CN118608552B - Building feature line extraction method based on 3D point cloud - Google Patents

Building feature line extraction method based on 3D point cloud Download PDF

Info

Publication number
CN118608552B
CN118608552B CN202410821067.5A CN202410821067A CN118608552B CN 118608552 B CN118608552 B CN 118608552B CN 202410821067 A CN202410821067 A CN 202410821067A CN 118608552 B CN118608552 B CN 118608552B
Authority
CN
China
Prior art keywords
point
points
point cloud
boundary
straight line
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202410821067.5A
Other languages
Chinese (zh)
Other versions
CN118608552A (en
Inventor
姜长磊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shandong Xingtu Information Technology Co ltd
Rizhao Polytechnic
Original Assignee
Shandong Xingtu Information Technology Co ltd
Rizhao Polytechnic
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shandong Xingtu Information Technology Co ltd, Rizhao Polytechnic filed Critical Shandong Xingtu Information Technology Co ltd
Priority to CN202410821067.5A priority Critical patent/CN118608552B/en
Publication of CN118608552A publication Critical patent/CN118608552A/en
Application granted granted Critical
Publication of CN118608552B publication Critical patent/CN118608552B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/13—Edge detection
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/70—Denoising; Smoothing
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/143—Segmentation; Edge detection involving probabilistic approaches, e.g. Markov random field [MRF] modelling
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
    • G06V10/763—Non-hierarchical techniques, e.g. based on statistics of modelling distributions
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10028—Range image; Depth image; 3D point clouds
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20076—Probabilistic image processing
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30181—Earth observation

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Software Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Probability & Statistics with Applications (AREA)
  • Health & Medical Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Computing Systems (AREA)
  • Artificial Intelligence (AREA)
  • Multimedia (AREA)
  • Image Analysis (AREA)

Abstract

本发明涉及建筑特征分析技术领域，具体涉及基于三维点云的建筑物特征线提取方法，包括以下步骤：使用三维激光扫描仪获取建筑物的点云数据；对采集到的点云数据进行预处理，包括删除孤立点和噪声点；采用改进的RANSAC算法检测建筑物的立面，确定拟合平面；基于多尺度统计分析方法去除立面点云中的离群点；将去噪后的点云数据投影到拟合平面上；使用自适应alpha shape算法提取立面点云的边界；采用密度峰值聚类算法将边界点云分离成多个簇；使用改进的RANSAC算法从每个簇中提取边界点云的直线段；交点计算。本发明，能够快速准确地从三维点云数据中提取出建筑物的轮廓特征线，显著提高了提取效率和准确性，适用于智慧城市建设中的高精度三维建模需求。

The present invention relates to the technical field of building feature analysis, and in particular to a method for extracting building feature lines based on three-dimensional point clouds, comprising the following steps: using a three-dimensional laser scanner to obtain point cloud data of a building; preprocessing the collected point cloud data, including deleting isolated points and noise points; using an improved RANSAC algorithm to detect the facade of the building and determine a fitting plane; removing outliers in the facade point cloud based on a multi-scale statistical analysis method; projecting the denoised point cloud data onto the fitting plane; using an adaptive alpha shape algorithm to extract the boundary of the facade point cloud; using a density peak clustering algorithm to separate the boundary point cloud into multiple clusters; using an improved RANSAC algorithm to extract straight line segments of the boundary point cloud from each cluster; and calculating intersections. The present invention can quickly and accurately extract the contour feature lines of a building from three-dimensional point cloud data, significantly improving the extraction efficiency and accuracy, and is suitable for the high-precision three-dimensional modeling requirements in smart city construction.

Description

Building characteristic line extraction method based on three-dimensional point cloud
Technical Field
The invention relates to the technical field of building feature analysis, in particular to a three-dimensional point cloud-based building feature line extraction method.
Background
With the continuous advancement of smart city construction, digital twin technology has become one of its core technological means. The digital twin city aims to accurately model and simulate the real city by a digital means, so that city management and operation efficiency is improved. In this process, the building is an important component of the city, and the reconstruction of the three-dimensional model and the feature line extraction thereof are always hot spots and difficult problems of research. The high-precision building characteristic line extraction is crucial for constructing a city information model (CI M), so that the accuracy of city planning and design can be improved, and the method plays an important role in building maintenance, disaster response and the like.
In the aspect of building characteristic line extraction, the prior art mainly adopts a Hough transformation algorithm, a region growing method and a deep learning method. However, these methods have some limitations. For example, the Hough transform algorithm is inefficient in processing large-scale point cloud data, the region growing method is sensitive to the selection of an initial point and is susceptible to noise, and the deep learning method is excellent in a specific scene, but the training data is required to be large and the computational resource is required to be high. In addition, the problems of poor robustness and insufficient precision of noise data generally exist in the methods, and the requirements of high-precision three-dimensional modeling are difficult to meet.
Aiming at the problems, it is important to provide a building characteristic line extraction method based on three-dimensional point cloud.
Disclosure of Invention
Based on the above purpose, the invention provides a three-dimensional point cloud-based building characteristic line extraction method.
The building characteristic line extraction method based on the three-dimensional point cloud comprises the following steps of:
s1, acquiring point cloud data, namely acquiring the point cloud data of a building by using a three-dimensional laser scanner;
s2, preprocessing the point cloud, namely preprocessing the acquired point cloud data, wherein the preprocessing comprises deleting isolated points and noise points;
S3, detecting the vertical face of the building, namely detecting the vertical face of the building by adopting an improved RANSAC algorithm, and determining a fitting plane;
s4, removing outliers, namely removing outliers in the vertical point cloud based on a multi-scale statistical analysis method;
s5, point cloud projection, namely projecting the denoised point cloud data onto a fitting plane to ensure the extraction precision of the characteristic lines;
s6, boundary extraction, namely extracting boundaries of the vertical point cloud by using a self-adaptive ALPHA SHAPE algorithm;
s7, boundary clustering, namely separating boundary point clouds into a plurality of clusters by adopting a density peak clustering algorithm;
s8, extracting straight line segments, namely extracting the straight line segments of the boundary point cloud from each cluster by using an improved RANSAC algorithm;
S9, calculating intersection points, namely obtaining the intersection points by extending adjacent straight line segments to obtain the integral characteristic line of the building facade;
s10, checking the accuracy of the extracted characteristic lines.
Further, the step S3 specifically includes:
s31, setting initial parameters of a RANSAC algorithm, wherein the initial parameters comprise iteration times N, an interior point threshold d and fitting accuracy E;
S32, a dynamic sampling strategy is to dynamically adjust the number of sample points under different sampling scales, so that the adaptability of the algorithm to building facades of different scales is improved;
S33, preliminary elevation detection, namely randomly selecting S points as sample point sets in each iteration, and calculating candidate plane parameters;
s34, estimating inner points, namely calculating the distances from all points to the plane by using parameters of the candidate plane, wherein the distances are defined as: Wherein d i is the distance from point i to the plane, (x i,yi,zi) is the coordinates of the point, and (a, b, c, d) is the parameters of the plane;
S35, selecting interior points, namely if d i is less than or equal to epsilon, counting the number of the interior points by taking the point i as the interior point, and selecting a candidate plane with the largest number of the interior points;
s36, plane fitting optimization, namely performing plane fitting on the selected inner point set by using a least square method, and optimizing plane parameters to enable the plane parameters to meet the following formula:
wherein (x i,yi,zi) is the coordinates of the inner points, and n is the number of the inner points;
S37, self-adaptive adjustment, namely, after each iteration, adjusting the number of sampling points and the iteration times according to the interior point proportion, wherein the formula is N=log (1-p)/log (1- (1-epsilon) s), p is the expected confidence coefficient, epsilon is the interior point proportion, and S is the number of sampling points.
Further, the step S4 specifically includes:
S41, neighborhood definition, namely defining a neighborhood for each point in the point cloud data set Wherein the method comprises the steps ofIs a set of all points within a sphere centered on point p i and having a radius r k, where k is a multi-scale parameter;
S42, calculating the distance d ij：dij＝‖pi-pj II of each point p i to all points in the neighborhood, wherein II represents Euclidean distance,
S43, calculating the average distance of each point under different scales, wherein the average distance is expressed as:
Wherein, Is the number of points in the neighborhood, k represents the kth scale;
S44, outlier detection, namely, the multi-scale average distance of all points obeys Gaussian distribution, and the global average distance mu k and the standard deviation sigma k are calculated;
S45, identifying outliers, namely identifying points with multi-scale average distances exceeding a threshold value in the point cloud data set as outliers, namely, threshold=mu k+t·σk, wherein t is a set standard deviation multiple;
And S46, removing the outliers, namely deleting all points identified as the outliers to obtain a point cloud data set after the outliers are removed.
Further, the step S5 specifically includes:
S51, determining parameters of a fitting plane, namely obtaining a parameter equation of the fitting plane through a RANSAC algorithm, wherein the equation is in the form of a.x+b.y+c.z+d=0, and a, b, c and d are parameters of the fitting plane;
S52, calculating the distance between the point cloud and the plane, namely calculating the vertical distance d Pi between each denoised point P (x s,ys,zs) and the fitting plane, wherein the calculation mode is the same as the calculation mode for calculating the distance between all points and the plane:
S53, calculating projection points, namely projecting each point P (x s,ys,zs) onto a fitting plane to obtain projection points P '(x s′,ys′,zs'), wherein the calculation is expressed as:
xs′＝xs-a·t;
ys′＝ys-b·t;
zs′＝zs-c·t;
and S54, integrating projection points P '(x s′,ys′,zs') of all points P (x s,ys,zs) to form a projected point cloud data set, and performing accuracy verification on the projected point cloud data set to ensure that errors in the projection process are within a preset range so as to ensure the accuracy of feature line extraction.
Further, the extracting the boundary in S6 specifically includes:
S61, setting alphashape initial parameters of an algorithm, including an initial radius alpha 0;
s62, neighborhood definition, defining a neighborhood for each point in the point cloud data set Wherein, Is a set of all points within a sphere centered on point p i and having a radius of α 0;
s63, calculating the neighborhood density, namely calculating the density of the points in the neighborhood of each point:
Wherein, Is the number of points in the neighborhood;
S64, self-adaptive radius adjustment, namely adjusting alpha radius alpha i of each point according to the density of the point cloud data, wherein the formula is as follows: wherein ρ max is the maximum density in the point cloud dataset;
s65, constructing alphashape by using the self-adaptive adjusted radius alpha i, and rolling a sphere with the radius alpha i on the point cloud data set to form a boundary;
And S66, boundary extraction, namely recording alphashape boundary points formed by scrolling, and extracting boundaries of the vertical point cloud.
Further, after the boundary is extracted, boundary optimization is further included, and the boundary optimization optimizes the extracted boundary through a gradient descent algorithm, so that the boundary precision is improved.
Further, the boundary clustering in S7 specifically includes:
S71, calculating Euclidean distances among all points in the boundary point cloud to form a distance matrix D= { D ij }, wherein D ij represents the distance between the point P i and the point P j;
S72, calculating the density, namely calculating the local density rho i of each point P i, wherein the local density is defined as the number of points of the point P i within a distance threshold delta, and the points are expressed as:
S73, calculating a distance peak value, namely calculating the minimum distance delta i from each point P i to the point with higher local density, wherein the minimum distance delta i is expressed as: for the point with the highest local density, delta i＝maxj(dij is set;
S74, selecting the point with the highest local density rho and distance peak delta as a clustering center;
S75, other points are distributed to the cluster where the nearest cluster center is located, and each point is distributed according to the distance and density between the point and the cluster center, wherein the formula is as follows:
Cluster (P i)＝argminC(diC), wherein C represents the Cluster center, and d iC is the distance from the point P i to the Cluster center;
and S76, optimizing the clusters, namely optimizing the initially allocated clusters, adjusting the attribution of points to maximize the density of each cluster, and outputting the finally allocated clusters as a plurality of clusters of boundary point clouds.
Further, the step S8 specifically includes:
S81, respectively processing boundary point clouds of each cluster, and dividing a point cloud data set into a plurality of clusters { C T }, wherein T represents the serial number of the cluster;
S82, selecting sample points, namely randomly selecting S points in each cluster as a sample point set, and calculating candidate straight line parameters;
S83, fitting a candidate straight line, namely fitting the candidate straight line by using a least square method according to a sample point set, and calculating a parameter equation of the candidate straight line, wherein the parameter equation is expressed as y=mx+b, m is the slope of the straight line, and b is the intercept of the straight line;
S84, evaluating inner points, namely calculating the vertical distance d line from each point to the straight line by using parameters of the candidate straight line, wherein the formula is as follows: If d line(Pi) is less than or equal to E, taking the point P i as an inner point, and counting the number of the inner points;
S85, selecting the interior points, namely selecting the candidate straight line with the largest number of the interior points as the best straight line of the current cluster;
S86, linear fitting optimization, namely performing linear fitting on the selected inner point set by using a least square method, and optimizing linear parameters, wherein the formula is as follows:
wherein (x i,yi) is the coordinates of the inner points, and n is the number of the inner points;
S87, extracting straight line segments, namely extracting optimized straight line segments, and extending adjacent straight line segments to intersection points of the optimized straight line segments to form complete boundary straight line segments;
S88, self-adaptive adjustment, namely after each iteration, adjusting the number of sampling points and the iteration times according to the inner point proportion, wherein the formula is as follows:
N=log (1-p)/log (1- (1-e) s), where p is the desired confidence, e is the interior point ratio, s is the number of sampling points;
s89, verifying the accuracy of the extracted straight line segment, ensuring that the extracted straight line segment meets the requirement of building boundary detection, and adjusting the RANSAC parameter according to the requirement to perform iterative optimization.
Further, the step S9 specifically includes:
S91, straight line segment fitting, namely, fitting a plurality of straight line segments to boundary point cloud data in each cluster by using a modified RANSAC algorithm, and marking the straight line segments as L i, wherein a parameter equation of L i is as follows:
y=m ix+bi, where m i is the slope of the ith straight line segment, and b i is the intercept of the ith straight line segment;
S92, extending each fitted straight line segment L i, and marking the extended straight line segment as L i';
S93, calculating intersection points of the adjacent extension straight line segments L i 'and L i+1', wherein the intersection point coordinates (x ij,yij) are obtained by solving the following equation:
yij＝mixij+bi;
S94, filtering the intersection points, namely filtering all the calculated intersection points to remove abnormal intersection points which do not accord with the outline of the building;
S95, constructing a characteristic line, namely sequentially connecting the filtered intersection points to construct an integral characteristic line of the building facade;
S96, optimizing the constructed characteristic line, ensuring the smoothness and the precision of the characteristic line, and smoothing the characteristic line by using a curve fitting algorithm, wherein the smoothing is as follows:
initializing an intersection set, namely sequentially arranging all intersection points (x ij,yij) obtained by extending adjacent straight-line segments to form the intersection set;
Selecting a curve fitting method, namely selecting a curve fitting algorithm suitable for the contour of a building, such as B spline curve fitting or polynomial fitting;
Setting curve fitting parameters, namely setting relevant parameters such as node number of a B spline curve or the order of a polynomial according to a selected fitting method;
constructing a fitting curve model by using a selected curve fitting method;
calculating a fitting curve, namely calculating parameters of a curve fitting model according to the intersection point set to obtain the fitting curve;
Smoothing the curve, namely smoothing the fitted curve to remove noise and sharp turning points on the curve;
And verifying the fitting degree of the fitted smooth curve and the actual building outline, and adjusting fitting parameters according to the requirements to ensure the accuracy and smoothness of the characteristic line.
Further, the step S10 specifically includes:
Comparing the extracted characteristic lines with the original point cloud data, and calculating the distance from each characteristic line point to the corresponding original point cloud point;
calculating error indexes, namely calculating error indexes of the characteristic line and the original point cloud data, wherein the error indexes comprise average error, root mean square error and maximum error;
and (3) error analysis, namely analyzing the error index obtained by calculation and judging whether the precision of the characteristic line meets the requirement.
The invention has the beneficial effects that:
the invention provides a high-efficiency and accurate building characteristic line extraction method by combining an improved RANSAC algorithm, a self-adaptive ALPHA SHAPE algorithm, a density peak clustering algorithm and a gradient descent algorithm. Through multi-scale sampling, dynamic parameter adjustment and multi-stage optimization, the adaptability and the precision of the algorithm to building facades of different scales are improved. Particularly, by utilizing a self-adaptive ALPHA SHAPE algorithm and a density peak clustering algorithm, noise and outliers in the point cloud data are effectively processed, and the integrity and the accuracy of feature line extraction are ensured. The method can rapidly and accurately extract the outline characteristic line of the building from the three-dimensional point cloud data, remarkably improves the extraction efficiency and accuracy, and is suitable for the high-precision three-dimensional modeling requirement in the construction of smart cities.
After the characteristic line is extracted, the characteristic line is smoothed by adopting a curve fitting algorithm, so that the smoothness and the precision of the characteristic line are ensured, the noise and sharp turning points on the curve are removed, and the continuity and the smoothness of the characteristic line are ensured. The method remarkably improves the quality of the characteristic lines, enables the extracted characteristic lines to be more in line with the outline of an actual building, improves the precision and the attractiveness of the three-dimensional model, and provides a solid foundation for high-precision building modeling.
The invention ensures that the extracted characteristic line has high accuracy and reliability through detailed precision checking steps. And calculating error indexes (comprising average error, root mean square error and maximum error) through the comparison of the characteristic lines and the original point cloud data, and comprehensively evaluating the accuracy and the error distribution condition of the characteristic lines by combining with a manual check sum error distribution map.
Drawings
In order to more clearly illustrate the invention or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only of the invention and that other drawings can be obtained from them without inventive effort for a person skilled in the art.
Fig. 1 is a schematic flow chart of an extraction method according to an embodiment of the invention.
Detailed Description
The present invention will be further described in detail with reference to specific embodiments in order to make the objects, technical solutions and advantages of the present invention more apparent.
As shown in fig. 1, the three-dimensional point cloud-based building feature line extraction method comprises the following steps:
S1, acquiring point cloud data of a building by using a three-dimensional laser scanner, wherein the acquisition comprises the following steps of:
selecting a three-dimensional laser scanner with high precision and high resolution to ensure that the acquired point cloud data has sufficient detail and accuracy;
setting scanning parameters of a scanner, including scanning resolution, scanning range and scanning angle, according to the height, area and structural characteristics of a building;
Selecting a plurality of scanning positions to ensure that the outer vertical surface of the whole building can be covered, and reducing shielding and blind areas as much as possible;
scanning is carried out at each scanning position according to the set scanning parameters, and point cloud data of the building are obtained;
Integrating and calibrating the point cloud data acquired from the plurality of scanning positions, and storing the integrated and calibrated point cloud data in a storage device for backup so as to prevent data loss;
s2, preprocessing the point cloud, namely preprocessing the acquired point cloud data, including deleting isolated points and noise points, specifically including:
performing preliminary processing on the acquired point cloud data by using open source point cloud processing software;
cutting out relatively complete partial point cloud data;
deleting isolated points and noise points, and ensuring the integrity and purity of point cloud data;
S3, detecting the vertical face of the building, namely detecting the vertical face of the building by adopting an improved RANSAC algorithm, and determining a fitting plane;
s4, removing outliers, namely removing outliers in the vertical point cloud based on a multi-scale statistical analysis method;
s5, point cloud projection, namely projecting the denoised point cloud data onto a fitting plane to ensure the extraction precision of the characteristic lines;
s6, boundary extraction, namely extracting boundaries of the vertical point cloud by using a self-adaptive ALPHA SHAPE algorithm;
s7, boundary clustering, namely separating boundary point clouds into a plurality of clusters by adopting a density peak clustering algorithm;
s8, extracting straight line segments, namely extracting the straight line segments of the boundary point cloud from each cluster by using an improved RANSAC algorithm;
S9, calculating intersection points, namely obtaining the intersection points by extending adjacent straight line segments to obtain the integral characteristic line of the building facade;
s10, checking the accuracy of the extracted characteristic lines.
S3 specifically comprises:
s31, setting initial parameters of a RANSAC algorithm, wherein the initial parameters comprise iteration times N, an interior point threshold d and fitting accuracy E;
S32, a dynamic sampling strategy is to dynamically adjust the number of sample points under different sampling scales, so that the adaptability of the algorithm to building facades of different scales is improved;
S33, preliminary elevation detection, namely randomly selecting S points as sample point sets in each iteration, and calculating candidate plane parameters;
s34, estimating inner points, namely calculating the distances from all points to the plane by using parameters of the candidate plane, wherein the distances are defined as: Wherein d i is the distance from point i to the plane, (x i,yi,zi) is the coordinates of the point, and (a, b, c, d) is the parameters of the plane;
S35, selecting interior points, namely if d i is less than or equal to epsilon, counting the number of the interior points by taking the point i as the interior point, and selecting a candidate plane with the largest number of the interior points;
s36, plane fitting optimization, namely performing plane fitting on the selected inner point set by using a least square method, and optimizing plane parameters to enable the plane parameters to meet the following formula:
wherein (x i,yi,zi) is the coordinates of the inner points, and n is the number of the inner points;
S37, self-adaptive adjustment, namely after each iteration, adjusting the number of sampling points and the iteration times according to the inner point proportion, wherein the formula is N=log (1-p)/log (1- (1-epsilon) s), wherein p is the expected confidence coefficient, epsilon is the inner point proportion, S is the number of sampling points, verifying the accuracy of a fitting plane, ensuring that the fitting plane meets the requirement of building elevation detection, and adjusting the RANSAC parameter according to the requirement to perform iterative optimization.
S4 specifically comprises the following steps:
S41, neighborhood definition, namely defining a neighborhood for each point in the point cloud data set Wherein the method comprises the steps ofIs a set of all points within a sphere centered on point p i and having a radius r k, where k is a multi-scale parameter;
S42, calculating the distance d ij：dij＝‖pi-pj II of each point p i to all points in the neighborhood, wherein II represents Euclidean distance,
S43, calculating the average distance of each point under different scales, wherein the average distance is expressed as:
Wherein, Is the number of points in the neighborhood, k represents the kth scale;
S44, outlier detection, namely, the multi-scale average distance of all points obeys Gaussian distribution, and the global average distance mu k and the standard deviation sigma k are calculated and expressed as:
wherein n is the total point number of the point cloud data set;
S45, identifying outliers, namely identifying points with multi-scale average distances exceeding a threshold value in the point cloud data set as outliers, namely, threshold=mu k+t·σk, wherein t is a set standard deviation multiple;
And S46, removing the outliers, namely deleting all points identified as the outliers to obtain a point cloud data set after the outliers are removed.
Outliers in the point cloud data are effectively removed through a multi-scale statistical analysis method. The multi-scale neighborhood definition and distance calculation ensure the comprehensiveness and accuracy of outlier detection, and the outlier can be accurately identified by calculating the multi-scale average distance and assuming that the multi-scale average distance is subjected to Gaussian distribution. Setting a reasonable threshold ensures the effectiveness of outlier removal. The method can effectively improve the purity and the integrity of the point cloud data, and provides a high-quality data basis for the subsequent feature line extraction.
S5 specifically comprises the following steps:
S51, determining parameters of a fitting plane, namely obtaining a parameter equation of the fitting plane through a RANSAC algorithm, wherein the equation is in the form of a.x+b.y+c.z+d=0, and a, b, c and d are parameters of the fitting plane;
S52, calculating the distance between the point cloud and the plane, namely calculating the vertical distance d Pi between each denoised point P (x s,ys,zs) and the fitting plane, wherein the calculation mode is the same as the calculation mode for calculating the distance between all points and the plane:
S53, calculating projection points, namely projecting each point P (x s,ys,zs) onto a fitting plane to obtain projection points P '(x s′,ys′,zs'), wherein the calculation is expressed as:
xs′＝xs-a·t;
ys′＝ys-b·t;
zs′＝zs-c·t;
and S54, integrating projection points P '(x s′,ys′,zs') of all points P (x s,ys,zs) to form a projected point cloud data set, and performing accuracy verification on the projected point cloud data set to ensure that errors in the projection process are within a preset range so as to ensure the accuracy of feature line extraction.
The boundary extraction in S6 specifically includes:
S61, setting alphashape initial parameters of an algorithm, including an initial radius alpha 0;
s62, neighborhood definition, defining a neighborhood for each point in the point cloud data set Wherein, Is a set of all points within a sphere centered on point p i and having a radius of α 0;
s63, calculating the neighborhood density, namely calculating the density of the points in the neighborhood of each point:
Wherein, Is the number of points in the neighborhood;
S64, self-adaptive radius adjustment, namely adjusting alpha radius alpha i of each point according to the density of the point cloud data, wherein the formula is as follows: wherein ρ max is the maximum density in the point cloud dataset;
s65, constructing alphashape by using the self-adaptive adjusted radius alpha i, and rolling a sphere with the radius alpha i on the point cloud data set to form a boundary;
And S66, boundary extraction, namely recording alphashape boundary points formed by scrolling, and extracting boundaries of the vertical point cloud.
After boundary extraction, boundary optimization is further included, and the boundary optimization optimizes the extracted boundary through a gradient descent algorithm, so that the boundary precision is improved, and the method specifically comprises the following steps:
Boundary point initial set: initializing boundary point set { P i }, wherein P i＝(xi,yi,zi) from boundary point set extracted from adaptive alphashape algorithm;
The energy function is defined as an energy function E for defining a boundary point set, which is used for measuring the smoothness and the precision of the boundary point set, and the formula is as follows: Wherein, Is a neighborhood point set of points P i, lambda is a regularization parameter,Is an initial boundary point;
calculating the gradient of the energy function E with respect to the boundary point set { P i }, calculating the gradient The formula is:
Gradient descent iteration, namely updating a boundary point set { P i } by using a gradient descent method, wherein the formula is as follows:
Wherein, The boundary point position at the t-th iteration is defined, and eta is the learning rate;
Judging whether the change of the energy function E is smaller than a preset threshold E, if so, stopping iteration, otherwise, returning to the step to continue iteration;
And taking the final converged boundary point set { P i } as the optimized boundary output.
The boundary clustering in S7 specifically includes:
S71, calculating Euclidean distances among all points in the boundary point cloud to form a distance matrix D= { D ij }, wherein D ij represents the distance between the point P i and the point P j;
S72, calculating the density, namely calculating the local density rho i of each point P i, wherein the local density is defined as the number of points of the point P i within a distance threshold delta, and the points are expressed as:
S73, calculating a distance peak value, namely calculating the minimum distance delta i from each point P i to the point with higher local density, wherein the minimum distance delta i is expressed as: for the point with the highest local density, delta i＝maxj(dij is set;
S74, selecting the point with the highest local density rho and distance peak delta as a clustering center, wherein the points are usually in a high-density area and are far away from other high-density points;
S75, other points are distributed to the cluster where the nearest cluster center is located, and each point is distributed according to the distance and density between the point and the cluster center, wherein the formula is as follows:
Cluster (P i)＝argminC(diC), wherein C represents the Cluster center, and d iC is the distance from the point P i to the Cluster center;
and S76, optimizing the clusters, namely optimizing the initially allocated clusters, adjusting the attribution of points to maximize the density of each cluster, and outputting the finally allocated clusters as a plurality of clusters of boundary point clouds.
S8 specifically comprises the following steps:
S81, setting initial parameters of a RANSAC algorithm, including iteration times N, an interior point threshold d and fitting accuracy E, respectively processing boundary point clouds of each cluster, and dividing a point cloud data set into a plurality of clusters { C T }, wherein T represents the number of the clusters;
S82, selecting sample points, namely randomly selecting S points in each cluster as a sample point set, and calculating candidate straight line parameters;
S83, fitting a candidate straight line, namely fitting the candidate straight line by using a least square method according to a sample point set, and calculating a parameter equation of the candidate straight line, wherein the parameter equation is expressed as y=mx+b, m is the slope of the straight line, and b is the intercept of the straight line;
S84, evaluating inner points, namely calculating the vertical distance d line from each point to the straight line by using parameters of the candidate straight line, wherein the formula is as follows: If d line(Pi) is less than or equal to E, taking the point P i as an inner point, and counting the number of the inner points;
S85, selecting the interior points, namely selecting the candidate straight line with the largest number of the interior points as the best straight line of the current cluster;
S86, linear fitting optimization, namely performing linear fitting on the selected inner point set by using a least square method, and optimizing linear parameters, wherein the formula is as follows:
wherein (x i,yi) is the coordinates of the inner points, and n is the number of the inner points;
S87, extracting straight line segments, namely extracting optimized straight line segments, and extending adjacent straight line segments to intersection points of the optimized straight line segments to form complete boundary straight line segments;
S88, self-adaptive adjustment, namely after each iteration, adjusting the number of sampling points and the iteration times according to the inner point proportion, wherein the formula is as follows:
N=log (1-p)/log (1- (1-e) s), where p is the desired confidence, e is the interior point ratio, s is the number of sampling points;
s89, verifying the accuracy of the extracted straight line segment, ensuring that the extracted straight line segment meets the requirement of building boundary detection, and adjusting the RANSAC parameter according to the requirement to perform iterative optimization.
S9 specifically comprises:
S91, straight line segment fitting, namely, fitting a plurality of straight line segments to boundary point cloud data in each cluster by using a modified RANSAC algorithm, and marking the straight line segments as L i, wherein a parameter equation of L i is as follows:
y=m ix+bi, where m i is the slope of the ith straight line segment, and b i is the intercept of the ith straight line segment;
S92, extending each fitted straight line segment L i, and marking the extended straight line segment as L i';
S93, calculating intersection points of the adjacent extension straight line segments L i 'and L i+1', wherein the intersection point coordinates (x ij,yij) are obtained by solving the following equation:
yij＝mixij+bi;
S94, filtering the intersection points, namely filtering all the calculated intersection points to remove abnormal intersection points which do not accord with the outline of the building;
S95, constructing a characteristic line, namely sequentially connecting the filtered intersection points to construct an integral characteristic line of the building facade;
S96, optimizing the constructed characteristic line, ensuring the smoothness and the precision of the characteristic line, and smoothing the characteristic line by using a curve fitting algorithm, wherein the smoothing is as follows:
initializing an intersection set, namely sequentially arranging all intersection points (x ij,yij) obtained by extending adjacent straight-line segments to form the intersection set;
Selecting a curve fitting method, namely selecting a curve fitting algorithm suitable for the contour of a building, such as B spline curve fitting or polynomial fitting;
Setting curve fitting parameters, namely setting relevant parameters such as node number of a B spline curve or the order of a polynomial according to a selected fitting method;
constructing a fitting curve model by using a selected curve fitting method;
calculating a fitting curve, namely calculating parameters of a curve fitting model according to the intersection point set to obtain the fitting curve;
Smoothing the curve, namely smoothing the fitted curve to remove noise and sharp turning points on the curve;
And verifying the fitting degree of the fitted smooth curve and the actual building outline, and adjusting fitting parameters according to the requirements to ensure the accuracy and smoothness of the characteristic line.
S10 specifically comprises the following steps:
Comparing the extracted characteristic lines with the original point cloud data, and calculating the distance from each characteristic line point to the corresponding original point cloud point;
calculating error indexes, namely calculating error indexes of the characteristic line and the original point cloud data, wherein the error indexes comprise average error, root mean square error and maximum error;
analyzing the error index obtained by calculation, and judging whether the precision of the characteristic line meets the requirement;
The error distribution diagram can be drawn, the distribution condition of errors on each point of the characteristic line is displayed, the identification of the region with larger errors is facilitated, the parameters of the characteristic line extraction algorithm are adjusted according to error analysis, and the extraction precision of the characteristic line is optimized.
It will be appreciated by persons skilled in the art that the above discussion of any embodiment is merely exemplary and is not intended to imply that the scope of the invention is limited to these examples, that combinations of technical features in the above embodiments or in different embodiments may also be implemented in any order, and that many other variations of the different aspects of the invention as described above exist, which are not provided in detail for the sake of brevity.
The present invention is intended to embrace all such alternatives, modifications and variances which fall within the broad scope of the appended claims. Therefore, any omission, modification, equivalent replacement, improvement, etc. of the present invention should be included in the scope of the present invention.

Claims (9)

1.基于三维点云的建筑物特征线提取方法，其特征在于，包括以下步骤：1. A method for extracting building feature lines based on three-dimensional point clouds, comprising the following steps: S1，点云数据采集：使用三维激光扫描仪获取建筑物的点云数据；S1, point cloud data acquisition: use a 3D laser scanner to obtain point cloud data of the building; S2，点云预处理：对采集到的点云数据进行预处理，包括删除孤立点和噪声点；S2, point cloud preprocessing: preprocessing the collected point cloud data, including deleting isolated points and noise points; S3，建筑物立面检测：采用改进的RANSAC算法检测建筑物的立面，确定拟合平面；S3, building facade detection: using the improved RANSAC algorithm to detect the building facade and determine the fitting plane; S4，离群点去除：基于多尺度统计分析方法去除立面点云中的离群点,具体包括：S4, outlier removal: remove outliers from the facade point cloud based on multi-scale statistical analysis methods, including: S41，邻域定义：为点云数据集中的每个点定义一个邻域其中是以点pi为中心、半径为rk的球体内的所有点的集合，其中k为多尺度参数；S41, Neighborhood definition: Define a neighborhood for each point in the point cloud dataset in is the set of all points in a sphere with point pi as the center and radius rk , where k is the multi-scale parameter; S42，距离计算：对于每个点pi，计算其到邻域内所有点的距离dij：dij＝‖pi-pj‖，其中，‖·‖表示欧几里得距离， S42, distance calculation: for each point p i , calculate its distance d ij to all points in the neighborhood: d ij = ‖p i -p j ‖, where ‖·‖ represents the Euclidean distance, S43，多尺度平均距离：计算每个点在不同尺度下的平均距离，表示为：S43, multi-scale average distance: Calculate the average distance of each point at different scales, expressed as: 其中，是邻域内点的数量，k表示第k个尺度； in, is the number of points in the neighborhood, k represents the kth scale; S44，离群点检测：所有点的多尺度平均距离服从高斯分布，计算全局平均距离μk和标准差σk；S44, outlier detection: The multi-scale average distance of all points obeys Gaussian distribution, and the global average distance μ k and standard deviation σ k are calculated; S45，离群点识别：将点云数据集中多尺度平均距离超过阈值的点识别为离群点，即：threshold＝μk+t·σk，其中，t为设定的标准差倍数；S45, outlier identification: identify points in the point cloud data set whose multi-scale average distance exceeds a threshold as outliers, that is, threshold = μ k + t·σ k , where t is a set multiple of the standard deviation; S46，离群点去除：删除所有被识别为离群点的点，得到去除离群点后的点云数据集；S46, outlier removal: deleting all points identified as outliers to obtain a point cloud data set after the outliers are removed; S5，点云投影：将去噪后的点云数据投影到拟合平面上，以确保特征线提取的精度；S5, point cloud projection: project the denoised point cloud data onto the fitting plane to ensure the accuracy of feature line extraction; S6，边界提取：使用自适应alpha shape算法提取立面点云的边界；S6, boundary extraction: use the adaptive alpha shape algorithm to extract the boundary of the facade point cloud; S7，边界分簇：采用密度峰值聚类算法将边界点云分离成多个簇；S7, boundary clustering: the density peak clustering algorithm is used to separate the boundary point cloud into multiple clusters; S8，直线段提取：使用改进的RANSAC算法从每个簇中提取边界点云的直线段；S8, straight line segment extraction: use the improved RANSAC algorithm to extract the straight line segments of the boundary point cloud from each cluster; S9，交点计算：通过延长相邻直线段求取其交点，得到建筑物立面的整体特征线；S9, intersection calculation: by extending adjacent straight line segments to obtain their intersection points, the overall characteristic line of the building facade is obtained; S10，精度校验：对提取的特征线进行精度校验。S10, precision check: perform precision check on the extracted feature lines. 2.根据权利要求1所述的基于三维点云的建筑物特征线提取方法，其特征在于，所述S3具体包括：2. The method for extracting building feature lines based on three-dimensional point clouds according to claim 1, wherein S3 specifically comprises: S31，初始参数设定：设定RANSAC算法的初始参数，包括迭代次数N、内点阈值d和拟合精度∈；S31, initial parameter setting: setting the initial parameters of the RANSAC algorithm, including the number of iterations N, the inlier threshold d and the fitting accuracy ∈; S32，动态采样策略：在不同的采样规模下动态调整样本点的数量，提高算法对不同规模建筑立面的适应性；S32, dynamic sampling strategy: dynamically adjust the number of sample points under different sampling scales to improve the adaptability of the algorithm to building facades of different scales; S33，初步立面检测：在每次迭代中，随机选择s个点作为样本点集，计算候选平面参数；S33, preliminary elevation detection: In each iteration, s points are randomly selected as sample point sets to calculate candidate plane parameters; S34，内点评估：使用候选平面的参数，计算所有点到平面的距离，定义为：其中，di为点i到平面的距离，(xi,yi,zi)为点的坐标，(a,b,c,d)为平面的参数；S34, interior point evaluation: Using the parameters of the candidate plane, calculate the distance from all points to the plane, defined as: Where d i is the distance from point i to the plane, (x i , y i , z i ) are the coordinates of the point, and (a, b, c, d) are the parameters of the plane; S35，内点选择：如果di≤∈，则点i为内点,统计内点的数量，选择内点数量最多的候选平面；S35, interior point selection: if d i ≤∈, then point i is an interior point, count the number of interior points, and select the candidate plane with the largest number of interior points; S36，平面拟合优化：对选定的内点集使用最小二乘法进行平面拟合，优化平面参数；S36, plane fitting optimization: using the least square method to fit the selected internal point set to optimize the plane parameters; S37，自适应调整：在每次迭代后，根据内点比例调整采样点数量和迭代次数，公式为：N＝log(1-p)/log(1-(1-∈)s)，其中，p为期望的置信度，∈为内点比例，s为采样点数量。S37, adaptive adjustment: after each iteration, the number of sampling points and the number of iterations are adjusted according to the inlier ratio, the formula is: N = log(1-p)/log(1-(1-∈) s ), where p is the desired confidence, ∈ is the inlier ratio, and s is the number of sampling points. 3.根据权利要求2所述的基于三维点云的建筑物特征线提取方法，其特征在于，所述S5具体包括：3. The building feature line extraction method based on three-dimensional point cloud according to claim 2, characterized in that said S5 specifically comprises: S51，确定拟合平面参数：通过RANSAC算法得到拟合平面的参数方程，形式为：a·x+b·y+c·z+d＝0，其中，a,b,c和d为拟合平面的参数；S51, determining the fitting plane parameters: obtaining the parameter equation of the fitting plane by the RANSAC algorithm, in the form of: a·x+b·y+c·z+d=0, where a, b, c and d are the parameters of the fitting plane; S52，点云平面距离计算：对于去噪后的每个点P(xs,ys,zs)，计算其到拟合平面的垂直距离dPi；S52, point cloud plane distance calculation: for each point P ( xs , ys , zs ) after denoising, calculate its vertical distance dPi to the fitting plane; S53，投影点计算：将每个点P(xs,ys,zs)投影到拟合平面上，得到投影点P′(xs′，ys′，zs′)；S53, projection point calculation: project each point P ( xs , ys , zs ) onto the fitting plane to obtain the projection point P' ( xs ', ys ', zs '); S54，投影点云集成：将所有点P(xs,ys,zs)的投影点P′(xs′，ys′，zs′)集成，形成投影后的点云数据集，对投影后的点云数据集进行精度验证，确保投影过程中的误差在预设范围内，以保证特征线提取的精度。S54, projected point cloud integration: integrate the projection points P′(xs ′ , ys ′ , zs ′ ) of all points P( xs , ys , zs ) to form a projected point cloud dataset, and perform accuracy verification on the projected point cloud dataset to ensure that the error in the projection process is within a preset range to ensure the accuracy of feature line extraction. 4.根据权利要求1所述的基于三维点云的建筑物特征线提取方法，其特征在于，所述S6中的边界提取具体包括：4. The method for extracting building feature lines based on three-dimensional point clouds according to claim 1, wherein the boundary extraction in S6 specifically comprises: S61，设定alphashape算法的初始参数，包括初始半径α0；S61, setting initial parameters of the alphashape algorithm, including an initial radius α 0 ; S62,邻域定义：为点云数据集中的每个点定义一个邻域其中，是以点pi为中心、半径为α0的球体内的所有点的集合；S62, Neighborhood definition: Define a neighborhood for each point in the point cloud dataset in, is the set of all points in a sphere with a center of point pi and a radius of α0 ; S63，邻域密度计算：计算每个点邻域内点的密度：S63, neighborhood density calculation: Calculate the density of points in the neighborhood of each point: 其中，是邻域内点的数量； in, is the number of points in the neighborhood; S64，自适应半径调整：根据点云数据的密度调整每个点的alpha半径αi，公式为：其中,ρmax为点云数据集中最大密度；S64, adaptive radius adjustment: adjust the alpha radius α i of each point according to the density of the point cloud data. The formula is: Among them, ρ max is the maximum density in the point cloud dataset; S65，使用自适应调整后的半径αi构建alphashape，将半径αi的球在点云数据集上滚动，形成边界；S65, constructing an alphashape using the adaptively adjusted radius α i , and rolling a ball of radius α i on the point cloud dataset to form a boundary; S66，边界提取：记录alphashape滚动形成的边界点，提取立面点云的边界。S66, boundary extraction: record the boundary points formed by rolling the alphashape and extract the boundary of the facade point cloud. 5.根据权利要求4所述的基于三维点云的建筑物特征线提取方法，其特征在于，所述边界提取后，还包括边界优化，所述边界优化通过梯度下降算法优化提取的边界，提高边界精度。5. The method for extracting building feature lines based on three-dimensional point clouds according to claim 4 is characterized in that after the boundary extraction, it also includes boundary optimization, and the boundary optimization optimizes the extracted boundary through a gradient descent algorithm to improve the boundary accuracy. 6.根据权利要求1所述的基于三维点云的建筑物特征线提取方法，其特征在于，所述S7中的边界分簇具体包括：6. The method for extracting building feature lines based on three-dimensional point clouds according to claim 1, wherein the boundary clustering in S7 specifically comprises: S71，计算边界点云中所有点之间的欧几里得距离，形成距离矩阵D＝{dij}，其中dij表示点Pi和点Pj之间的距离；S71, calculating the Euclidean distances between all points in the boundary point cloud to form a distance matrix D={d ij }, where dij represents the distance between point Pi and point Pj ; S72，密度计算：为每个点Pi计算其局部密度ρi，局部密度定义为点Pi在距离阈值δ内的点数，表示为： S72, density calculation: for each point Pi, calculate its local density ρi , the local density is defined as the number of points within the distance threshold δ of the point Pi , expressed as: S73，距离峰值计算：计算每个点Pi到具有更高局部密度点的最小距离δi，表示为：对于局部密度最高的点，设定δi＝maxj(dij)；S73, distance peak calculation: calculate the minimum distance δ i from each point Pi to a point with a higher local density, expressed as: For the point with the highest local density, set δ i = max j (d ij ); S74，选择局部密度ρ和距离峰值δ最高的点作为聚类中心；S74, selecting the point with the highest local density ρ and distance peak δ as the cluster center; S75，将其他点分配到最近的聚类中心所在的簇中，每个点根据其与聚类中心的距离和密度进行分配，公式为：S75, assign other points to the cluster where the nearest cluster center is located. Each point is assigned according to its distance and density from the cluster center. The formula is: Cluster(Pi)＝argminC(diC)，其中，C表示聚类中心，diC为点Pi到聚类中心的距离；Cluster(P i )=argmin C (d iC ), where C represents the cluster center and d iC is the distance from point P i to the cluster center; S76，簇的优化：对初始分配的簇进行优化，调整点的归属，使得每个簇的密度最大化，将最终分配后的簇作为边界点云的多个簇输出。S76, cluster optimization: optimize the initially allocated clusters, adjust the affiliation of the points, maximize the density of each cluster, and output the finally allocated clusters as multiple clusters of the boundary point cloud. 7.根据权利要求1所述的基于三维点云的建筑物特征线提取方法，其特征在于，所述S8具体包括：7. The method for extracting building feature lines based on three-dimensional point clouds according to claim 1, wherein S8 specifically comprises: S81，对每个簇的边界点云分别进行处理，将点云数据集分割成多个簇{CT}，其中T表示簇的编号；S81, processing the boundary point cloud of each cluster separately, dividing the point cloud data set into multiple clusters {C T }, where T represents the cluster number; S82，样本点选择：在每个簇中，随机选择s个点作为样本点集，计算候选直线参数；S82, sample point selection: In each cluster, randomly select s points as the sample point set and calculate the candidate line parameters; S83，候选直线拟合：根据样本点集，使用最小二乘法拟合候选直线，计算候选直线的参数方程,表示为：y＝mx+b，其中，m为直线的斜率，b为直线的截距；S83, candidate straight line fitting: according to the sample point set, the candidate straight line is fitted using the least square method, and the parameter equation of the candidate straight line is calculated, which is expressed as: y=mx+b, where m is the slope of the straight line and b is the intercept of the straight line; S84，内点评估：使用候选直线的参数，计算每个点到直线的垂直距离dline，公式为：如果dline(Pi)≤∈，则点Pi为内点，统计内点的数量；S84, interior point evaluation: Using the parameters of the candidate line, calculate the vertical distance d line from each point to the line. The formula is: If d line (P i )≤∈, then point P i is an interior point, and the number of interior points is counted; S85，内点选择：选择内点数量最多的候选直线作为当前簇的最佳直线；S85, interior point selection: select the candidate straight line with the largest number of interior points as the best straight line of the current cluster; S86，直线拟合优化：对选定的内点集使用最小二乘法进行直线拟合，优化直线参数，公式为：S86, Line Fitting Optimization: Use the least squares method to fit the selected interior point set to optimize the line parameters. The formula is: 其中，(xi,yi)为内点的坐标，n为内点的数量； Where (x i ,y i ) are the coordinates of the interior points, and n is the number of interior points; S87，直线段提取：提取优化后的直线段，将相邻直线段延长至其交点，形成完整的边界直线段；S87, straight line segment extraction: extract the optimized straight line segment, extend the adjacent straight line segments to their intersection points, and form a complete boundary straight line segment; S88，自适应调整：在每次迭代后，根据内点比例调整采样点数量和迭代次数，公式为：S88, adaptive adjustment: After each iteration, the number of sampling points and the number of iterations are adjusted according to the ratio of inliers. The formula is: N＝log(1-p)/log(1-(1-∈)s)，其中，p为期望的置信度，∈为内点比例，s为采样点数量；N = log(1-p)/log(1-(1-∈) s ), where p is the expected confidence, ∈ is the proportion of internal points, and s is the number of sampling points; S89，验证提取的直线段的精度，确保其满足建筑边界检测的要求，并根据需要调整RANSAC参数进行迭代优化。S89, verify the accuracy of the extracted straight line segment to ensure that it meets the requirements of building boundary detection, and adjust the RANSAC parameters for iterative optimization as needed. 8.根据权利要求7所述的基于三维点云的建筑物特征线提取方法，其特征在于，所述S9具体包括：8. The method for extracting building feature lines based on three-dimensional point clouds according to claim 7, wherein S9 specifically comprises: S91，直线段拟合：对每个簇中的边界点云数据使用改进的RANSAC算法拟合出若干条直线段，记为Li，其中Li的参数方程为：S91, straight line segment fitting: Use the improved RANSAC algorithm to fit the boundary point cloud data in each cluster to obtain several straight line segments, denoted as Li , where the parameter equation of Li is: y＝mix+bi，其中，mi为第i条直线段的斜率，bi为第i条直线段的截距；y＝m i x+b i , where mi is the slope of the i-th straight line segment and b i is the intercept of the i-th straight line segment; S92，将每条拟合的直线段Li进行延长，延长后的直线段记为Li′；S92, extending each fitted straight line segment Li , and the extended straight line segment is recorded as Li ′; S93，计算相邻延长直线段Li′和Li+1′的交点，交点坐标(xij,yij)，通过解以下方程组得到：mixij+bi＝mi+1xij+bi+1，解得：S93, calculate the intersection of the adjacent extended straight line segments Li ′ and Li +1 ′. The intersection coordinates ( xij , yij ) are obtained by solving the following equations: mixij + bi =mi +1xij + bi +1 . The solution is: yij＝mixij+bi； yij = mixij + bi ; S94，交点过滤：对所有计算得到的交点进行过滤，去除不符合建筑物轮廓的异常交点；S94, intersection filtering: filtering all calculated intersections to remove abnormal intersections that do not conform to the building outline; S95，特征线构建：将过滤后的交点按顺序连接，构建出建筑物立面的整体特征线；S95, feature line construction: connect the filtered intersection points in order to construct the overall feature line of the building facade; S96，对构建的特征线进行优化，确保其光滑度和精度，使用曲线拟合算法对特征线进行平滑处理。S96, optimizes the constructed feature lines to ensure their smoothness and accuracy, and uses a curve fitting algorithm to smooth the feature lines. 9.根据权利要求1所述的基于三维点云的建筑物特征线提取方法，其特征在于，所述S10具体包括：9. The method for extracting building feature lines based on three-dimensional point clouds according to claim 1, wherein S10 specifically comprises: 将提取的特征线与原始点云数据进行对比，计算每个特征线点到其对应原始点云点的距离；Compare the extracted feature lines with the original point cloud data, and calculate the distance from each feature line point to its corresponding original point cloud point; 计算误差指标：计算特征线与原始点云数据的误差指标，包括平均误差、均方根误差和最大误差；Calculate error indicators: Calculate the error indicators between feature lines and original point cloud data, including average error, root mean square error and maximum error; 误差分析：对计算得到的误差指标进行分析，判断特征线的精度是否符合要求。Error analysis: Analyze the calculated error indicators to determine whether the accuracy of the characteristic line meets the requirements.
CN202410821067.5A 2024-06-24 2024-06-24 Building feature line extraction method based on 3D point cloud Active CN118608552B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202410821067.5A CN118608552B (en) 2024-06-24 2024-06-24 Building feature line extraction method based on 3D point cloud

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202410821067.5A CN118608552B (en) 2024-06-24 2024-06-24 Building feature line extraction method based on 3D point cloud

Publications (2)

Publication Number Publication Date
CN118608552A CN118608552A (en) 2024-09-06
CN118608552B true CN118608552B (en) 2025-02-28

Family

ID=92558938

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202410821067.5A Active CN118608552B (en) 2024-06-24 2024-06-24 Building feature line extraction method based on 3D point cloud

Country Status (1)

Country Link
CN (1) CN118608552B (en)

Families Citing this family (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN119251175B (en) * 2024-09-20 2025-03-28 湖南曼吉电子科技有限公司 Production quality inspection method, device and system for special-shaped parts
CN118857134B (en) * 2024-09-23 2024-12-06 江苏省建筑工程质量检测中心有限公司 Building deformation monitoring system and method based on three-dimensional laser scanning technology
CN119180965B (en) * 2024-11-22 2025-11-18 浙江初驰科技有限公司 A method and apparatus for multi-scale point cloud feature extraction based on feature point neighborhood division
CN119693442B (en) * 2024-12-03 2026-03-27 合肥工业大学 A rapid method for identifying and detecting pothole volume in road repair
CN119783236B (en) * 2025-03-11 2025-08-05 北京工业大学 Building modeling method, device, equipment and storage medium
CN119832170B (en) * 2025-03-18 2025-07-08 江苏濠汉信息技术有限公司 Gaussian modeling method combined with laser scanning point cloud data processing
CN119862519B (en) * 2025-03-18 2025-05-30 山东同生测绘地理信息有限公司 Geographic surveying and mapping data processing and analysis method and system based on big data
CN120612359B (en) * 2025-05-19 2026-03-10 安徽大学 Point cloud prefabricated wall size detection method based on deep learning
CN120470137B (en) * 2025-07-14 2025-09-23 中铁科学研究院集团有限公司 Detection system for falling off of shale filling broken belt lining in wide and thick earthquake region of tunnel
CN120970539A (en) * 2025-07-17 2025-11-18 北京市第三建筑工程有限公司 A Threshold Zoning Detection Method for Building Formwork Flatness Based on Normal Vector Constraint RANSAC

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112802041A (en) * 2021-02-05 2021-05-14 北京工业大学 Ground laser point cloud building contour line extraction method based on improved Alpha Shapes algorithm
CN116824379A (en) * 2023-07-04 2023-09-29 嵩山实验室 A progressive optimization method for laser point cloud building contours based on multi-dimensional features

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115761172A (en) * 2022-10-10 2023-03-07 哈尔滨工程大学 A 3D reconstruction method for single building based on point cloud semantic segmentation and structure fitting
CN116883754A (en) * 2023-07-20 2023-10-13 成都市勘察测绘研究院(成都市基础地理信息中心) Building information extraction method for ground LiDAR point cloud
CN117194945B (en) * 2023-07-25 2026-01-09 西北工业大学 A data-driven method for recognizing the shape of arbitrary group targets

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112802041A (en) * 2021-02-05 2021-05-14 北京工业大学 Ground laser point cloud building contour line extraction method based on improved Alpha Shapes algorithm
CN116824379A (en) * 2023-07-04 2023-09-29 嵩山实验室 A progressive optimization method for laser point cloud building contours based on multi-dimensional features

Also Published As

Publication number Publication date
CN118608552A (en) 2024-09-06

Similar Documents

Publication Publication Date Title
CN118608552A (en) Building feature line extraction method based on 3D point cloud
CN119672718B (en) Object labeling method and system based on laser point cloud
CN110866934B (en) Method and system for segmentation of complex point cloud based on normative coding
CN103077529A (en) System for analyzing plant leaf features based on image scanning
CN111598780A (en) Terrain adaptive interpolation filtering method suitable for airborne LiDAR point cloud
CN118628682B (en) Automatic modeling method of building 3D model based on 3D point cloud
CN114898118A (en) Automatic statistical method and system for power transmission line house removal amount based on multi-source point cloud
CN116363520B (en) A landscape ecological detection system for urban green space planning
CN119691412B (en) River channel measurement data preprocessing method based on artificial intelligence
CN120179850A (en) A Knowledge Base Intelligent Retrieval System
CN120807853B (en) Construction method and system for engineering building component model based on BIM
CN120612481A (en) Wheat organ 3D point cloud segmentation method based on dynamic voxel and category perception
CN120451973B (en) A method and system for calculating cell wall grayscale, hollowness and thickness
CN113496159B (en) Multi-scale convolution and dynamic weight cost function smoke target segmentation method
CN117152344A (en) Tunnel surrounding rock structural surface analysis method and system based on photo reconstruction point cloud
CN120953231A (en) IC Carrier Detection Method Based on Surface State Image Extraction
CN120374882A (en) Automatic modeling and updating method for dynamic evolution of urban morphology based on parameter iteration
CN111723737B (en) A target detection method based on multi-scale matching strategy deep feature learning
CN117809161B (en) Star-masking refractive index profile data quality evaluation system based on multi-source data
CN119048834A (en) Corn quality identification method, device, computer equipment and storage medium
CN112819810B (en) A Calculation Method of Roundness of Sand Particles
CN114677428A (en) Detection method of ice coating thickness of power transmission line based on UAV image processing
CN118583347B (en) A stress detection method for an organic electroluminescent device
CN116681898B (en) Image recognition and positioning method
CN119478922B (en) Method, system and storage medium for evaluating silkworm pupa status for making silkworm pupa

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
TA01 Transfer of patent application right
TA01 Transfer of patent application right

Effective date of registration: 20250114

Address after: No. 16 Yantai North Road, Donggang District, Rizhao City, Shandong Province, China 276825

Applicant after: RIZHAO POLYTECHNIC

Country or region after: China

Applicant after: Shandong Xingtu Information Technology Co.,Ltd.

Address before: Room 717, 7th Floor, Qilu Software Park Building, South End of Xinyu Road, Jinan Area, China (Shandong) Pilot Free Trade Zone, Qingdao City, Shandong Province, China 250098

Applicant before: Shandong Xingtu Information Technology Co.,Ltd.

Country or region before: China

GR01 Patent grant
GR01 Patent grant