阅读说明：本技术 一种基于TerraSolid参数阈值自动选择的地面点云滤波方法 (Ground point cloud filtering method based on Terrasolide parameter threshold automatic selection ) 是由 黄漪 赵文 张恒 韩祖杰 王�华 宁新稳 范登科 于 2019-09-20 设计创作，主要内容包括：本发明公开了一种基于TerraSolid参数阈值自动选择的地面点云滤波方法,包括：1)点云数据预处理,去除点云中的噪声点,包括多次回波的首次回波和中间回波点、低点和空中点；2)参数阈值自动选择,包括对最大房屋尺寸、迭代角、迭代距离三个关键滤波参数进行均匀采样,编辑并运行TerraSolid宏命令,获取不同滤波参数阈值所得到的地面点数,然后对地面点数进行Logistic曲线拟合,计算斜率最大点,最后得到滤波参数推荐阈值；3)根据推荐阈值,执行地面点滤波算法,分离地面点。本发明的方法能够自动选择滤波参数阈值,减少人工工作量,提高地面点云滤波的效率。(The invention discloses a ground point cloud filtering method based on Terras solid parameter threshold automatic selection, which comprises the following steps: 1) point cloud data are preprocessed, and noise points in the point cloud are removed, wherein the noise points comprise first echoes and middle echo points of multiple echoes, low points and empty middle points; 2) the automatic selection of the parameter threshold comprises the steps of uniformly sampling three key filtering parameters including the maximum house size, the iteration angle and the iteration distance, editing and operating a Terrasolide macro command, obtaining ground points obtained by different filtering parameter thresholds, then carrying out Logistic curve fitting on the ground points, calculating a maximum slope point, and finally obtaining a filtering parameter recommended threshold; 3) and according to the recommended threshold value, executing a ground point filtering algorithm and separating the ground points. The method can automatically select the filtering parameter threshold, reduce the manual workload and improve the efficiency of ground point cloud filtering.)

1. A ground point cloud filtering method based on Terrasolide parameter threshold automatic selection comprises the following steps:

s1, point cloud data preprocessing: removing noise points in the point cloud, including separating a first echo and a middle echo point of a plurality of echoes, and separating a low point and an aerial point;

s2, automatically selecting the parameter threshold, comprising the following steps:

s2-1, data sampling, namely uniformly sampling threshold values of three key filtering parameters of the maximum house size, the iteration angle and the iteration distance in the filtering algorithm, wherein the number of the samples is N respectively₁,N₂,N₃Obtaining N ═ N₁×N₂×N₃A group filtering parameter threshold;

s2-2, editing the macro command, adopting each group of filtering parameter threshold values obtained in the step S2-1, aiming at the preprocessed point cloud data obtained in the step S1, automatically and sequentially executing a ground point cloud filtering algorithm to obtain different filtering parameter threshold valuesNumber of ground points s corresponding to value₁,s₂,...,s_N；

S2-3, Logistic curve fitting, comprising:

a) sorting the N groups of filtering parameter thresholds from small to large according to the number s of the ground points, and numbering according to a natural number sequence i which is 1, 2.

b) To ground point data s_iAnd a natural number sequence i ═ 1, 2.., N, normalized;

c) performing Logistic curve fitting on the data subjected to normalization processing in the step b), wherein the Logistic curve fitting comprises performing linear transformation on a fitting formula, and performing linear fitting by using a least square method to obtain fitting parameters;

s2-4, selecting an optimal filtering parameter threshold: calculating the slope of a straight line from a point on a Logistic curve to an origin; calculating the corresponding ground point number when the slope is maximum, and selecting a recommended filtering parameter threshold value according to the ground point number;

and S3, separating the ground points, and executing a ground point filtering algorithm according to the recommended filtering parameter threshold value obtained in the step S2-4 to separate the ground points and the non-ground points.

2. The ground point cloud filtering method according to claim 1, wherein the separating of low points in step S1 is to separate lower points from adjacent points, and to eliminate points significantly lower than the ground.

3. The ground point cloud filtering method according to claim 2, wherein the separated empty points in step S1 are points removed that are significantly higher than the average elevation of surrounding points.

4. The ground point cloud filtering method according to claim 3, wherein the normalization in step b) is performed by the following formula:

wherein:

n is the number of filtering parameter threshold groups;

i is the ith group of filtering parameter threshold value sample number;

s_ithe number of ground points obtained by adopting the ith group of filtering parameter threshold values.

5. The ground point cloud filtering method according to claim 4, wherein the formula for performing Logistic curve fitting in step c) is as follows:

wherein:

e is a natural base number in mathematics;

and a and b are parameters to be fitted.

6. The ground point cloud filtering method according to claim 5, wherein the linear transformation is calculated by:

and (4) performing linear fitting on the formula (3) by using a least square method to obtain fitting parameters a and b.

7. The ground point cloud filtering method according to claim 6, wherein the formula for calculating the slope of the straight line in step S2-4 is:

Technical Field

The invention belongs to the field of surveying and mapping remote sensing, and particularly relates to a ground point cloud filtering method based on Terrasolide parameter threshold automatic selection.

Background

An airborne laser radar (Light Detection and Ranging, abbreviated as LIDAR) is a novel remote sensing device mainly integrating technologies such as laser Ranging, global positioning, inertial navigation and the like, and can directly acquire a three-dimensional coordinate point set of measurement codes. The method for acquiring the high-precision three-dimensional digital ground model by using the LIDAR is widely applied to the fields of basic mapping, digital cities, forest resource investigation and the like. The point cloud data acquired by LIDAR techniques includes ground point cloud data and non-ground point cloud data. In order to obtain a Digital Elevation Model (DEM), a ground point cloud and a non-ground point cloud are separated by a filtering method. Ground point cloud filtering is a necessary step for processing LIDAR point cloud data and is also the basis for subsequent application.

The Terras-based solid series software is a set of widely applied commercialized LIDAR point cloud data processing software, is a plug-in system developed by Terras-based solid based on MicroStation, and mainly comprises modules such as Terras, Terra model and Terra Photo. The filtering classification principle of Terrasolid is based on a progressive encryption method based on an irregular triangulation network (TIN) proposed by Axelsson in 2000, sparse TIN is generated by initial seed points, and points meeting a set threshold condition are gradually added into the TIN through iterative processing, wherein the specific documents are as follows:

Vosselman George.Slope based filtering of laser altimetry data[J].International Archives of Photogrammetry&Remote Sensing,2000,xxxiii,935–942.

the key of the method lies in the selection of the threshold of the filtering parameter, the main filtering parameter comprises the maximum house size, the iteration angle and the iteration distance, and the proper threshold needs to be selected according to experience. For the filtering process of the LIDAR point cloud data of different terrains, the setting of the artificial experience threshold value often needs repeated tests, so that the workload of point cloud filtering is increased, and the processing efficiency is low. Therefore, how to automatically select the corresponding filtering parameter threshold value by using the acquired data information and reduce the manual subjective and empirical intervention is still an urgent problem to be solved.

Disclosure of Invention

The technical problem to be solved by the invention is to provide a ground point cloud filtering method based on TerraSolid parameter threshold automatic selection, aiming at the problems of low efficiency and strong subjectivity of manual parameter threshold setting, automatically selecting a corresponding filtering parameter threshold according to a sampling data calculation result, and improving the efficiency and precision of ground point cloud filtering.

Therefore, the technical scheme of the invention is as follows:

a ground point cloud filtering method based on Terrasolide parameter threshold automatic selection comprises the following steps:

s1, point cloud data preprocessing, comprising the following steps:

s1-1, separating the first echo and the intermediate echo point of the multiple echoes;

s1-2, separating low points, separating lower points from adjacent points, and eliminating points which are obviously lower than the ground and are possibly wrong;

s1-3, separating the empty center points, and eliminating points obviously higher than the average elevation of the surrounding points;

s2, automatically selecting the parameter threshold, comprising the following steps:

s2-1, data sampling, namely uniformly sampling threshold values of three key filtering parameters of the maximum house size, the iteration angle and the iteration distance in the filtering algorithm, wherein the number of the samples is N respectively₁,N₂,N₃Obtaining N ═ N₁×N₂×N₃A group filtering parameter threshold;

s2-2, editing Terrasolid Macro (Macro) commands, and for each group of filtering parameter threshold values obtained in the step S2-1, automatically and sequentially executing a ground point cloud filtering algorithm for the preprocessed point cloud data obtained in the step S1 to obtain ground point numbers S corresponding to different filtering parameter threshold values₁,s₂,...,s_N；

S2-3, Logistic curve fitting, comprising:

a) sorting N groups of filtering parameter thresholds from small to large according to the number s of ground points, and numbering according to a natural number sequence i which is 1, 2.

b) To ground point data s_iAnd a natural number sequence i ═ 1, 2.., N, normalized using the following formula:

c) will be provided withAnd

performing Logistic curve fitting, wherein the fitting formula is as follows:

in the formula, a and b are parameters to be fitted.

And (3) performing linear transformation on the formula (2) to obtain:

performing linear fitting on the formula (3) by using a least square method to obtain fitting parameters a and b;

s2-4, selecting an optimal filtering parameter threshold, including:

1) calculating the slope of a straight line from a point on a Logistic curve to an origin point by the following formula

2) Calculating the corresponding s value when the slope k is maximum, and selecting the value closest to the s value

The corresponding group of filtering parameter threshold values are used as recommended filtering parameter threshold values;

and S3, separating the ground points, and executing a ground point filtering algorithm by adopting a Terrasolide macro command according to the recommended filtering parameter threshold value obtained in the step S2-4 to separate the ground points from the non-ground points.

The ground point cloud filtering algorithm has the following beneficial results:

(1) in the existing filtering operation based on Terra monolithic software, the value of a key filtering parameter is often selected according to manual experience and repeated tests are needed, so that the workload of point cloud filtering is increased and the processing efficiency is low. The method of the invention automatically selects the optimal filtering parameter threshold value by utilizing the acquired sampling data information, reduces the intervention of artificial subjective experience, and improves the efficiency of ground point cloud filtering.

(2) For point cloud data of different terrains and large ranges, filtering by taking a single filtering parameter threshold obviously cannot meet the precision requirement. The method has the characteristics of strong automation and high calculation speed, so that different terrains can be partitioned, one window datum is selected for each partition, the filtering parameter threshold is selected by using the method, and then the partition filtering is performed, so that the accuracy of ground point cloud filtering is improved.

Drawings

FIG. 1 is a flow chart of the terraSold parameter-based ground point cloud filtering method of the present invention;

FIG. 2 is an ISPRS test point cloud data in an embodiment of the present invention;

FIG. 3 is a Terrasolide point cloud preprocessing command in an embodiment of the present invention;

FIG. 4 is a Terrasolid ground point filter command in an embodiment of the invention;

FIG. 5 is a Macro file in an embodiment of the present invention;

FIG. 6 is a graphical illustration of a Logistic curve fit in an embodiment of the invention;

FIG. 7 is a diagram illustrating maximum slope selection in an embodiment of the present invention;

FIG. 8 illustrates ground point filtering results in an embodiment of the present invention;

FIG. 9 is a schematic diagram of an error accuracy analysis in an embodiment of the present invention.

Detailed Description

The technical solution of the present invention is specifically described below with reference to the accompanying drawings and specific embodiments.