basic shape of the road can be clearly displayed, and the error is large in the dry flat area.

The method for solving the problems needs to select a proper constraint condition parameter combination according to specific terrain, ground features and space distribution characteristics thereof in the actual point cloud filtering operation, and adopts a progressive filtering method to possibly achieve a satisfactory result, and under quite a lot of conditions, continuous ground needs to be divided into a plurality of blocks according to the ground fluctuation rule, and respective parameter combinations are selected according to the block characteristics;

the pseudo scanning line method is a data structure which reorganizes laser points which are discretely distributed in two dimensions on a horizontal plane into a one-dimensional linear continuously distributed point sequence and is called a pseudo scanning line.

Pseudo scan line based filtering is to use elevation discontinuity information to distinguish between ground and non-ground points. The basic idea is as follows: the height difference between two points is caused by the undulation of the natural topography and the height of the ground feature together. The greater the difference in height between two adjacent points, the less likely it is that this difference in height is due to natural terrain, more likely it is that the higher point is on the terrain and the lower point is on the ground, i.e.: suppose there are two adjacent laser foot points p ₁ And p ₂ ，p ₁ Is a ground point, p ₂ Is its neighborhood. If their height value h ₁ And h ₂ The conditions are satisfied: h is a total of ₂ -h ₁ ≤Δh _max ×d(Δh _max Is the tolerance of the elevation difference, d is the horizontal distance between them), then p is considered to be ₂ Is also the ground point, otherwise, p is considered to be ₂ Is a non-ground point.

The two-dimensional filtering problem is simplified into the one-dimensional filtering problem based on the pseudo scanning line filtering algorithm, the algorithm is simple in structure, the calculated amount of filtering is effectively reduced, the accuracy is guaranteed, and meanwhile, the algorithm only needs two filtering parameters, and automation is easy to achieve. However, the local neighborhood two-dimensional filter mostly assumes that the lowest elevation point in the neighborhood is a ground point, and when the ground points are few, such a filtering method is often ineffective. And the filtering algorithm based on the pseudo scanning lines can always ensure that each filtering window contains ground points, can obtain smaller first-class errors and total errors, and can accurately extract the topographic points. In a flat area, the filtering effect of the pseudo scanning line is very good, and in a steep terrain area, the error of the pseudo scanning line is controlled within a small range. However, in the case of steep slopes and areas with severe elevation changes or large objects to be filtered, the elevation threshold and the size of the filter window are usually reduced in order to obtain reliable results; in urban areas, in order to filter out large buildings completely, the size of the filter window needs to be increased properly so that the size of the filter window is not smaller than the maximum size of the building. At present, the selection of these two parameters cannot be fully automated, and the method is yet to be further improved.

Based on a multi-resolution direction prediction filtering method, point clouds of laser foot points of the unmanned aerial vehicle Lidar positioned on different ground objects show different elevation differences, and ground points and non-ground points can be distinguished by means of elevation abrupt changes between adjacent laser foot points.

For a certain distance range, if the difference value between the current point pi and the predicted value in all directions is greater than the maximum height difference limit difference under the distance condition, the point is a ground object point, otherwise, the point is a ground point.

Wavelet layering method; the wavelet transformation is introduced into unmanned aerial vehicle Lidar data filtering, and data filtering based on the wavelet transformation is achieved. In general, the ground points of the point cloud data of the unmanned aerial vehicle Lidar appear as the lowest elevation points in the local area. Thus, the raw data may be divided into windows of a particular size, and then a point of lowest elevation may be selected in each window to form a new data description. The ground points are networked to form a rough terrain surface. And then, the rough terrain surface is used as a reference surface, and filtering is carried out on the next layer to obtain more ground points. It is necessary to ensure that there is at least one ground point in each of the segmented windows, which requires that the segmented windows be sufficiently large. A window slightly larger than the largest building area within the target area is used as a measure for the top level data description. Next, the window size of the first layer data description is determined. The window size described by the first layer data is taken as the smallest area of the man-made building in the target area.

The wavelet hierarchical filtering algorithm firstly carries out layering, and then the rough interest domain is transmitted to the next layer to be used as an initial value of the interest domain of the current layer, so that the calculation time is reduced, and the accuracy of a processing result is improved. In addition, the selection of the scale of the segmentation window is also important, the scale of the data description of the top layer is slightly larger than the largest building area in the target area, and the scale of the window described by the data description of the first layer is used for selecting the size of the smallest artificial building area in the target area. In addition, the wavelet layered filtering algorithm also needs to consider more detail in the aspects of data initial value selection and discrimination rules, and eliminate gross errors in data.

The vegetation and building classification method is applied, and point cloud classification is carried out by using echo information.

At present, methods for extracting vegetation and buildings based on LiDAR point clouds mainly comprise a supervised learning classification method and a point cloud segmentation level extraction method, wherein the supervised learning classification method mainly utilizes a machine learning method to train and learn feature samples to generate a classifier for extracting buildings, the performance of the methods depends on the selection of the samples, and the cost is high. The latter is based on the auxiliary information provided by the LiDAR system or the three-dimensional topological relation of the point cloud.

As a preferable aspect of the artificial intelligence based level sensing apparatus of the present invention, the step S10111 further includes the following steps:

s101111: and (5) processing the ground point model, and modeling the automatically classified ground points. And if unreasonable triangular meshes exist, manually dividing points which are not separated into ground points until no unreasonable triangular meshes exist. The regions with abrupt elevation changes can be processed;

s101112: and obtaining elevation information, separating out contour line key points on the finely classified ground point model, automatically generating a contour line by using software, setting parameters such as minimum area, smoothness and contour distance, and removing small self-moving circles to generate the contour line of a scale required by the project. After the contour lines are exported, the key points of the model are exported into an ENZ format, and the contour lines and the elevation points are edited by using CASS software, so that the elevation element acquisition of the topographic map is completed.

Step S2 further includes:

as a preferable scheme of the artificial intelligence-based level sensing apparatus of the present invention, wherein S201: and checking a data source before construction.

Based on the technical characteristics: before construction, five planes are adopted to carry an UCXp-WA aerial camera to obtain aerial images of an application area, the obtaining time is 2016 years and 12 months, the resolution of original aerial images is 0.20m, digital elevation model data with the grid interval of 0.5m in the aerial image range are extracted through image preprocessing and space-three encryption, abnormal points and ground surface ground objects are removed through three-dimensional editing, then ground point (star, las) results are obtained, and an orthographic image module is used for quickly embedding the image data and the digital elevation model.

As a preferable scheme of the artificial intelligence based level sensing apparatus of the present invention, wherein S202: checking a data source of the completion construction period,

based on the technical characteristics: in the engineering completion period, a flying horse unmanned aerial vehicle is adopted to carry a SONYILCE-5100 aerial camera to obtain aerial images of the application area, the obtaining time is 2021 years and 8 months, the resolution ratio of the original aerial image is 0.10m, and the orthographic images can be quickly spliced by adopting a flying horse unmanned aerial vehicle steward intelligent jigsaw module through air-to-three processing, dem generation and quick embedding processes.

As a preferable scheme of the artificial intelligence-based level sensing apparatus of the present invention, wherein S303: the data sources are summarized.

Based on the technical characteristics: and summarizing the data of the two.

Preferably, the step S4 further includes the steps of:

s401: and (4) creating a project, creating a new project by using software, setting project parameters, and building a uniform data platform for subsequent calculation.

S402: importing point cloud data; point cloud data is created to the pointClouds right key CreatePoint cloud in the Prospector.

Determining basic point cloud information: specifying a name, a point cloud style (Single) and a point cloud layer (V-SITE-SCAN) in an information dialog box;

in the SourceData dialog, sourceData selects "create new point cloud database", selects "LAS" in the "point cloud file format", adds data, and assigns a corresponding coordinate system. The name is entered under the "specify new point cloud database" under the "new point cloud database". Wherein, the 'coordinate system of point cloud database' and the 'coordinate system of current figure' are set to be matched.

Point cloud database coordinate system modification: this value is obtained from the point cloud source file, and the coordinate system can be modified after importing the data into the point cloud database.

Modifying the coordinate system of the current graph: this value is obtained from the "units and bands" tab of the "graphic settings" dialog.

S403: adding point cloud data to the curved surface, designating source data, verifying point cloud parameters and creating a point cloud object.

Advantageous effects

Compared with the prior art, the invention has the advantages that:

1. the method comprises the steps of calculating the gradient of each point, dividing original point cloud data into blocks according to topographic statistical characteristics, processing each block according to a filtering algorithm based on gradient change to obtain a ground point set of each block, and splicing each block according to characteristic points of an overlapped area to obtain a complete ground point set. Thus, different blocks can obtain different filtering thresholds, so that the singularity of the thresholds is avoided, and the classification error is reduced.

2. And finally, splicing the blocks according to the characteristic points of the overlapped area to obtain a complete ground point set. Thus, different blocks can obtain different filtering thresholds, so that the singularity of the thresholds is avoided, and the classification error is reduced.

3. In the construction completion period, a flying horse unmanned aerial vehicle is adopted to carry an SONYILCE-5100 aerial camera to obtain aerial images of the application area, the obtaining time is 2021 years and 8 months, the resolution ratio of the original aerial image is 0.10m, and the fast splicing of the orthographic images can be realized by adopting a flying horse unmanned aerial vehicle housekeeper intelligent jigsaw module through air-to-three processing, dem generation and fast embedding processes.

4. In the earth and stone square balance, the overall earth and stone square balance result in various constructions such as a field, a side slope, a road and the like needs to be comprehensively considered, and the earth and stone square quantity of the flat field occupies a relatively large proportion. The utility model provides an amount of earth and stone that unmanned aerial vehicle Lidar point cloud to project area calculated the gained compares with the amount of earth and stone of water conservation monitoring quarterly newspaper (annual newspaper), and the practical application effect of analysis unmanned aerial vehicle Lidar data in the estimation of amount of earth and stone. In the calculation process, the calculation range is obtained from a water and soil conservation partition map provided by a design institute, and coordinates are unified.

Drawings

FIG. 1 is a flow chart of laser data preprocessing of the present invention;

FIG. 2 is a schematic diagram of the first and last echoes of plants and the first and last echoes of building edges according to the present invention;

FIG. 3 is a schematic flow chart of a building edge extraction method based on local multi-feature point cloud classification according to the present invention;

FIG. 4 is a diagram of importing point cloud data according to the present invention;

FIG. 5 is a diagram of the invention specifying basic point cloud information;

FIG. 6 is a schematic diagram of the operation of specifying source data according to the present invention;

FIG. 7 is a schematic diagram of an image setting operation according to the present invention;

FIG. 8 is a schematic diagram of verification point cloud parameters according to the present invention;

FIG. 9 is a cloud of points added to a curved surface according to the present invention;

FIG. 10 is a schematic view of the operation of the invention in the area designated or custom-constructed curved surface;

FIG. 11 is a schematic diagram of information forming a curved surface according to the present invention;

FIG. 12 is a diagram of the creation of a design surface according to the present invention;

FIG. 13 is a diagram of the creation of element lines from objects according to the present invention;

FIG. 14 is a design surface construction show view of the present invention;

FIG. 15 is a schematic diagram illustrating the operation of adding boundary to a design surface according to the present invention.

FIG. 16 is a schematic diagram of the operation of the present invention to create a volumetric surface.

Detailed Description

In the description of the present invention, it is to be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", and the like, indicate orientations and positional relationships based on those shown in the drawings, and are used only for convenience of description and simplicity of description, and do not indicate or imply that the equipment or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus, should not be considered as limiting the present invention.

In the description of the present invention, "a plurality" means two or more unless specifically defined otherwise.

In the description of the present invention, it should be noted that, unless otherwise explicitly specified or limited, the terms "mounted", "provided", "fitted/connected", "connected", and the like, are to be interpreted broadly, such as "connected", which may be fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in a specific case to those of ordinary skill in the art.

Referring to fig. 1-16, the present invention provides a technical solution:

example 1

The method for measuring and calculating the excavation and filling stone volume of the power grid engineering by using the high-density laser point cloud technology comprises the following steps:

s1: carrying a laser radar on the unmanned aerial vehicle, and collecting high-density laser point cloud data;

step S1 further comprises the steps of:

s101: and applying for handling aviation flight batch documents according to the aerial photography area. Selecting a suitable site along the line as a take-off and landing base, signing a guarantee protocol with an air traffic control unit, coordinating an airspace and flying in batches according to the related air traffic control unit;

s102: three steps of design, self-checking and checking are needed, the accuracy of data can be ensured in the design process of the flight band, and the flight band is better applied to engineering application;

s103: in the unmanned aerial vehicle Lidar data processing, after noise point removal and coordinate conversion are carried out on a point cloud preprocessing result acquired by an airborne laser radar, point cloud automatic classification is carried out, the classified result is manually checked and corrected, and finally a corresponding classification result is output according to requirements;

s2: automatically classifying the point cloud, and then manually rechecking and performing data internal processing;

s3: utilizing a GIS to define a calculation range;

s4: and inputting the data into a data system, calculating the volume change by using the data in the two periods, and calculating the earth and stone volume.

Example 2

And point cloud data are preprocessed, and the laser point cloud data are preprocessed to obtain the space coordinates of the point cloud data and provide data for post-production DEM. The data required for processing includes: original laser point cloud data, POS resolving data and the like. The processing content comprises laser point cloud, laser point cloud coordinate conversion and the like.

The working process of the laser point cloud data processing comprises the following steps: utilizing a POS resolving result and calibration field point cloud to eliminate an eccentric angle error, utilizing calibration field line shape measuring points to eliminate a torsion error and an elevation point to eliminate a distance error, and finally carrying out coordinate conversion on the calibrated point cloud data to generate point cloud data meeting requirements

Removing point cloud noise;

in the process of collecting three-dimensional point cloud data, the laser radar is influenced by various factors, so that certain noise can be generated when the data are obtained. In actual work, besides self-measurement errors, the device can be influenced by external environments, such as the shielding of a measured target, obstacles, the surface material of the measured target and other influence factors; in addition, some local large-scale noise cannot be filtered by the same method because the local large-scale noise is far away from the target point cloud.

Noise is a point which has no relation to the description of the target information and is not useful for the subsequent reconstruction of the entire three-dimensional scene. However, in an actual point cloud data processing algorithm, it is not easy to distinguish noise points from target points with characteristic information, and some characteristic information is inevitably lost in the denoising process due to many external factors. A good point cloud filtering algorithm not only has high real-time requirement, but also well retains the characteristic information of the model when denoising. The denoising algorithm with better effect can be provided only by applying the characteristics of the noise points of the point cloud data thoroughly.

The point cloud data is in an unstructured data format, the point cloud data obtained by scanning the laser radar is influenced by the distance between an object and the radar, the distribution has nonuniformity, the point cloud data of the object close to the radar is densely distributed, and the point cloud data of the object far away from the radar is sparsely distributed. In addition, the point cloud data has characteristics of disorder and asymmetry, which causes the point cloud data to lack a clear and uniform data structure during data representation, and increases the difficulty of processing such as segmentation and identification of subsequent point clouds. The neural network is used as an end-to-end network structure, data which is usually processed is conventional input data, such as sequences, images, videos, 3D data and the like, disordered data such as a point set cannot be directly processed, when point cloud data is processed by convolution operation, the convolution directly discards shape information of the point cloud, and only sequence information of the point cloud is reserved.

Classifying point cloud data; in Terrasolide software, the point cloud data classification map layers are as follows: 1) Defaults; 2) Group; 3) vegetation; 4) Building; 5) Lowpoint; 6) Modelkypoids. Wherein, the Default layer is a point cloud data layer temporarily stored in the operation process; group (Ground point) is mainly a point for storing and reflecting the real landform of the Ground (manually built embankments, earth dikes, step roads, naturally formed large-scale soil pits, soil piles and other deposits), and the part of manually built hydraulic structures such as earth ridges, retaining dams, dry dikes, water gates and the like connected with the Ground is regarded as a Ground point; vegetation (Vegetation points) are points where surface Vegetation is mainly stored, grasslands, shrubs, bamboo forests, nurseries, young forests, gardens, woodlands and the like are regarded as Vegetation points; building points are mainly points for storing earth surface buildings, and houses, greenhouses and the like are regarded as Building points.

And determining an optimal filter function according to the change of the terrain gradient, wherein for a given height difference value, the probability that the laser foot point with a large height value belongs to the ground point is smaller along with the reduction of the distance between the two points.

The slope-based filtering algorithm has the characteristics of simplicity in calculation, strong adaptability and the like, but the terrain slope is required to be known in advance and the size of a windowed window is required to be determined, a selected point must be compared with all other points to determine whether the point is a ground point or not, slope calculation needs to be carried out on each point in the whole data set, therefore, the calculated amount is increased, the speed is reduced, original point cloud data are partitioned according to the terrain statistical characteristics, then each partition is processed according to the slope change-based filtering algorithm to obtain each block of data ground point set, and finally, all blocks are spliced according to the feature points of an overlapped area to obtain the complete ground point set. Thus, different blocks can obtain different filtering thresholds, so that the singularity of the thresholds is avoided, and the classification error is reduced.

TIN is an important model representing digital elevation, often used to store the proximity relationships between spatially discrete points. For irregularly distributed elevation points, an unordered point set described as a plane can be formally formed, and a common method for converting the point set into the TIN is to construct a Delaunay triangulation network of the point set, where adjacent points in the Delaunay triangulation network are also adjacent in the TIN model.

The filtering algorithm is to filter the ground object points by utilizing parameters such as a height difference critical value condition, the number of adjacent points meeting the condition and the like by utilizing the elevation mutation relation of the ground object adjacent point clouds in the TIN model;

the method for solving the problems needs to select a proper constraint condition parameter combination according to specific terrain, ground features and space distribution characteristics thereof in actual point cloud filtering operation, and can achieve a satisfactory result only by adopting a progressive filtering method.

And (4) point cloud data is mapped, a ground point model is processed, and elevation information is obtained.

Example 3

The ground point filtering method application comprises a filtering algorithm based on mathematical morphology, a filtering method based on gradient, an unmanned aerial vehicle Lidar point cloud filtering algorithm based on TIN, a pseudo scanning line algorithm, wavelet layering and the like, wherein the filtering algorithm of mathematical morphology is used for extracting ground points when the mathematical morphology theory is used for filtering unmanned aerial vehicle Lidar point cloud data, the traditional mathematical morphology 'on' operator is improved, and the filtering algorithm is processed according to the following procedures:

and (5) carrying out discrete point corrosion treatment. Traversing unmanned aerial vehicle Lidar point cloud data, opening a window with the size of w multiplied by w by taking any point as a center, comparing the elevation of each point in the window, and taking the minimum elevation value in the window as the elevation after corrosion;

and (5) discrete point expansion treatment. And traversing the unmanned aerial vehicle Lidar point cloud data again, and expanding the corroded data by using a structural window with the same size. The method comprises the following steps of opening a window with the size of w multiplied by w by taking any point as a center, replacing an original elevation value with an etched elevation value at the moment, comparing the elevations of all points in the window, and taking the maximum elevation value in the window as the expanded elevation;

and (4) extracting ground points. Let Zp be the original elevation of point p and t be the threshold, and at the end of each expansion operation, make a judgment as to whether the point is a ground point. And if the absolute value of the difference between the expanded elevation value of the point p and the original elevation value Zp of the point p is less than or equal to the threshold value t, the point p is considered as a ground point, otherwise, the point p is considered as a non-ground point.

A filtering algorithm based on gradient change; determining an optimal filtering function according to the change of the terrain gradient, wherein for a given height difference value, the probability that a laser foot point with a large height value belongs to a ground point is smaller along with the reduction of the distance between two points; the slope-based filtering algorithm has the characteristics of simplicity in calculation, strong adaptability and the like, but the terrain slope is required to be known in advance and the size of a windowed window is required to be determined, a selected point must be compared with all other points to determine whether the point is a ground point or not, slope calculation needs to be carried out on each point in the whole data set, therefore, the calculated amount is increased, the speed is reduced, original point cloud data are partitioned according to the terrain statistical characteristics, then each partition is processed according to the slope change-based filtering algorithm to obtain each block of data ground point set, and finally, all blocks are spliced according to the feature points of an overlapped area to obtain the complete ground point set. Thus, different blocks can obtain different filtering thresholds, so that the singularity of the thresholds is avoided, and the classification error is reduced.

Irregular triangulation network (TIN) based filtering; TIN is an important model representing digital elevation, often used to store the proximity relationships between spatially discrete points. For irregularly distributed elevation points, an unordered point set described as a plane can be formally formed, and a common method for converting the point set into the TIN is to construct a Delaunay triangulation network of the point set, where adjacent points in the Delaunay triangulation network are also adjacent in the TIN model.

The filtering algorithm is to filter the ground object points by utilizing parameters such as a height difference critical value condition, the number of the adjacent points meeting the condition and the like by utilizing the elevation mutation relation of the ground object adjacent point cloud in the TIN model.

Suppose a non-empty point Cloud Pt _ Cloud is given, two threshold conditions of height difference (threshold _ h) and the number of adjacent points (threshold _ vn) are given according to the distribution of regional terrain, buildings, vegetation and the like and the elevation change situation, and two arrays of Filtered and Unfiltered are defined to record Filtered points and Unfiltered points respectively.

The irregular triangulation network (TIN) based method is a two-dimensional neighborhood search based method, and the calculation amount and the algorithm complexity are relatively large. In general, the filtering effect of the algorithm is good for the ground objects with sudden elevation changes due to obvious sudden elevation changes between tall buildings and vegetation and the adjacent ground points, but an excessive error is generated when filtering the shrubs or the short ground objects. Experimental data (data including coverage types of buildings, roads, vegetation, dry lands, reservoirs and the like) on undulating mountain areas such as Majianhua and the like are filtered based on a TIN filtering algorithm, and results show that the filtering effect of the ground objects with relatively obvious high-range mutation such as the vegetation, the buildings and the like is better, in the obtained DEM, the undulation change of the terrain is basically maintained, the basic shape of the road can be clearly displayed, and the error is larger in the dry and flat areas.

In the actual point cloud filtering operation, a proper constraint condition parameter combination needs to be selected according to specific terrain, ground features and space distribution characteristics of the terrain and the ground features, a more satisfactory result can be achieved by adopting a progressive filtering method, under quite a lot of conditions, continuous ground needs to be divided into a plurality of blocks according to the ground fluctuation rule, and respective parameter combinations are selected according to the characteristics of the blocks.

The pseudo scanning line method is a data structure which reorganizes laser points which are discretely distributed in two dimensions on a horizontal plane into a one-dimensional linear continuously distributed point sequence and is called a pseudo scanning line. Wu et al propose a filtering method based on a pseudo scan line based on a one-dimensional neighborhood search of the pseudo scan line.

Pseudo scan line based filtering is to use elevation discontinuity information to distinguish between ground and non-ground points. The basic idea is as follows: the height difference between two points is caused by the undulation of the natural topography and the height of the ground feature together. If the height difference between two adjacent points is larger, thenThe less likely this height difference is due to natural terrain, the more likely it is that the higher points are located on the terrain and the lower points are located on the ground, i.e.: suppose there are two adjacent laser foot points p ₁ And p ₂ ，p ₁ Is a ground point, p ₂ Is its neighborhood. If their height value h ₁ And h ₂ The conditions are satisfied: h is ₂ -h ₁ ≤Δh _max ×d(Δh _max Is the tolerance of the elevation difference, d is the horizontal distance between them), then p is considered to be ₂ Is also the ground point, otherwise, p is considered to be ₂ Is a non-ground point.

The pseudo scanning line filtering algorithm is based on, the two-dimensional filtering problem is simplified into the one-dimensional filtering problem, the algorithm is simple in structure, the calculated amount of filtering is effectively reduced, the accuracy is guaranteed, meanwhile, the algorithm only needs two filtering parameters, and automation is easy to achieve. However, the local neighborhood two-dimensional filter mostly assumes that the lowest elevation point in the neighborhood is a ground point, and when the ground points are few, such a filtering method is often ineffective. And the filtering algorithm based on the pseudo scanning lines can always ensure that each filtering window contains ground points, can obtain smaller first-class errors and total errors, and can accurately extract the topographical points. In a flat area, the filtering effect of the pseudo scanning line is very good, and in a region with steep terrain, the error of the pseudo scanning line is controlled within a small range. However, in steep slopes and areas with severe elevation changes or where large objects are filtered, the elevation threshold and the size of the filter window are typically reduced in order to obtain reliable results; in urban areas, in order to filter out large buildings completely, the size of the filter window needs to be increased properly so that the size of the filter window is not smaller than the maximum size of the building. At present, the selection of these two parameters cannot be fully automated, and the method is yet to be further improved.

Based on a multi-resolution direction prediction filtering method, point clouds of laser foot points of the unmanned aerial vehicle Lidar positioned on different ground objects show different elevation differences, and ground points and non-ground points can be distinguished by means of elevation abrupt changes between adjacent laser foot points.

The idea of the direction prediction method is as follows: for a certain distance range, if the difference value between the current point pi and the predicted value in all directions is greater than the maximum height difference limit difference under the distance condition, the point is a ground object point, otherwise, the point is a ground point.

Wavelet layering method; the wavelet transformation is introduced into unmanned aerial vehicle Lidar data filtering, and data filtering based on the wavelet transformation is achieved. In general, the ground points of the point cloud data of the unmanned aerial vehicle Lidar appear as elevation lowest points in a local area. Thus, the raw data may be divided into windows of a particular size, and then a point of lowest elevation may be selected in each window to form a new data description. The ground points are networked to form a rough terrain surface. And then, the rough terrain surface is used as a reference surface, and filtering is carried out on the next layer to obtain more ground points. It is necessary to ensure that there is at least one ground point in each of the segmented windows, which requires that the segmented windows be sufficiently large. A window slightly larger than the largest building area within the target area is used as a measure for the top level data description. Next, the window size of the first layer data description is determined. The window size described by the first layer data is taken as the smallest area of the man-made building in the target area.

Example 4

The wavelet layered filtering algorithm is to layer first, and then to transmit the rough interest domain to the next layer as the initial value of the interest domain of the current layer, thereby reducing the calculation time and improving the accuracy of the processing result, but the judgment result of each layer is affected by the previous layer, and if the processing of the previous layer has errors, the errors can cause the judgment of the data point type of the next layer to have errors. In addition, the selection of the scale of the segmentation window is also important, the scale of the data description of the top layer is slightly larger than the largest building area in the target area, and the scale of the window described by the data description of the first layer is used for selecting the size of the smallest artificial building area in the target area. In addition, the wavelet layered filtering algorithm also needs to consider more detail in the aspects of data initial value selection and discrimination rules, and eliminate gross errors in data.

The method mainly comprises a supervised learning classification method and a point cloud segmentation level extraction method, wherein the method mainly comprises the steps of training and learning a characteristic sample by using a machine learning method to generate a classifier to extract the building, the performance of the method depends on the selection of the sample, and the cost is high. The latter is based on the auxiliary information provided by the LiDAR system or the three-dimensional topological relation of the point cloud.

Example 5

And (5) processing a ground point model, namely modeling the automatically classified ground points (DEM), and carrying out manual intervention on the observation model. And if unreasonable triangulation networks exist, points which are not separated are manually classified to ground points until no unreasonable triangulation networks exist. Parameters or algorithms can be adjusted for the regions with abrupt elevation changes, and automatic classification of small areas can be carried out again;

and obtaining elevation information, separating out contour line key points on the finely classified ground point model, automatically generating a contour line by using software, setting parameters such as minimum area, smoothness and contour distance, and removing small self-moving circles to generate the contour line of a scale required by the project. And after the contour lines are exported, exporting the key points of the model into an ENZ format, and editing the contour lines and the elevation points by using CASS software to finish the acquisition of the elevation elements of the topographic map.

Example 6

The method comprises the steps of checking a data source before construction, adopting five airplanes to carry a UCXp-WA aerial camera to obtain aerial images of an application area before construction, obtaining the aerial images of the application area in 2016 (12) years and with the resolution of 0.20m of original aerial images, extracting digital elevation model data with the grid interval of 0.5m in the aerial range through image preprocessing and aerial encryption, obtaining ground point (las) results after abnormal points and ground surface ground objects are removed through three-dimensional editing, and utilizing an orthographic image module to quickly inlay the image data and the digital elevation model.

And (3) checking a data source of a finished construction period, and in the construction completion period, adopting a flying horse unmanned aerial vehicle to carry a SONYILCE-5100 aerial camera to obtain aerial images of the application area, wherein the obtaining time is 2021 years and 8 months, the resolution ratio of the original aerial image is 0.10m, and adopting a flying horse unmanned aerial vehicle manager intelligent jigsaw module to realize the quick splicing of the orthographic images through the air-to-three processing, dem generation and quick inlaying processes.

And summarizing the data source, and summarizing the data of the data source and the data of the data source.

Example 7

And (4) creating a project, creating a new project by using software, setting project parameters, and building a uniform data platform for subsequent calculation.

Importing point cloud data; point cloud data is created to the pointClouds right key CreatePoint cloud in the Prospector.

Determining basic point cloud information: specifying a name, a point cloud style (Single) and a point cloud layer (V-SITE-SCAN) in an information dialog box;

in the SourceData dialog, sourceData selects "create new point cloud database", selects "LAS" in the "point cloud file format", adds data, and assigns a corresponding coordinate system. The name is entered under the "specify new point cloud database" under the "new point cloud database". Wherein, the 'coordinate system of point cloud database' and the 'coordinate system of current figure' are set to match.

Point cloud database coordinate system modification: this value is obtained from the point cloud source file, and the coordinate system can be modified after importing the data into the point cloud database.

Modifying the coordinate system of the current graph: this value is obtained from the "units and bands" tab of the "graphic settings" dialog.

Adding point cloud data to the curved surface, specifying source data, verifying point cloud parameters and creating a point cloud object,

in the Summary dialog, the collection in the "properties" table is expanded to ensure that the properties match the previously specified properties. If the property values do not match, the link to the left of the dialog box is used to return to the previous page. And displaying a notice being processed in a status bar at the lower right corner, and if the notice that the processing is finished shows that the point cloud database is processed and the point cloud object is created.

Clicking the point cloud tab and right clicking AddPointstoSurface opens a guide window.

The name of the surface can be entered in the "surfaces options" page, the style of the surface is selected, and the next step is clicked. Here, point cloud points may also be added to an existing surface in the current graph.

A designated region or custom bounding surface can be selected on the "Regionoptions" page, clicking on "Next".

In the "Summary" page, the collection in the "Properties" table is expanded to ensure that the properties match the previously specified properties. If the property values do not match, please return to the previous page using the link to the left of the dialog box. Curved contours and triangulation will be displayed in the graph, and curved objects will be displayed on the "browse" tab in "tool space".

Selecting the curved surface, right key-object viewer, the three-dimensional effect of the curved surface can be viewed.

Design surface construction

(1) Creating an empty-design surface

(2) Creating element lines from objects

Selecting a design site borderline, creating a key line from the object, and specifying an elevation of the key line, wherein the elevation can be given an estimated value or according to a design elevation.

(3) Adding element lines to the design surface

And selecting an element line, adding the element line as a characteristic line to the design curved surface, defaulting other parameters, clicking to determine, namely adding the element line to the curved surface, selecting the curved surface, and allowing the design curved surface to be seen as a plane in the object viewer.

(4) Example 8 adding bounds to design surface, selecting element lines, fixing design Range to design Red line

Analysis of results

In earth and stone square balance, the overall earth and stone square balance result in various constructions such as fields, side slopes, roads and the like needs to be comprehensively considered, and the earth and stone square quantity of flat fields has a relatively large proportion. The utility model provides an amount of earth and stone that unmanned aerial vehicle Lidar point cloud calculated the gained to project area compares with the amount of earth and stone of water conservation monitoring quarterly newspaper (annual newspaper), analyzes the practical application effect of unmanned aerial vehicle Lidar data in the estimation of amount of earth and stone. In the calculation process, the calculation range is obtained from a water and soil conservation partition map provided by a design institute, and coordinates are unified.

In earth and stone square balance, the overall earth and stone square balance result in various constructions such as fields, side slopes, roads and the like needs to be comprehensively considered, and the earth and stone square quantity of flat fields has a relatively large proportion. The utility model provides an amount of earth and stone that unmanned aerial vehicle Lidar point cloud calculated the gained to project area compares with the amount of earth and stone of water conservation monitoring quarterly newspaper (annual newspaper), analyzes the practical application effect of unmanned aerial vehicle Lidar data in the estimation of amount of earth and stone. In the calculation process, the calculation range is obtained from a water and soil conservation partition map provided by a design institute, and coordinates are unified.

The foregoing shows and describes the general principles, principal features, and advantages of the invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention, and that various changes and modifications may be made without departing from the spirit and scope of the present invention, which fall within the scope of the claimed invention. The scope of the invention is defined by the appended claims and equivalents thereof.

Claims (7)

1. The method for measuring and calculating the excavation and filling stone volume of the power grid engineering by using the high-density laser point cloud technology is characterized by comprising the following steps of:

s1: carrying a laser radar by an unmanned aerial vehicle, and collecting high-density laser point cloud data before and after a project excavating and filling area;

s2: automatically classifying the point cloud, and then manually rechecking and performing data internal processing;

s3: utilizing a GIS to define a calculation range;

and S4, inputting the data into a data system, calculating volume variation by using the data before and after excavation, and calculating the volume of earth and stone.

2. The method for measuring and calculating the excavation and filling stone volume of the power grid engineering by using the high-density laser point cloud technology as claimed in claim 1, wherein the method comprises the following steps: the step S1 further includes the steps of:

s101: unmanned aerial vehicle Lidar data processing carries out noise point through the data to airborne laser radar obtains and gets rid of, coordinate conversion back, carries out point cloud automatic classification, and the manual work is examined, is revised to the achievement after classifying, at last according to the classification achievement that requires output corresponding.

3. The method for measuring and calculating the excavation and filling stone volume of the power grid engineering by using the high-density laser point cloud technology as claimed in claim 2, wherein the method comprises the following steps: the step S101 further includes the steps of:

s1011: point cloud data preprocessing;

s1012: removing point cloud noise;

s1013: classifying point cloud data;

s1014: and (4) point cloud data is mapped, a ground point model is processed, and elevation information is obtained.

4. The method for measuring and calculating the earth and rock excavation and filling in the power grid engineering by using the high-density laser point cloud technology as claimed in claim 3, wherein the method comprises the following steps: the step S1033 further includes the steps of:

s10111: and filtering the point cloud preprocessing information, wherein the filtering comprises slope-based filtering based on mathematical morphology, unmanned aerial vehicle Lidar point cloud filtering based on TIN, pseudo scanning line filtering and wavelet layered filtering.

5. The method for measuring and calculating the excavation and filling stone volume of the power grid engineering by using the high-density laser point cloud technology as claimed in claim 4, wherein the method comprises the following steps: the step S10111 further includes the steps of:

s101111: and (5) processing the ground point model, and modeling the automatically classified ground points. And if unreasonable triangular meshes exist, manually dividing points which are not separated into ground points until no unreasonable triangular meshes exist. The regions with abrupt elevation changes can be processed;

s101112: and acquiring elevation information, separating out contour line key points on the finely classified ground point model, automatically generating a contour line by using software, setting parameters such as minimum area, smoothness and contour distance, removing small self-moving circles, and generating the contour line of a scale required by the project. And after the contour lines are exported, exporting the key points of the model into an ENZ format, and editing the contour lines and the elevation points by using CASS software to finish the acquisition of the elevation elements of the topographic map.

6. The method for measuring and calculating the excavation and filling stone volume of the power grid engineering by using the high-density laser point cloud technology as claimed in claim 1, wherein the method comprises the following steps: the step S2 further includes:

s201: checking a data source before construction;

s202: checking a data source of a completed construction period;

s303: the data sources are summarized.

7. The method for measuring and calculating the excavation and filling stone volume of the power grid engineering by using the high-density laser point cloud technology as claimed in claim 1, wherein the method comprises the following steps: the step S4 further includes the steps of:

s401: creating a project;

s402: importing point cloud data;

s403: adding point cloud data to the curved surface, designating source data, verifying point cloud parameters and creating a point cloud object.

Images